A vector space model approach for web attack classification using machine learning technique

Abstract

Web applications usage is increasing in online services in many ways in our day-to-day life. Business service providers have started deploying their business over the web through various e-commerce applications online. The growth of online web application increases the web complexity and vulnerability in terms of security which is a major concern in the current web security research. The extensive growth of various types of web attacks is a severe threat to web security. HTTP requests are usually secret code into a web attack spread through the injection and allow them to perform malicious actions on remote systems to execute arbitrary commands. This paper proposes an efficient approach for web attack classification, using a vector space model approach (VSMA), to improve the detection and classification accuracy. It is able to automatically classify the attacks from valid requests to detect the specific web attacks. The evaluation measure shows high precision and low recall rates than the existing classifiers in comparison.