Detection and classification of malicious URLs based on machine learning models

Abstract

There has been a shift in the last few years that saw a rapid increase in eservices that need the transmission of sensitive and private information over the internet. Along with the number of people utilizing the internet, crime-attacks are expanding quickly. Malicious attacks using malicious URLs are becoming the simplest way to compromise the security chain's weakest link. Machine learning is an extremely well-liked method for identifying malicious or fake URLs. Tree based classification ML models may be utilized to detect non-benign URLs by training them by utilizing a dataset that contains common malicious URLs. This paper explains the lexical features in a URL that may be utilized to train the machine learning models. It also calculates the accuracy rate of different models and selects one for prediction based on the calculated accuracies.