Study on Lexical Analysis of Malicious URLs using Machine Learning

The act of phishing has an impact all over the world in the past decade. The numbers of attacks are increasing day-by-day while gaining access over the internet. The attacks are done with minimum efforts and low cost. Due to this, the attackers can do maximum number of phishing attacks in less time. The protection of users connected to the internet is so cumbersome and hard. Detecting malicious URLs is one way to mitigate the act of phishing. A mandatory solution can be performed by using information extracted only from a URL. In this paper, the study consists of three phases: first phase includes lexical features evaluation from existing studies as well as novel features evaluation to find out an optimal feature vector, second phase includes classification algorithm identification which provides a classification model to compared with existing studies and the third phase includes the study of first two phases to produce and evaluate a scalable classification system which evaluates the malicious URLs obtained from various sources.