Automated Detection of Deceptive Online Product Reviews Using Supervised Learning Techniques

The rapid proliferation of e-commerce platforms has made online product reviews a central determinant of consumer purchasing behavior. However, the emergence of deceptive or fake reviews—generated by bots, paid agents, or competitors—significantly threatens the reliability of such feedback. This paper presents a comprehensive machine learning–based system for the automated detection of fake product reviews leveraging Natural Language Processing (NLP) techniques. Using a balanced dataset of 40,432 reviews (20,216 genuine labeled CG; 20,216 deceptive labeled OR) spanning 10 Amazon product categories, the proposed system applies a structured NLP pipeline comprising tokenization, stop-word removal, stemming, and lemmatization for text normalization. Feature extraction is performed using Bag-of-Words (CountVectorizer) and TF-IDF representations. Six supervised classification algorithms—Logistic Regression, Random Forest, Decision Tree, Naïve Bayes, K-Nearest Neighbors (KNN), and Support Vector Machine (SVM)—are systematically trained, evaluated, and compared. Experimental results demonstrate that SVM achieves the highest classification accuracy of 88.5%, significantly outperforming all other models. Comprehensive evaluation using accuracy, precision, recall, and F1-score metrics confirms the robustness of the proposed framework. The study also includes a real-time review classification module, demonstrating its readiness for integration into live e-commerce moderation pipelines.