On the Approximate Maximum Likelihood Estimation in Stochastic Model of SQL Injection Attacks

SQL injection is one of vulnerabilities of web application and there is a risk of an unauthorized access if web application has this vulnerability. A lot of heuristic prevention and detection methods of SQL injection attacks have been developed, however, theoretical performance analysis is difficult about them. In this paper, an approximate maximum likelihood estimation of parameter of a stochastic model which imitates the property of SQL injection attacks would be proposed. Furthermore, its effectiveness would be considered by investigating relationships between our proposed method and linear classification methods such as soft confidence weighted learning (SCW) and support vector machine (SVM).