Supervised Machine Learning-Based Model for Automated Fault Localization and Intrusion Resilience in Web Applications

The rapid digital transformation of modern society has positioned web applications at the center of economic, social, and governmental activities. From online banking and e-commerce platforms to healthcare systems and e-government services, web-based systems constitute a critical infrastructure that supports daily operations across sectors. Cyberattacks targeting web applications continue to evolve in scale and sophistication. Common attack vectors include SQL injection, cross-site scripting (XSS), cross-site request forgery (CSRF), remote code execution, distributed denial-of-service (DDoS), and zero-day exploits. These attacks often exploit latent software faults, insecure coding practices, improper input validation, misconfigurations, or weak authentication mechanisms. Therefore, there is a pressing need for a data-driven, supervised machine learning framework capable of automatically detecting, classifying, and localising faults associated with security vulnerabilities in web applications. This paper, explores the application of supervised machine learning techniques for automated fault localisation in web applications, with the ultimate objective of enhancing intrusion resilience. The study adopts a quantitative, experimental research approach, as the primary objective is to develop and empirically evaluate supervised machine learning models for intrusion detection and automated fault localisation using numerical performance indicators. This study draws on two publicly available network intrusion detection benchmarks, UNSW-NB15 and CICIDS-2017, together with a purpose-built synthesised web application log dataset. The study proposes an integrated framework design that unifies supervised intrusion detection and automated fault localisation into a single pipeline. The findings of this study support four principal, appropriately hedged conclusions. First, within the scope of this study’s own experimental protocol, supervised machine learning, particularly Random Forest ensemble methods, appears more effective for intrusion detection in web applications than the traditional rule-based approach evaluated here.

Keywords: Structured Query Language (SQL), Machine Learning, Web, Fault Localization.