A Superficial Learning Strategy to Identify Email Spams Based on Hybrid Data Filtration and Classification Methodology

Spam email detection is a critical task in improving email security, especially with the increasing volume of spam emails that threaten users' privacy and productivity. In this study, we propose a hybrid approach combining Gated Recurrent Units (GRU) and Support Vector Machine (SVM) to identify spam emails effectively. The GRU component captures temporal dependencies in the email text, while SVM classifies the extracted features into spam or non-spam categories. The model was evaluated using various performance metrics, including accuracy, precision, recall, and F1-score, and compared with other models like standalone GRU, SVM, CNN, and Naive Bayes. The proposed hybrid GRU-SVM model achieved a superior accuracy of 96.8%, outperforming GRU (93.2%), SVM (91.5%), CNN (94.1%), and Naive Bayes (89.4%). Additionally, the model demonstrated a high true positive rate (98.2%) and a low false negative rate (1.8%) on a spam-heavy dataset, indicating its robustness in identifying spam emails. The results suggest that the hybrid model can significantly enhance spam email detection systems by improving classification accuracy and reducing false positives.