Machine Learning-Based Phishing Detection System
Abstract
Phishing attacks are still one of the most important threats to cybersecurity, exploiting human weaknesses to illicitly obtain sensitive information such as credit card numbers, personal data, and passwords. These attacks are generally carried out by misleading emails or websites that mimic legitimate sources, which can have severe consequences, such as financial losses, identity theft and data breaches within the organization. To address this growing concern, we have developed a phishing detection system using a random forest (RF) model. The model has been trained on significant Mendeley datasets_2020 and has demonstrated considerable advantages in accurately detecting phishing attempts. By analyzing the critical features of the site's URL, the system can distinguish between legitimate and malicious sites. Our comprehensive evaluation showed a high 99.4% accuracy and makes it a reliable tool for phishing detection. We have integrated the system into Chrome's web extension, allowing real-time detection and improving user protection. The paper highlights the potential of machine learning in cybersecurity and offers opportunities for future research and development to improve phishing detection through advanced ML techniques and larger, more diverse datasets.