Online marketplaces increasingly suffer from deceptive product reviews that manipulate customer perception and distort purchasing decisions. Traditional rule-based and manual moderation approaches struggle to detect sophisticated opinion spam, especially as review volumes grow exponentially across e-commerce platforms. The proposed system, Fake Product Review Detection Using Machine Learning, introduces an automated text analytics pipeline for identifying deceptive reviews using supervised learning techniques. The system processes raw review text through data preprocessing stages including tokenization, stop-word removal, normalization, and stemming, followed by feature extraction using TF-IDF vectorization. Multiple classification algorithms such as Logistic Regression, Naïve Bayes, and Support Vector Machine (SVM) are evaluated to determine optimal performance. A trained model is integrated into a Flask-based web application that enables real-time review classification as Fake or Genuine. The system architecture ensures seamless interaction between preprocessing, feature engineering, model inference, and user interface components. Performance evaluation conducted on a labeled dataset demonstrates an accuracy of 85%, with balanced precision and recall values, confirming reliable detection capability. The modular Python-based implementation ensures scalability, maintainability, and ease of deployment on standard computing environments. This approach enhances trustworthiness in online review ecosystems by providing an efficient, intelligent, and automated fake review detection solution.
Sivasankari et al. (Thu,) studied this question.