Machine Learning Approach on Cyberstalking Detection in Social Media Using Naive Bayes and Decision Tree
Universiti Tun Hussein Onn Malaysia, Auni Filzah Md Nasir, Khairul Amin Mohamad Sukri, Universiti Tun Hussein Onn Malaysia
Abstract
Social media has increased the chance to communicate through many things such as video calls and can be connected globally. But there is also a growth in the vulnerability of the system. With this advancement, some threat is bound to happen. Cyberbullying is one of the social issues that users deliberately and tenaciously misuse social media. It became an issue because most cases affect the victim's mental health. Before, detecting these crimes only has traditionally used linguistic features, but cyberbullying on social media has more than that. Therefore, technologies today may play an important role in detecting cyberstalking on social media by using Machine Learning (ML). In this paper, cyberbullying detection will use the ML algorithm, which is Naïve Bayes and Decision Tree, and compare which algorithm is better to detect. ML has a wide range of methods that allow systems to quickly access the data and learn from it to make decisions for complicated problems. Cyberstalking has been concerning as it psychologically affects the victims. An experimental result indicates that Naïve Bayes algorithms achieve the best accuracy, which is 0.958.