PREDICTIVE MODELING FOR WEB SERVER VULNERABILITY DETECTION USING MACHINE LEARNING

In 2023, the Indonesian government's websites faced significant challenges with web defacement attacks, totaling 189 incidents, with the highest number, 31, occurring in January. Web defacement, which exploits vulnerabilities in web servers to alter or delete web page content, poses serious risks to data integrity and user privacy. This study addresses these challenges by proposing a comprehensive framework for identifying and evaluating security vulnerabilities in website technologies using machine learning techniques. The framework integrates data collection, preprocessing, training and modeling, and analysis into a seamless process. Data is collected using a Website Technology Crawler to identify technology stacks and Common Vulnerabilities and Exposures (CVE) databases to gather information on known vulnerabilities. The research highlights the Random Forest Classifier as the most effective model, achieving an impressive accuracy of 98%, with precision and recall scores of approximately 0.98. These metrics underscore the model's capability to accurately identify and distinguish between safe and exploitable vulnerabilities. Key features such as `cve_number`,
`version`, and `product_name` were critical indicators of exploitability, significantly enhancing predictive accuracy. Comparative analysis showed that the K-Nearest Neighbors (KNN) classifier also performed well, with an accuracy of 97%, while the Support Vector Classifier (SVC) had a slightly lower accuracy of 88%. The Random Forest model's ROC curve, with an AUC score of 0.99, highlighted its exceptional ability to discriminate between positive and negative classes. The deployment of the Random Forest model in a real-time prediction platform demonstrated its practical applicability, offering an efficient approach to vulnerability management. This research contributes to cybersecurity by providing a systematic and reliable approach to vulnerability detection, significantly improving the proactive identification and mitigation of exploitable vulnerabilities in Website Technologies.

Keyword: Website Vulnerabilities, Machine Learning, Random Forest, Support Vector Machine, K-Nearest Neighbors, Cyber Security