Classification of social media posts during the Corona crisis based on their polarity

The Iranian people were confronted with a range of emotions during the Covid-19 crisis, which they shared on social media platforms. Social media played a crucial role in disseminating information and reflecting public sentiment during the pandemic. Consequently, governments and health organizations worldwide recognized the importance of analyzing social media data. Many researchers have examined these data using different approaches worldwide. This study focuses on the polarity analysis and classification of messages posted on social media during the COVID-19 crisis. The study analyzed messages shared by Persian-language users on social networks using natural language processing and deep learning techniques. Various deep learning methods, including convolutional neural networks (CNN), long short-term memory (LSTM), and fuzzy-LSTM were used to classify the data as positive or negative polarity. The three-layer deep convolutional neural network achieved the highest accuracy of 72.29%. Finally, a comprehensive comparison of the different networks used was conducted across multiple aspects.