URL-Based Phishing Attacks Detection Using Machine Learning

In this paper, we tackle the challenge of real-time detection of social engineering attacks, specifically focusing on phishing Uniform Resource Locator (URL) classification. We apply machine learning (ML) techniques to determine the most effective classification model. Multiple machine learning models, including k-nearest Neighbors (kNN), Naive Bayes, Decision Tree, Random Forest, and Logistic regression are implemented and evaluated based on their accuracy, recall values, F1-score and ROC-AUC values to identify the most effective classifier. Our results show that the Random Forest model achieves the highest accuracy among the selected ones, reaching 96 %. This underscores its potential as a reliable tool for mitigating social engineering threats in real-time scenarios. This model is integrated into an iOS application using Core ML for real-time phishing attack detection. The app enables users to enter URLs and immediately receive phishing or legitimate classifications, facilitating quick responses. This paper presents the potential of combining machine learning with mobile development to strengthen user security and proactively combat phishing attacks.