PhiDMA – A phishing detection model with multi-filter approach

Abstract

Phishing remains a basic security issue in the cyberspace. In phishing, assailants steal sensitive information from victims by providing a fake site which looks like the visual clone of a legitimate site. Phishing shall be handled using various approaches. It is established that single filter methods would be insufficient to detect different categories of phishing attempts. This paper provides a multilayer model to detect phishing, titled as PhiDMA(Phishing Detection using Multi-filter Approach). The PhiDMA model incorporates five layers: Auto upgrade whitelist layer, URL features layer, Lexical signature layer, String matching layer and Accessibility Score comparison layer. A prototype implementation of the proposed PhiDMA model is built with an accessible interface so that persons with visual impairments shall access it without any barrier. The result from the experiment shows that the model is capable to detect phishing sites with an accuracy of 92.72%.