Enhanced Fake Review Detection: An Ensemble Approach Integrating Stylistic Analysis, Contextual Insights, And Attribute Correlation

This paper presents a robust ensemble technique that integrates attribute correlation with contextual and stylistic evaluations to enhance fake review identification. Our approach significantly boosts review classification precision by combining powerful machine learning models, including Random Forest, Support Vector Machine (SVM), and Logistic Regression, with advanced anomaly detection techniques such as One-Class SVM and Isolation Forest. The results demonstrate a notable increase in detection accuracy from 89.1% to 96.9% and an improvement in F1 Score from 0.88 to 0.97 through the application of deep learning architectures like Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Self-Organizing Maps (SOM). This method shows strong capability in distinguishing between authentic and counterfeit reviews. Future research will explore real-time implementation, feature set extension, optimization via hyperparameter tuning, and validation across various review platforms. The goal is to foster a more transparent and credible online review ecosystem by addressing ethical concerns and incorporating user feedback.