SQL injection detection method based on convolution neural network algorithm

Abstract

The invention discloses a SQL injection detection method based on a convolution neural network algorithm. The method comprises the following steps: step one, performing text word-segmentation processing; step two, extracting a text vector: performing vocabulary model training on a training sample after the word segmentation processing by using a CBOW algorithm in a Word2Vec tool, converting the text data into the text vector; step three, training a detection model: designing a convolution neural network structure; selecting a convolution layer, a pooling layer and an activation function as parameters; inputting the text vector extracted from the step 2 into the convolution neural network to perform model training to obtain a detection model; step four, performing SQL injection detection: inputting the to-be-detected SQL injection instance into the detection model of the step three, or abandoning the data packet. Through the SQL injection detection method disclosed by the invention, the feature vector of an attack sample is extracted through the own characteristics of the CNN algorithm, the false alarm rate of the loophole detection is reduced, and a recognition rate for the variation attack is improved.

Description

technical field [0001] The invention relates to a method for detecting SQL injection based on a convolutional neural network algorithm. Background technique [0002] With the rapid development of computer technology and the Internet, Web applications have risen rapidly, and network security issues have attracted much attention. SQL injection vulnerability attack is one of the most popular hacking script attack methods on the Internet. SQL injection has the characteristics of high attack damage, multiple types, fast mutation, and hidden attack. Therefore, the detection and defense of SQL injection vulnerabilities has always been the focus of attention in the field of web security. [0003] There are many SQL injection methods, which can be divided into joint query injection, error injection, tautology attack, etc. based on different attack methods. The main principle of the SQL injection detection model is to intercept the communication data between the client and the Web s...

AI Technical Summary

Problems solved by technology

For example, Li Hongling et al. proposed the SVM algorithm to train the injection detection model in terms of solving the applicability of the detection method and improving the accuracy of injection detection; Yang Lianqun et al. proposed to combine the hidden Markov model to reduce the false positive rate of SQL injection detection; Zhang Zhichao et al. A BP neural network training detection model is proposed. These detection features are fast and efficient, but its recognition rate for mutation attacks is low

[0004] In order to solve the problem of the low detection rate of mutation attacks, the inventor proposed a SQL injection detection method based on convolutional neural network algorithm

Images