Detecting Malicious URLs using Data Analytics

Abstract

Phishing is a type of social engineering hack in which attackers trick victims into providing their login information on a form that sends the information to a malicious server. The fast developing subfield of data science is dependent on the application of machine learning as a central component. One vs All Classification is a strong machine learning technique that may be used to detect dangerous URLs by classifying them as belonging to one of several categories, such as malware, phishing, or spam. This algorithm can be used to detect malicious URLs by classifying them as belonging to one of these categories. Training numerous binary classifiers is required for this method. Each of these classifiers is responsible for differentiating one category from the others. The suggested approach makes use of One Vs All Classification in order to obtain a high level of accuracy in the detection of dangerous URLs. It is possible to train the system using a tiny dataset that just includes the pertinent aspects of the URLs, such as the domain and the path. In conclusion, one vs All Classification is a promising strategy for detecting malicious URLs. It can deliver high accuracy even with tiny datasets and low computational resources, which is a significant advantage over existing approaches.