Advancing Automated Social Engineering Detection with Oversampling-Based Machine Learning
Fatima Boumahdi, Mohamed Abdelkarim Remmide, Narhimène Boustia · International Journal of Security and Networks · 2024
Social engineering attacks have surged with the increased reliance on online interactions. However, detecting these subtle deceptions remains challenging. This study proposes a novel machine learning approach to enhance social engineering attack detection. We analyse well-known models (support vector machines and XGBoost) in imbalanced datasets and employ oversampling techniques (SMOTE-ENN) to address class imbalance issues. Experimental results demonstrate that the oversampled SVM model outperforms other techniques, achieving over 99% accuracy in attack detection. Statistical analysis via ANOVA confirms the significant improvement in detection performance compared to previous methods. This research contributes to the development of automated and reliable systems for identifying social engineering attacks, enhancing the security of online communications.