In the past years, the Internet has been widely used for a variety of applications. Although the systems have become more complex, have also become more complex and widespread, This article aims to detect SQL injection attacks by focusing on extracting effective features from several appropriate data sets, and tries to recognize these types of attacks more effectively. In this context, using machine learning algorithms, including logistic regression and random forest, he has performed various experiments on three data sets and tested manual and automatic feature selection processes with the relevant algorithms. It has also evaluated whether some features are more important than others and displayed the results. From the results obtained in this research, it was found that feature extraction from data sets with high specialized knowledge in this field and selection of optimized basic tables to determine features leads to high accuracy results and we reached 99.57% accuracy
)