A Computational Sentiment Analysis for Emotion Classification in Occasion-Related English Language Tweets

Abstract

In the age of digital communication, social media platforms like Twitter have become essential spaces for individuals to express emotions, particularly during significant national and religious events. This study investigates how well machine learning classifiers identify emotions in tweets that are posted on particular occasions. The research aims to determine the impact of sentiment classification accuracy and assess the performance of different classification models in identifying emotions in tweets shared during major events. The seven main emotions—joy, sadness, anger, fear, disgust, surprise, and neutral were represented in a dataset that was carefully gathered from twelve distinct events. To categorize emotions, three popular machine learning classifiers Multinomial Naïve Bayes (MNB), Random Forest (RF), and Logistic Regression (LR) were utilized, and their results were assessed using four important metrics: accuracy, precision, recall, and F1-score. The results demonstrated that Logistic Regression (LR) outperformed other classifiers, achieving the highest accuracy of 92%. Random Forest (RF) followed closely, with an accuracy of 91%, maintaining robust results in all emotional categories. Multinomial Naïve Bayes (MNB), with an accuracy of 89%, proved effective in cases where emotions had strong keyword associations. Logistic Regression emerged as the most effective classifier for sentiment analysis of occasional tweets, Random Forest provided a strong alternative, and Multinomial Naïve Bayes remained useful for keyword-driven sentiment detection.