Phishing Website Detection Based on Multi-Feature Stacking
Qiang Hu, Hangxia Zhou, Qian Liu · 2021 2nd International Conference on Artificial Intelligence and Computer Engineering (ICAICE) · 2021
Aiming at the problems of low accuracy and high computational resource consumption of most current phishing website detection technologies, a stacking phishing website detection method combining four base learners is proposed. And an HTML string embedding feature is designed based on the Transformer to replace artificial features, which converts the HTML document of the website into a multi-dimensional vector, and combines filtered URL features to detect the phishing website. On the 100,000-level data set, the accuracy rate reaches 98.52%, and the F1 value reaches 98.81%. And after introducing the HTML string embedding features designed in this article, compared to only using URL features, each performance index has been significantly improved.