PhishSKaPe: A Content based Approach to Escape Phishing Attacks
Ankit Kumar Jain, Sakshi Parashar, Palak Katare, Isha Sharma
Abstract
A type of cyberattack, viz. phishing attack exposes a user’s credentials to malicious entities leading to breach of privacy. The proposed approach involves employing a search engine to match the domain name of the site under scrutiny with the sites that come up as result of our search query. Here, our search query is comprised of features viz. domain name and keywords consisting of title, meta description data and body text. We initially see how conventional TF-IDF (Term Frequency-Inverse Document Frequency) performs in classifying websites as phishing or legitimate. We then apply a weighted heuristic proposed in detail in this report by assigning different weights to different tag data and manipulate the TF-IDF score accordingly to improve the performance of our phishing detector. Experiments results illustrate that proposed search based method with weighted TF-IDF performs better than conventional TF-IDF.