Int J Performability Eng ›› 2018, Vol. 14 ›› Issue (12): 3167-3173.doi: 10.23940/ijpe.18.12.p26.31673173

Previous Articles     Next Articles

Method for Detecting Javascript Code Obfuscation based on Convolutional Neural Network

Wei Jiangab*(), Huiqiang Wanga, and Keke Wua   

  1. a College of Computer Science and Technology, Harbin Engineering University, Harbin, 150080, China
    b College of Computer Science and InformationEngineering, Harbin Normal University, Harbin, 150080, China
  • Revised on ; Accepted on
  • Contact: Jiang Wei


Malicious webpage attacks occur frequently, and most of the JavaScript attack code is implemented through obfuscation. In order to further confront malicious webpage attacks, detecting JavaScript obfuscation scripts has become one of the most urgent issues to be addressed. This paper proposes a method for detecting JavaScript code obfuscation based on Convolutional Neural Networks (CNNs). Firstly, the character matrix feature method of Bigram is used to extract features of JavaScript code. Secondly, a CNN model is applied to the JavaScript code obfuscation detection, which overcomes the high requirement of the machine code learning and the low accuracy of the obfuscation feature extraction of JavaScript code. Finally, the simulation results show that this method can not only reduce the requirements for the features, but also effectively improve the accuracy of the detection of JavaScript code obfuscation.

Key words: JavaScript code obfuscation detection, convolutional neural networks, character matrix feature extraction, bigram