Machine Learning Based Detection for Compromised Accounts on Social Media Networks

Abstract

The proliferation of social networking platforms has led to a corresponding increase in the frequency and sophistication of cyberattacks targeting user accounts. Compromised accounts can be used to spread misinformation, launch phishing attacks, and steal personal information. This paper presents a novel approach to detecting compromised accounts on social networks. Our method leverages a combination of behavioral and linguistic features to identify anomalous activity that may indicate account compromise. Behavioral features include changes in posting frequency, interaction patterns, and location data. We employ machine learning algorithms to train models that can accurately classify accounts as compromised or legitimate based on these features. Our experiments demonstrate the effectiveness of our approach in detecting compromised accounts with high precision and recall. Furthermore, we explore the potential of incorporating graph-based techniques to analyze the social network structure surrounding compromised accounts. By examining the relationships between compromised accounts and their associated nodes, we can identify potential propagation paths and take proactive measures to mitigate the spread of malicious activity