Smart Defense: Machine Learning-Based Web Application Firewall

The rise in internet usage has been revolutionary, as nearly 8 billion people in the world in 2024—5.35 billion of them, or around 66% of the world's population, are active internet users. These statistics clearly show that the development of web applications has skyrocketed as they offer accessibility and flexibility, but it also introduces serious threats that can inflict considerable damage on these applications and systems and compromise the personal information of their users. Hence, leveraging security mechanisms is equally crucial. In this research, an abnormality-based web application firewall model was developed using machine learning-based classification algorithms, mainly XG Boost. The performance of different classification-based algorithms was evaluated by developing different models. The proposed model achieves higher performance in differentiating between normal and malicious requests. As per investigation, our proposed model serves to be effective in differentiating between normal and malicious requests by delivering an accuracy level of 94.2%.