Published Paper Details:

This study presents a hybrid cyberbullying detection model that integrates advanced machine learning algorithms--Linear Support Vector Machines (LSVM) and Recurrent Neural Networks (RNNs)--to address the growing issue of cyberbullying on social media platforms like Twitter and Facebook. As cyber threats become increasingly sophisticated, traditional security measures often fall short, making the application of machine learning crucial for real-time detection and mitigation. Utilizing a dataset of text expressions, the system achieves an impressive 85% classification accuracy with LSVM and an 81% feature extraction accuracy with RNN, outperforming both deep learning and traditional machine learning techniques. The research employs Object-Oriented Analysis and Design Methodology (OOADM) to create a modular and efficient software solution that processes real-time messages for cyberbullying characteristics. The results highlight the superior capabilities of RNN and LSVM in analyzing complex patterns, essential for effective cyberbullying detection. By combining technical advancements with human engagement, this study emphasizes the need for a multifaceted approach to combat cyberbullying. The proposed framework not only enhances cybersecurity measures but also contributes valuable insights into the intersection of machine learning and online safety. Ultimately, this research offers a proactive defense strategy, demonstrating the potential of cutting-edge AI methods to address emerging cyber threats effectively.