Classification of SQL injection, XSS and Path Traversal for Web Application Attack Detection

Abstract

Web application attack detection is one of the popular research areas during these years. SQL injection, XSS and path traversal attacks are the most commonly occurred types of web application attacks. The proposed system effectively classifies three attacks by random forest algorithm to ensure reasonable accuracy. Request length module is computed based on the certain length of the URL to analyze each record as normal or attack. Regular pattern analysis is emphasized on the content of URL and other features to analyze the certain attack patterns. ECML/PKDD standard web attack dataset is used in this system. Combination of random forest algorithm with request length and regex pattern analysis is proposed to outperform the accuracy.