본문 바로가기
  • Home

Explainable Multi-View Ensemble Model for Phishing Website Detection

  • Journal of Internet of Things and Convergence
  • Abbr : JKIOTS
  • 2025, 11(4), pp.143~149
  • Publisher : The Korea Internet of Things Society
  • Research Area : Engineering > Computer Science > Internet Information Processing
  • Received : July 12, 2025
  • Accepted : August 12, 2025
  • Published : August 31, 2025

Hye-won Hong 1 Ji-hee Park 1 Sanghoon Jeon 1

1수원대학교

Accredited

ABSTRACT

This study proposes a multiview ensemble-based phishing detection model to effectively respond to sophisticated phishing website attacks. A total of 113 features were categorized into six groups: URL, domain, directory, file, parameter, and network. Each group was independently trained using the LightGBM. The results were then integrated using a soft voting mechanism. The proposed model demonstrated superior performance over the individual models, achieving an accuracy of 96.0% and an F1-score of 94.9%. In addition, the SHAP-based explainable AI (XAI) technique was employed to quantitatively analyze and visualize the contribution of each feature group, thereby enhancing model interpretability and reliability. Generalization experiments on external datasets confirmed a stable detection performance, demonstrating the model’s applicability and scalability in real-world environments. We expect that the explainable multiview ensemble model developed in this study will contribute to the detection and prevention of phishing sites.

Citation status

* References for papers published after 2024 are currently being built.