A Markov Detection Tree-Based Centralized Scheme to Automatically Identify Malicious Webpages on Cloud Platforms

Abstract

The effective detection of malicious webpages plays a paramount role in ensuring the Web security on the Internet. However, the detection results of current methods are poor and their efficiency is low, and thus, it is important and challenging to design an efficient detection scheme that can improve the accuracy of classification of malicious webpages. To overcome this challenge, a Markov detection tree scheme is proposed in this paper to automatically identify and classify malicious webpages, where the link relations of unified resource locators, the information gain ratio, and Markov decision process as well as decision tree are used to analyze malicious webpages simultaneously. To increase the detection accuracy for malicious webpages, two methods of filling missing values are presented to process the null attribute values of webpages. We compare the performance of our algorithms when the different methods are applied in terms of the information gain ratio, classification accuracy, and detection efficiency. Our experimental results show that the proposed methods can improve the accuracy and efficiency in the classification of malicious webpage detections.

Keywords

• Decision tree
• Markov decision process
• malicious web detection
• machine learning