On the Property of the Distribution of Symbols in SQL Injection Attack

SQL injection is an attack of type to insert malicious query via an input form on web site. If SQL injection attack were successful, there are the threats of unauthorized access, information leak or falsification of data for web applications driven database system. In the conventional studies, a lot of prevention and detection methods using pattern matching, parsing or machine learning have been developed. However, it is easy to evade detection by modifying the strings of SQL injection attacks. Therefore, it is very important to investigate the essence of SQL injection attacks for preventing such evasion of detection. In this study, we constructed the feature space by using the property of the distribution of SQL injection attack, and proposed an attack detection method. The result of this study is showing the importance concerning the construction of feature space.