关键词: 恶意URL, 投票策略, 机器学习, 分类算法, 多层过滤模型

Abstract: Malicious URL detection is always a hot research topic in the field of Web security. This paper proposes a malicious URL multilevel filtering detection model. This model contains 4 layers of filter: black and white list filter, Naive Bayesian filter, CART decision tree filter and Support Vector Machine filter. In this paper several key strategies of multilayer filtering model are discussed, including support vector machine filter layer voting strategy; filter order strategy and filtering threshold tuning strategy. Filter voting strategies are discussed in separate voting, parallel voting and weighted parallel voting three voting methods. The filter order strategy discusses the order of the four filters. Filter threshold strategy discusses the method of determining the threshold of the filter. The validity of the above methods is verified by experiments. According to the experimental results, this paper implements a Web application.

Key words: malicious URL, voting strategy, machine learning, classification algorithm, multi layer filtering model