Modern anti-spam email systems have faced a growing problem in the form of multi-modal campaigns that not only employ compelling textual information but also utilize structural manipulations; moreover, numerous existing high performing models lack explainability. In this research paper, we present a novel explainable approach based on gated fusion of text-based semantic features and metadata-related structural patterns, where the decision process is divided into three types of evidences – semantic-dominant, structural-dominant and mixed mode. For the evaluation of our proposed framework, the Assassin_clean data set has been selected and five different random seeds have been tested. Our results indicate that among the proposed neural models, Variant F performed best, scoring an average F1 score of 0.9712 and an average AUC score of 0.9971; additionally, it outperformed Variant E and demonstrated comparable results with respect to linear SVM. It was also noted that the mode of explanation might differ across the random seeds without deteriorating the predictive performance.
E. H. Tusher, M. A. Ismail, M. A. Rahman, A. H. Alenezi, and M. Uddin, “Email spam: A comprehensive review of optimization techniques in detection systems,” IEEE Access, vol. 12, pp. 1-24, 2024.
K. I. Roumeliotis et al., “Next-generation spam filtering: Comparative fine-tuning of LLMs, NLPs, and CNN models for email classification,” Electronics, vol. 15, no. 3, p. 630, 2026.
R. Meléndez, M. Ptaszyński, and F. Masui, “Comparative investigation of traditional machine-learning models and transformers for phishing and spam detection,” Applied Sciences, vol. 15, no. 6, p. 3396, 2025.
P. H. Kyaw et al., “A systematic review of deep learning techniques for phishing email detection,” Electronics, vol. 14, no. 5, p. 812, 2025.
J. L. Wilk-Jakubowski et al., “Machine learning and neural networks for phishing detection: A systematic review,” Sensors, vol. 24, no. 9, p. 2951, 2024.
A. Alhuzali, A. Alloqmani, M. Aljabri, and F. Alharbi, “In-depth analysis of phishing email detection: Evaluating the robustness of modern deep learning approaches,” Information, vol. 15, no. 1, p. 45, 2024.
H. Asliyuksek et al., “A comparative evaluation of a multimodal approach for spam email classification using DistilBERT and structured metadata,” Computers & Security, vol. 148, Art. no. 104312, 2025.
M. Hosseinzadeh et al., “Improving phishing email detection performance through hybrid deep learning architecture and feature fusion,” Expert Systems with Applications, vol. 254, Art. no. 124565, 2024.
C. Patra et al., “Phishing email detection using vector similarity search and transformer-based word embedding,” Computers, Materials & Continua, vol. 82, no. 3, pp. 5229-5250, 2025.
M. Murhej et al., “Multimodal framework for phishing attack detection and mitigation through behavior analysis using deep learning,” Engineering Applications of Artificial Intelligence, vol. 134, Art. no. 108791, 2024.
S. Rashed and C. Ozcan, “A novel dual-layer deep learning architecture for phishing and spam email detection,” Electronics, vol. 14, no. 2, p. 205, 2025.
M. A. Uddin et al., “An explainable transformer-based model for phishing email detection: A large language model approach,” IEEE Access, vol. 14, pp. 1-18, 2026.
A. Sharma, S. Rani, and M. Shabaz, “A comprehensive review of explainable AI in cybersecurity: Decoding the black box,” Journal of Information Security and Applications, vol. 80, Art. no. 103796, 2024.
M. A. Uddin, M. N. Islam, L. Maglaras, H. Janicke, and I. H. Sarker, “ExplainableDetector: Exploring transformer-based phishing email detection with interpretable evidence,” IEEE Access, vol. 13, pp. 1-16, 2025.
M. Ibrahim et al., “Phishing email detection using BERT and RoBERTa,” Computation, vol. 14, no. 2, p. 46, 2026.
A. M. Vulfin et al., “A multimodal phishing website detection system using explainable artificial intelligence techniques,” Knowledge-Based Systems, vol. 298, Art. no. 112019, 2024.
Y. Kim, “Convolutional neural networks for sentence classification,” in Proc. EMNLP, pp. 1746-1751, 2014.
S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Computation, vol. 9, no. 8, pp. 1735-1780, 1997.
D. Bahdanau, K. Cho, and Y. Bengio, “Neural machine translation by jointly learning to align and translate,” in Proc. ICLR, 2015.
D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in Proc. ICLR, 2015.
