Hybrid Neural Framework for Cloud E-Commerce Security: Integrating LSTM-OCSVM, DTW-NN, and PSL

Online shopping is changing through the emergence of cloud-based marketplaces. This provides flexibility and efficiency. However, heavy usage of these has resulted in a high level of cybersecurity threats, such as fraud, data breaches, and unauthorized access. This work proposes a Hybrid Neural System, which builds upon DTW-NN behavior alignment, PSL logical potential thinking, and LSTM-OCSVM anomaly detection. This integrates adaptive and accurate real-time cyber risk mitigation to improve cloud commerce security. The proposed framework achieves 93.1% accuracy, 92.5% precision, 94.0% recall, 93.3% F1-score, and 94.5% AUC-ROC compared with traditional methods, thus being more effective in fraud detection and anomalies. The deep learning, sequential analysis, and potential reasoning allow the system to be more robust in security aspects, minimize false alerts, and adapt to changes in cyber threats, which improves the reliability of the platform and user trust in cloud-based transactions.