JavaScript Malicious Code Detection System

People are more inclined to use fast and convenient Internet service at this stage. At the same time, because people on their own internet security awareness is weak, or because the Internet has a strong open and vulnerable to attacks by hackers. Hackers have a variety of attacks, in numerous attacks, a very popular way to attack is the Java Script script directly as attack vectors, using browser vulnerabilities and plugin vulnerabilities, users of the Internet access to malicious attacks. This embedded in the web page Java Script script attacks on the Internet in the form of an enormous threat to the security of the internet.People on the Java Script script attacks used some methods, such as traditional static code based on the anti virus engine detection methods mainly used feature code matching technology to detect Java Script script Trojan. However, this method has some limitations, it is difficult to detect the malicious code after encryption, while the static characteristics of the library will continue to increase due to time changes, so that the detection performance significantly reduced. Or as dynamic detecting method based on execution, tracking Java Script code on the stack memory and to judge the existence of malicious operation, this method also has certain limitations, in the batch analysis, the efficiency is low. Therefore, it is needed to propose an efficient and static detection algorithm, which is based on the regression prediction algorithm based on machine learning, which can effectively detect the malicious Java Script script code.This paper proposes a Java Script script detection method, which can be achieved on the Java Script script malicious classification. This method firstly proposes a detection method using a dynamic and static combined detection model. Its implementation method is: The traditional static detection method is optimized, which is suitable for the characteristics of the detection system, optimize the algorithm of support vector machine in machine learning, and realize the high efficiency of the static detection method. Meanwhile it is a virtual execution based technique is proposed as a model of advanced detection. Finally, the efficiency and effectiveness of the method are proved by experiments.The detection system has a high detection performance, compared with the same kind of reverse virus engine, the test has the confusion encryption Java Script when the effect is more remarkable and provide browser, anti virus engine interface, has a high application value.