Published Paper Details:

Abstract: Phishing is a major cybersecurity threat that tricks users into revealing sensitive information through deceptive websites. Traditional methods like blacklists and browser alerts often fail to detect newly crafted or obfuscated phishing URLs. This project presents a Phishing Detection System using Hybrid Machine Learning, which combines multiple supervised algorithms--Logistic Regression, SVM, Decision Tree, Random Forest, Gradient Boosting, and XGBoost--within an ensemble voting framework. URLs are converted into numerical vectors using TF-IDF, allowing the models to detect hidden patterns and anomalies. A majority voting mechanism enhances detection accuracy and system robustness. The solution is implemented with a user-friendly interface using Streamlit, enabling real-time URL classification without relying on external APIs or blacklists. It is lightweight, scalable, and capable of operating offline, making it ideal for integration into browsers, email systems, or enterprise security tools. Tested on real-world datasets, the system achieves over 95% accuracy while maintaining high performance under load. This project highlights the effectiveness of hybrid machine learning in phishing detection and offers scope for future improvements, such as deep learning integration and browser plugin development.