Artificial Intelligence Based Malicious Social Bots’ Detection Model

Abstract

The widespread adoption of Online Social Networks (OSNs) has led to an alarming increase in spam content, fake accounts, and bot activity, posing significant risks to user privacy and platform integrity. To address these issues, this work proposes a novel detection framework based on Deep Learning Convolutional Neural Networks (DLCNN). The method focuses on identifying suspicious clickstream sequences and classifying user accounts as legitimate or fraudulent. By leveraging the feature extraction capabilities of convolutional layers and a supervised classification algorithm, the model effectively captures behavioral patterns associated with malicious activity. Extensive simulation results show that the proposed DLCNN model significantly outperforms existing state-of-the-art machine learning techniques. The proposed model demonstrated superior performance in terms of precision (97.2%), recall (96.1%), and F1-score (96.6%) as compared to Random Forest. This advancement contributes to the field by offering a more robust and scalable solution for real-time bot and spam detection. The proposed approach can be applied to various OSN platforms, improving user safety, data security, and the overall reliability of social network ecosystems.