FRAUDULENT ONLINE AUCTION BIDDERS’ DETECTION SYSTEM USING NEURAL NETWORK

Abstract

The difficulties in distinguishing between genuine and fake bidders who manipulate auction prices by placing a deceptive bid has become a core challenge, as fraudulent activities like shill bidding can undermine the integrity of auction platforms. To ameliorate this issue, an AI detection model which is based on Neural Network has been developed to improve the accuracy and reliability of fraudulent bidding identification. The system is designed to analyse the bid patterns, feedback scores, bid amount, and bidding frequencies. The AI system was also trained using benchmark dataset that include both real and simulated bidding activities, which allow it to recognise subtle deviations from normal bidding behaviours. Various performance metrics such as accuracy, precision, recall, and loss function were deployed to evaluate the model’s performance. The results show that the Neural Network AI detection system achieved a high level of accuracy in identifying fake bidders, with overall accuracy rate of 94%. The loss function used was categorical Cross-entropy, which minimized the error in prediction during the training phase. The system demonstrated effective learning, as the loss values decreased significantly after several epochs of training, confirming that the model was optimized to detect fraudulent behaviour.