Optimized bidirectional gated recurrent unit-convolutional neural network for enhanced social network security

The rapid growth of social networks has significantly transformed digital communication, facilitating large-scale information sharing. However, this expansion has also introduced complex security risks, including phishing, spam, fake accounts, cyberbullying, AI-generated misinformation, and sophisticated bot networks. Traditional security approaches often fail to address the dynamic and multifaceted nature of these threats. To tackle these challenges, we introduce an Optimized Bidirectional Gated Recurrent Unit and Convolutional Neural Network (O-Bi-GRU-DCNN) model designed specifically for social network security. This model combines the sequential learning power of Bidirectional Gated Recurrent Units (Bi-GRU) to capture contextual dependencies in social media data with the feature extraction capabilities of Deep Convolutional Neural Networks (CNN). Additionally, Archimedes Optimization (ArO) is used to enhance the model's accuracy. Experimental results reveal that the proposed model outperforms existing security techniques in detecting malicious activities, spam, and fake accounts, achieving an impressive accuracy of 98.9%. This approach offers a robust and scalable solution to enhance security across social network platforms. Graphical Abstract