Hybrid method integrating SQL-IF and Naïve Bayes for SQL injection attack avoidance

Abstract

Web applications are the objects most targeted by attackers. The technique most often used to attack web applications is SQL injection. This attack is categorized as dangerous because it can be used to illegally retrieve, modify, delete data, and even take over databases and web applications. To prevent SQL injection attacks from being executed by the database, a system that can identify attack patterns and can learn to detect new patterns from various attack patterns that have occurred is required. This study aims to build a system that acts as a proxy to prevent SQL injection attacks using the Hybrid Method which is a combination of SQL Injection Free Secure (SQL-IF) and Naïve Bayes methods. Tests were carried out to determine the level of accuracy, the effect of constants (K) on SQL-IF, and the number of datasets on Naïve Bayes on the accuracy and efficiency (average load time) of web pages. The test results showed that the Hybrid Method can improve the accuracy of SQL injection attack prevention. Smaller K values and larger dataset will produce better accuracy. The Hybrid Method produces a longer average web page load time than using only the SQL-IF or Naïve Bayes methods.