Evaluación de algoritmos de minería de datos para detección y predicción de Ataques de Inyección SQL en Big Data : Evaluación de SVM para la detección y predicción deAtaques de Inyección SQL en Big Data.

Abstract

Web applications have grown exponentially in the last 15 years since they make lives more convenient due to their outstanding capabilities to meet users’requirements without neglecting their significant data storage capabilities. However, its great accessibility results in possible vulnerabilities. For example, vulnerabilities in user input functions. The most common attacks to exploit these vulnerabilities, according to OWASP Top Ten, are: SQL injection (SQLIAs) and cross site scripting (XSS). Therefore, any system or business organization of any kind is usually attacked by its users or third parties to access information with high confidentiality. Code analyzers can be implemented for entries made by users to prevent and discover SQL injection attacks. However, this technique is minimal against the ability of the attack. Therefore, this project allows the detection and prevention of SQL injection attacks through log analysis. In addition, its implementation was carried out with the CRISP-DM data mining methodology since it is the most common in the data mining area due to its flexibility and focus on detecting and data to find suspicious patterns. The algorithm implemented for the prediction was the SVM (Support Vector Machines), using the transaction log of the CSIRTEPN server as the primary information. For the data visualization interfaces, quantitative graphs were made using bar diagrams or others, which allow better visualization of whether the system has been under attack.