Bivariate Classification of Malware in Java']JavaScript using Dynamic Analysis

JavaScript is used as an attack vector to infect webpages to gain access to user's information. We present a tool that will dynamically analyze and perform bivariate classification of webpages as malicious or benign. We categorized the general behavior of JavaScript using datasets of known benign and malicious JavaScript by using a classifier which is trained on the basis of difference between function calls made by malicious and benign JavaScript and identification of Iframe tag in them. A Script is then matched to those categorizations to classify its behavior as malicious or benign. Here we have developed a light weight malicious JavaScript detection approach which can be used in real time as most of the existing techniques perform offline analysis.