基于多特征的JavaScript恶意代码检测方案

In recent years, the number of malicious code written in JavaScript has emerged in an endless stream, and the high-speed development of front-end technology has brought about malicious script attacks based on new technologies. Therefore, how to accurately identify various types of JavaScript malicious code is particularly important to ensure user security. This paper studies the JavaScript malicious script detection scheme based on machine learning, and improves the JavaScript malicious script detection scheme based on machine learning. It extracts multiple features from JavaScript code features, cross-features of HTML and new HTML5 malicious features. The classification model is constructed by comparing different classification algorithms. Finally, it is verified by experiments that the detection scheme can effectively and accurately detect JavaScript scripts and HTML pages containing malicious scripts.