Epidemic period emotion analysis using a deep learning based hybrid model on social media data

It is seen that social media is gaining more importance in human life day by day and has even become a decision-making body on many issues. Especially the coronavirus epidemic, which affected our lives at the end of 2019, has pushed people further away from each other socially. Every development that has occurred after the coronavirus epidemic has become a pandemic has been one of the most important agenda items of people. People's thoughts and feelings about vaccines, which are thought to end the epidemic, have become a subject that must be known. Perhaps the biggest proof of this is that the topics opened on this subject at the beginning of the pandemic are still up to date. It is as important to make the right meaning of these emotions as it is to obtain them. In this study, based on this idea, first of all, the data in 2020 and 2021 were taken from the topics that people focused on the most. In order for the data to have the correct emotion labels, many data preprocessing operations were performed on the data. We tried to create a hybrid model by using multiple models on top of each other to perform multi-class sentiment labeling. Social media posts were labeled using different classifiers such as positive, negative, neutral and also fear, sadness, fun, happy. According to the results of high-performance training operations, we focused on people's opinions about vaccines and vaccine brands. Some deep learning-based classifiers were used to increase the success rates of the study. It is assumed that this study can be successful in different areas related to the classification of social media data.