A Hybrid Approach for Detecting Phishing URLs: Integrating Rule-Based and Machine Learning Techniques

Phishing attacks continue to pose a major cybersecurity threat by taking advantage of weaknesses in internet platforms and tricking users into revealing confidential data. Although traditional Rule-Based detection methods were effective in detecting particular patterns in phishing URLs, they had difficulty with attack techniques that were constantly changing and becoming more complex. Machine Learning methods provided flexibility, but they frequently encountered challenges in efficient resource usage and immediate decision making.This research introduces a blended detection framework that combined the benefit of Rule-Based analysis in deciding in real-time and the flexibility of the Machine Learning Random Forest Classifier to improve the accuracy and dependability of detecting phishing URLs. A collection of benign and malicious URLs was utilized to test the framework, and the combined detection approach demonstrated strong performance metrics with 95.38% accuracy, 95.42% precision, 95.33% recall, and 95.37% F1-score. This showed its effectiveness in reducing false negatives, while ensuring consistent detection of malicious URLs.This research advanced phishing detection technologies by effectively addressing Rule-Based detection shortcomings and optimizing Machine Learning integration, providing a scalable and efficient solution to combat cyber threats.