AI-Powered Detection of Spear Phishing and Digital Arrest Attacks in E-Commerce
Phishing and digital arrest attacks pose significant threats to e-commerce security, necessitating advanced AI-driven detection mechanisms. In this context, this study proposed a hybrid deep learning model integrating BERT for feature extraction, Atrous Spatial Pyramid Pooling (ASPP) for multi-scale feature refinement, and a CNN-based classification network. The BERT encoder captures semantic and contextual features, while ASPP enhances spatial hierarchies using dilated convolutions. A CNN classifier with batch normalization and ReLU activation ensures robust classification. Extensive experiments on a Kaggle dataset demonstrate superior performance, achieving 97.81% accuracy, 98% precision, and an AUC of 0.9972, outperforming GRU, LSTM, RNN, and Transformer models.
Authors
- Shin‐Hung Pan (ORCID: https://orcid.org/0000-0002-1366-539X)
- Razaz Waheeb Attar (ORCID: https://orcid.org/0000-0001-9819-7945)
- Varsha Arya (ORCID: https://orcid.org/0000-0001-7549-4429)
- Ching‐Hsien Hsu
- Amal Hassan Alhazmi (ORCID: https://orcid.org/0000-0002-8525-0302)
- Ahmed Alhomoud (ORCID: https://orcid.org/0000-0003-2782-3106)
- Ayman Shawqi Alhazmi (ORCID: https://orcid.org/0009-0006-3143-6650)
Institutions
- Chandigarh University (IN)
- Princess Nourah bint Abdulrahman University (SA)
- Northern Border University (SA)
- Chaoyang University of Technology (TW)
- Asia University (TW)
- Umm al-Qura University (SA)
- University of Business and Technology (SA)
- University of Petroleum and Energy Studies (IN)
Publication Details
- Journal
- International Journal on Semantic Web and Information Systems
- Published
- 2025-06-29
- DOI
- https://doi.org/10.4018/ijswis.382475
- Primary Topic
- Spam and Phishing Detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00