Classifying Fake Profiles on X with Multinomial Naïve Bayes

Online social networks have experienced rapid growth, facilitating massive connections among users. However, this expansion has also led to a significant increase in fake users, putting user privacy at risk as these deceptive entities exploit the platform's accessibility to gather and misuse personal data. The proliferation of fake profiles poses a serious threat to the social fabric of the X world, potentially undermining user interactions. This research aims to classify fake profiles that mimic genuine accounts, making detection challenging. The proposed solution utilizes a Multinomial Naïve Bayes model, offering an innovative method to combat the issue of fake profiles. The model demonstrates promising results, achieving 94.67% accuracy, 90% precision, 90% recall, and a 95% area under the curve (AUC). The study concludes by calling for further research to expand the range of training examples for the model, enhancing its ability to accurately differentiate between fake and genuine X profiles. It also emphasizes the importance of maintaining key features within the model to ensure an accurate representation of the subtle differences between genuine and fake profiles.