Decoding the impact of emotions: machine learning insights on user interests in social networks

Abstract

This study investigates the correlation that exists between users' emotional states and their expressed interests in the context of social network. By utilizing cutting-edge machine learning methods, we set out to reveal the connections that underpin user behavior. Saudi Arabia is among the nations with the highest usage of X (previously Twitter). Several studies used the analysis of English tweets to determine the topic of interest and whether the user is passive or active. Studies that examined user interaction to ascertain interest have been conducted with reference to Arabic tweets. There are, however, few studies that track how an external factor, like emotions, affects interest over time. To investigate the relationship between interest and emotion, we used two models of supervised algorithms: Support Vector Machines (SVM) and Naïve Bayes. Once the topic of interests and emotions were classified, we discovered that the topic of interest had a higher accuracy than the emotion classifier because it had been applied to a sample of dataset. Furthermore, the SVM outperformed Naïve Bayes in terms of accuracy for classifying both topics of interest and emotions. Finally, the result indicates that the interests for specific user change over time according to the emotions.