A Hybrid Intelligent Framework for Phishing Website Detection Using SVM and LightGBM

Abstract

Phishing websites continue to pose a serious cybersecurity threat by deceiving users into revealing sensitive information such as login credentials, banking details, and personal data. Traditional blacklist-based detection techniques are ineffective against newly created phishing websites, necessitating intelligent machine learning solutions. This paper presents PhishShield, a hybrid phishing website detection framework that integrates Support Vector Machine (SVM) and Light Gradient Boosting Machine (LightGBM) to accurately classify legitimate and phishing websites. The proposed approach utilizes URL-based feature extraction and text preprocessing to generate meaningful representations for classification. SVM provides robust decision boundaries, while LightGBM enhances predictive performance through efficient gradient boosting. Experimental evaluation demonstrates that the hybrid framework achieves higher accuracy, precision, recall, and F1-score compared to conventional machine learning models. The system is implemented as a web-based application capable of real-time URL analysis, enabling users to identify malicious websites before accessing them. The proposed framework offers an efficient, scalable, and reliable solution for strengthening web security against evolving phishing attacks.