Detection of Phishing Attacks in Iranian E-Banking Websites Using a Neuro-Fuzzy System

The increasing accessibility of the Internet and the rise of e-commerce have led to a significant surge in online transactions facilitated by payment gateways. However, this convenience has also increased users’ vulnerability to cyberattacks, particularly phishing, where users are tricked into revealing sensitive information. To address this issue, this paper proposes an automatic phishing detection system using the Adaptive Network-based Fuzzy Inference System (ANFIS) and five key features from websites. The proposed system was evaluated using a dataset of 102 Iranian payment gateways, achieving high performance metrics with an accuracy of $\mathbf{9 9 \%}$, precision of $\mathbf{9 9 \%}$, recall of $\mathbf{9 9. 1 \%}$, and an $\mathbf{F 1}$-score of $\mathbf{9 8. 9 9 \%}$. The model demonstrates competitive accuracy while requiring fewer features than similar approaches. The methodology includes selecting relevant features, applying ANFIS, the use of SMOTE for balancing the dataset, and validating the model’s performance through rigorous testing.