Prediction of Narcissistic Behavior on Indonesian Twitter Using Machine Learning Methods

Abstract

Social media's explosive expansion in recent years has changed people's behavior, leading to an increase in narcissistic tendencies. Arrogance, the need to flaunt one's accomplishments, the urge to get approval from others, and excessive dreams about success, power, intelligence, attractiveness, and other attributes are characteristics of narcissistic behavior. Social media has evolved into a platform for showcasing accomplishments, particularly for people with narcissistic tendencies. Moreover, narcissistic trait is one of the three characteristics of the dark triad personality type. Several research have demonstrated that a variety of machine learning techniques can be used to predict dark triad personality traits and narcissism from social media posts. As some studies have suggested, this narcissistic behavior can further increase the level of cyberbullying in social media, while also have strong correlation with the use of chatbot for academic cheating. This research aims to build a prediction model for behavioral symptoms of narcissism based on posts on Twitter in Indonesian language, using the natural language processing technique and several basic machine learning methods (Nearest Neighbors, Naïve Bayes, Decision Tree, and Support Vector Machine), and then compare the results. We concluded that SVM model achieved the best performance, with Accuracy = 0.72 and F1 Score = 0.725.