A Systematic Review: Detecting Phishing Websites Using Data Mining Models

Abstract

As internet technology use is on the rise globally, phishing constitutes a considerable share of the threats that may attack individuals and organizations, leading to significant losses from personal and confidential information to substantial financial losses. Thus, much research has been dedicated in recent years to developing effective and robust mechanisms to enhance the ability to trace illegitimate web pages and to distinguish them from non-phishing sites as accurately as possible. Aiming to conclude whether a universally accepted model can detect phishing attempts with 100% accuracy, we conduct a systematic review of research carried out in 2018–2021 published in well-known journals published by Elsevier, IEEE, Springer, and Emerald. Those researchers studied different Data Mining (DM) algorithms, some of which created a whole new model, while others compared the performance of several algorithms. Some studies combined two or more algorithms to enhance the detection performance. Results reveal that while most algorithms achieve accuracies higher than 90%, only some specific models can achieve 100% accurate results.