Phishing Website Detection Using Hybrid Multi-Feature Classification
Abstract
Phishing is stealing sensitive information from users like, passwords, usernames, & debit/credit card details in a fraudulent way. Detection and Classification of such Phishing Websites are generally inefficient due to Low Accuracy rate and High False Positive rate with Novel Phishing Techniques. Hence, the need for development of Hybrid Multi-Feature based Prediction and Classification System plays a vital role in Preventing and Safeguarding users from online theft, fraud, and espionage. The Proposed System consists of Hybrid Prediction model using Random Forest and Logistic Regression Algorithms which are used in the Prediction of Phishing URLs more efficiently. A Decision Algorithm Module is used to check for URL/Domain feature-based detection which is incorporated with an URL Feature Classification model which uses XGBoost for classifying the Phishing URL based on the Domain based Features, HTML & Javascript based Features and Address Bar based Features of the URLs. The Hybrid system would detect phishing URLs more efficiently.