MALICIOUS QUERY RECOGNITION USING MACHINE LEARNING TECHNIQUES

Given the variety of database systems and the growing sophistication of attacks, identifying fraudulent SQL queries is a difficult and constantly changing cybersecurity task.Using domain-informed characteristics, this research proposes a machine learning-based system for identifying and categorizing fraudulent requests.GitHub and other public sources were used to generate a dataset of 88,213 malicious and benign SQL queries.Comprehensive data preparation, feature engineering, and the use of several classifiers, such as Random Forest, Support Vector Machine, Naive Bayes, and Multi-Layer Perceptron, were all part of the research.The Naive Bayes classifier outperformed the other models, with the maximum accuracy of 99% and sensitivity of 97%.The suggested solution shows how well expert-driven features and conventional machine learning methods may be used to detect SQL injection threats.The results highlight how these methods can improve web application security and provide a solid foundation for further research on different kinds of intrusions.