DEVELOPMENT OF METHODS FOR ANOMALY DETECTION IN WEB APPLICATIONS USING DATA MINING TECHNIQUES

Abstract

The article addresses the issue of automated anomaly detection in web applications amid increasing system complexity and data volumes. Three algorithms were developed and experimentally validated to enhance the accuracy and speed of anomaly detection. It was established that the time correlation module demonstrates the highest accuracy, the adaptive threshold monitoring ensures rapid response, and hybrid clustering offers balanced performance. Key implementation challenges, including computational complexity and the need for model adaptation, were identified. Recommendations for integrating the algorithms into monitoring systems were proposed, and prospects for developing optimized methods and exploring new types of threats were outlined.