Domain Detector - An Efficient Approach Of Machine Learning For Detecting Malicious Websites

Abstract

Phishers employ social engineering and mimic sites to trick users and organizations into divulging personal details such as account IDs, usernames, and passwords. Phishing URL detection, hence, in the face of this is of paramount significance. Machine learning and deep learning algorithms have been created to identify phishing URLs automatically. We use a Gradient Boosting Classifier which has been trained on a wide range of features and an extremely large corpus of data in our process. This enables the system to learn in real-time, reacting to new threats by incorporating recently detected phishing techniques, actual domain changes, and notes by experienced analysts. Our system analyzes the content of sites for harmful patterns and adds reputation-based features like domain age to aid in detection. With such sophisticated means, our system is highly resistant to phishing attacks preventing loss of funds and safeguarding confidential information.