Phishing classification models: Issues and perspectives
Hiba Zuhair, Ali Selamat · 2017
Never-ending phishing threats on cyberspace motivate researchers to develop more proficient phishing classification models to survive a supreme cyber-security with safe web services. However, such achievements remain incompetent in their performance against novel phish attacks. This is attributed to the induction factors of the classification model itself such as hybrid feature space, inactive learning on up-to-date data flows, and limited adaptation to the evolving phish attacks. In this light, this paper surveys the current achievements, studies their limitations, restates what induction factors need to boost for a successful real-time application. Consequently, future outlooks are recommended on how to devote well-performed anti-phishing scheme.