Malicious web page detection based on feature classification

摘要

Malicious code detection is a major concern in computer science community in this decade. With the rapid growth of web applications, web sites have been become the attacker's main target. Innocent users' machines become compromised by just visiting a malicious page. This paper presents a malicious web page detection based on static feature classification. We classified features into three groups: explicit features, replicated features, and miscellaneous features. We employed Greasemonkey to develop the detection script. It provides the alert when an innocent user is visiting a malicious page. The accuracy of our detection system is 97.9% with 1.42 % of false positive and 2.76% of false negative. The average detection time is 2.49 seconds per page.