Detecting malicious URLs using machine learning techniques

The World Wide Web supports a wide range of criminal activities such as spam-advertised e-commerce, financial fraud and malware dissemination. Although the precise motivations behind these schemes may differ, the common denominator lies in the fact that unsuspecting users visit their sites. These visits can be driven by email, web search results or links from other web pages. In all cases, however, the user is required to take some action, such as clicking on a desired Uniform Resource Locator (URL). In order to identify these malicious sites, the web security community has developed blacklisting services. These blacklists are in turn constructed by an array of techniques including manual reporting, honeypots, and web crawlers combined with site analysis heuristics. Inevitably, many malicious sites are not blacklisted either because they are too recent or were never or incorrectly evaluated. In this paper, we address the detection of malicious URLs as a binary classification problem and study the performance of several well-known classifiers, namely Naïve Bayes, Support Vector Machines, Multi-Layer Perceptron, Decision Trees, Random Forest and k-Nearest Neighbors. Furthermore, we adopted a public dataset comprising 2.4 million URLs (instances) and 3.2 million features. The numerical simulations have shown that most classification methods achieve acceptable prediction rates without requiring either advanced feature selection techniques or the involvement of a domain expert. In particular, Random Forest and Multi-Layer Perceptron attain the highest accuracy.