A Systemic Review of Machine Learning Approaches for Malicious URL Detection

New websites are emerging every day due to the growing popularity of the internet. No matter where you live, your occupation, or your age, web browsing has become an everyday activity for everyone. However, due to the growing internet use, website attacks have become common. A URL that contains hidden links is vulnerable and exploited by intruders for phishing, spam, DoS, DDoS, etc. Identifying and combating such malicious websites has been quite challenging due to the difficulty separating good from harmful websites. In this survey paper, various techniques used by researchers to detect malicious URLs are analyzed. Methods like the Blacklist method, Heuristic approach, and various research articles for malicious URL detection are discussed here. This paper presents malicious URL detection as a machine-learning task and categorizes and reviews literature studies that address the different aspects of the problem. Several well-known classifiers are discussed in this paper, including Naive Bayes, Support Vector Machines, Multi-Layer Decision Trees, and Random Forests, for detecting malicious URLs as a binary classification problem.