Enhancing Review Spam Detection with a Hybrid Approach Integrating Association Rule Mining and Convolutional Neural Networks

This study presents a novel spam review detection framework for eCommerce platforms that combines Association Rule Mining with Convolutional Neural Networks (CNN) to tackle the growing issue of digital review spam. As spam tactics evolve, traditional detection methods often struggle with accuracy, leading to high false positives or negatives. Our hybrid model harnesses the pattern recognition power of CNNs and the analytical strengths of Association Rule Mining to improve spam review detection. Trained on a comprehensive dataset of product reviews, the model was evaluated on metrics such as accuracy, precision, recall, and F1 score, achieving a notable accuracy of 94.5% and outperforming traditional and standalone machine learning approaches. This research not only introduces a robust solution for detecting spam reviews but also suggests potential for future studies to enhance this model across different languages, platforms, and real-time applications, contributing to safer and more reliable eCommerce environments.