Analyzing public sentiment on social media during FIFA world cup 2022 using deep learning and explainable AI

Abstract

Analysis of public sentiment is extremely useful for comprehending the responses of the general public during important events, and the FIFA World Cup 2022 was no exception. Within the scope of this study, we used deep learning models such as roBERTa, distilBERT, and XLNet to conduct an analysis of the views that were stated on Twitter during the first day of the tournament. These models were fine-tuned using a comprehensive dataset consisting of 30,000 tweets, which had been preprocessed. The performance of these models was assessed using measures such as accuracy, F1-score, precision, recall, etc. In addition, we used an Explainable AI known as Local Interpretable Model-Agnostic Explanations (LIME) so that we could better understand how model decisions were made in sentiment classification. Our research has shown that roBERTa is an excellent model for classifying sentiment, and it has also shown the significance of interpretability achieved using LIME. Our research enhances the understanding of sentiment analysis during major sports events and suggests future directions for research in this domain.