Enhancing Online Social Network Security for Effective Detection of Bots, Spam, and Fake Accounts

Abstract

Online social networks (OSNs) are becoming prevalent in the lives of individuals, with a significant portion of the public actively engaging with them. Consequently, users are encountering challenges pertaining to spam and fraudulent accounts as a result of the proliferation of unregulated online social networks. As a consequence of these attacks, the security of their personal information remains compromised. In order to address these issues, academics have put forth a range of machine learning techniques. However, these strategies proved to be ineffective in accurately detecting bots, spam, and false accounts. This research presents the utilization of the Deep Learning Convolutional Neural Network (DLCNN) as a contemporary technique for the efficient detection of suspicious clickstream sequences and bots, enabling the inclusion of options and the limitation of measurements. The classification method for account verification is employed to ascertain the veracity of target accounts, distinguishing between genuine and fraudulent identities. Based on the comprehensive simulation findings, it is evident that the proposed Deep Learning Convolutional Neural Network (DLCNN) exhibits reduced training time and achieves superior classification accuracy in comparison to existing state-of-the-art methodologies.