A Hybrid Phishing Website Detection Framework Using URL Feature Analysis and Visual Intelligence
Author(s)
Abstract
Phishing attacks have become one of the most significant cyber security threats in recent years, targeting users by creating fraudulent websites that closely resemble legitimate platforms. Traditional detection methods such as blacklist-based systems and URL analysis are insufficient to detect newly emerging and visually deceptive phishing websites. This paper proposes a Hybrid AI-Based Phishing Website Detection Framework that integrates both URL feature analysis and visual intelligence. The system extracts structural features from URLs and analyses them using machine learning techniques, while webpage screenshots are processed using a Convolutional Neural Network (CNN) to identify phishing patterns. A hybrid decision engine combines the outputs of both models to generate the final classification. The proposed system is implemented as a browser extension that enables real-time phishing detection and provides risk scores along with explanations. Experimental results show that the hybrid model achieves an accuracy of 94.3%,outperforming traditional single-method approaches. This approach enhance cyber security by providing a more reliable and intelligent phishing detection system
Keywords