Al-Driven Fraud Detection in E-Commerce Using Deep Learning and Graph Neural Networks

The rapid growth of e-commerce has transformed the global economy by enabling fast, accessible, and borderless digital transactions. At the same time, the increasing dependence on online platforms has intensified cybersecurity threats and created new opportunities for fraudulent activities. Modern e-commerce fraud extends beyond traditional payment theft and now includes account takeover attacks, synthetic identities, phishing campaigns, fake reviews, refund abuse, and bot-driven manipulation. Conventional rule-based fraud detection systems are no longer sufficient because they struggle to identify dynamic and evolving attack patterns in real time. In response to these challenges, Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL), and Graph Neural Networks (GNNs) have emerged as effective technologies for intelligent fraud prevention. This paper examines the role of AI-driven systems in modern e-commerce security, with particular emphasis on deep learning and graph-based intelligence. The study discusses the evolution of digital fraud, recent developments in AI-based detection techniques, behavioral analytics, explainable AI, and emerging cybersecurity challenges. In addition, a hybrid fraud detection framework is proposed that combines behavioral analysis, graph learning, anomaly detection, and deep learning to improve detection performance while reducing false positives. The findings indicate that intelligent hybrid AI architectures are becoming essential for securing next-generation digital commerce infrastructures.

کلیدواژه ها:

E-Commerce Security ، Fraud Detection ، Artificial Intelligence ، Deep Learning ، Graph Neural Networks ، Cybersecurity