JavaScript Malicious Codes Analysis Based on Naive Bayes Classification
Yongle Hao, Hongliang Liang, Daijie Zhang, Qian Zhao, Baojiang Cui · 2014 Ninth International Conference on P2P, Parallel, Grid, Cloud and Internet Computing · 2014
Given the security threats of JavaScript malicious codes attacks in the Internet environment, this paper presents a method that uses the Naive Bayes classification to analyze JavaScript malicious codes. The method uses many malicious and normal sample data, and trains the classifier using extended API symbol features with a high degree of predictability of malicious codes, which contain variable names, function names, string constants and comments extracted from the JavaScript codes. Experiments show that the analysis method of JavaScript malicious codes is effective and achieves high accuracy.