期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2021
卷号：99
期号：1
页码：147
出版社：Journal of Theoretical and Applied
摘要：Impacting businesses across the world, phishing remains today to be a serious problem: due to anonymous access to personal details, businesses and their consumers deal with the problems materialised out of fishing attacks, huge financial loss being one of these problems. Because of this, phishing needs to be identified and dealt with efficiently using intrusion detection techniques; such mechanisms are yet to be used. It is within this paper that, with the use of a newly-arisen method (Phishing Multi-Class, founded on the grounds of Association Rule) we will study the issue of predicting phishing websites. So as to weigh up the successful use of data mining algorithms using a publicly available dataset involving 10,068 incidents of legitimate and phishing websites, two experimental studies were conducted, in which the classifier model was built. In the first of these two studies, the capability of PMCAR (Phishing Multi-Class Association Rule) compared to three associative classification algorithms (CBA, MCAR, and FACA) was examined; additionally, five benchmark algorithms (SVM, LR, DT, and ANN) were assessed in the second experiment, so as to generalise the competence of utilising data mining for resolving the phishing websites detection issue. As a result of conducting these experiments, all data mining algorithms that were evaluated predict phishing websites with decent classification rate; and so we can conclude that, when looking to tackle the issue of predicting phishing websites, these can be successful methods.
