Explainable phishing website detection for secure and sustainable cyber infrastructure.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• Phishing attacks are a growing cybercrime threat impacting various sectors.
• Traditional phishing detection methods (blacklist, heuristic) are insufficient and resource-intensive.
• Previous research often neglects the importance of feature selection in detection models.

Purpose of the Study:

• To improve phishing website detection by applying feature selection techniques.
• To enhance model interpretability and precision using Shapley Additive Explanations (SHAP).
• To evaluate the effectiveness of different machine learning models for phishing detection.

Main Methods:

• Utilized a dataset of 11,000+ URLs with 30 features for phishing website detection.
• Applied feature selection techniques, specifically SHAP, to enhance URL-based detection models.
• Trained and tested Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Logistic Regression (LR), and K-Nearest Neighbor (KNN) models.

Main Results:

• The Random Forest model achieved the highest accuracy at 97%.
• SHAP analysis highlighted the most important features for improved detection and interpretability.
• Performance was evaluated using accuracy, precision, recall, and F1 score.

Conclusions:

• The proposed system provides an accurate and interpretable solution for phishing detection.
• Feature selection with SHAP significantly enhances the performance of machine learning models.
• The study contributes to a safer digital environment by improving defenses against phishing attacks.