Malicious URL Detection Using Machine Learning and Deep Learning Algorithms

Abstract

The majority of commercial enterprises rely on the World Wide Web's Internet services (WWW). These services are vulnerable to cyberattacks. The majority of cyberattacks occur when users click on malicious URLs. URL is an abbreviation for Uniform Resource Locator, which is the global address of documents and other resources on the World Wide Web. Malicious URLs are compromised URLs that are used for cyberattacks. URLs are used to access legitimate resources on the WWW. When used for other purposes, they endanger data availability, controllability, confidentiality, and integrity. The distribution of malware, illegal information, or illegal images via computers or networks is an example of cybercrime involving computer use to commit other crimes. Malicious URLs can be found in a variety of places, including pop-up windows, social media posts, emails, and texts. These links are created and shared by scammers in an effort to deceive users into clicking. People can be exposed to harmful software, viruses, and other dangerous stuff once they click on these websites. The proposed work in this paper considers multiclass Malicious URL detection and investigates the evaluation metrics of various Machine Learning and Deep Learning classifiers where the experimental results show that the highest performance of the Machine Learning Algorithm and Deep Learning Algorithm is Random Forest Classifier and Convolutional Neural Network (CNN) respectively. The experimental findings also indicate that the suggested URL features and behavior can considerably increase the capacity to recognize malicious URLs. This implies that the suggested methodology might be regarded as an effective method of identifying malicious URLs.