A State-of-the-Art Review on Phishing Website Detection Techniques
Abstract
Phishing websites are a major cybersecurity threat because they are designed to deceive users into sharing sensitive information, such as passwords, banking details, and personal data. These websites often imitate legitimate websites, making them difficult to identify through visual inspection alone. This paper presents a state-of-the-art review of techniques used to detect phishing websites. The review examines different approaches, including blacklist-based detection, heuristic-based methods, Machine Learning (ML), and Deep Learning (DL). It also discusses important website features, such as URL structure, domain information, webpage content, hyperlinks, and HTML characteristics, that help distinguish phishing websites from legitimate ones. Commonly used algorithms include Support Vector Machine (SVM), Decision Tree, Random Forest, XGBoost, Artificial Neural Networks (ANN), and Convolutional Neural Networks (CNN). The review further explores the applications, advantages, limitations, and evaluation metrics of these techniques. Existing research shows that AI-based detection methods have significant potential to improve phishing website identification and provide faster protection against online threats. However, challenges such as newly created phishing websites, changing attack strategies, imbalanced datasets, zero-day attacks, and model generalization remain important concerns. Future research should focus on real-time detection, explainable AI, updated and diverse datasets, lightweight detection systems, and hybrid approaches that combine multiple techniques to improve accuracy and reliability