Optimized Feature Selection and Quantum Convolutional Neural Network Driven Web Spam and Phishing Detection Framework

ABSTRACT The increasing prevalence of phishing and web spam poses significant threats to online users and organizations alike. Traditional detection systems often struggle with scalability, adaptability, and accuracy in identifying evolving cyber threats. This paper proposes a novel web spam and phishing detection framework leveraging Quantum Convolutional Neural Networks (Quantum CNN) combined with Doll Maker Optimization (DMO) for efficient feature selection. The integration of these advanced deep learning and quantum computing techniques results in a system that not only improves detection accuracy but also enhances interpretability—a critical concern in security applications. The proposed model achieves a detection accuracy of 97%, outperforming existing methods like Hybrid Deep Learning (99%), Random Forest (93.7%), and AMALS (92.6%) in accuracy, specificity (98%), and sensitivity (96%). Furthermore, it demonstrates superior scalability, making it highly suitable for real‐time applications across web, social media, and email platforms. This paper's contributions represent a significant step forward in cybersecurity, offering a scalable, interpretable, and highly efficient solution for combating phishing and web spam. The framework's ability to handle large, diverse datasets and provide insights into detection decisions underscores its practical utility in modern cybersecurity landscapes.