A Survey on Machine Learning and Deep Learning Techniques for Fake Review Detection

Abstract: Online reviews have become an essential source of information for consumers when evaluating products and services on digital platforms. However, the growing presence of deceptive or fake reviews has raised concerns about the reliability of these systems. Over the past decade, researchers have proposed various techniques to detect such reviews using natural language processing, machine learning, and deep learning methods. This paper presents a comprehensive survey of existing approaches for fake review detection, focusing on different categories of techniques including traditional machine learning models, neural network-based methods, and hybrid approaches that combine textual and behavioral features. The survey examines commonly used feature extraction techniques such as term frequency–inverse document frequency, sentiment analysis, and reviewer behavior analysis. It also discusses recent developments in deep learning models that capture contextual relationships within review text. In addition, the paper highlights key challenges faced in this domain, including limited availability of labeled datasets, data imbalance, and evolving strategies used by spammers. Finally, the study identifies important research gaps and suggests directions for future work, particularly in improving model interpretability and developing real-time detection systems. The findings provide a structured understanding of existing methods and their limitations, which can support further research in this area.

Keywords: Fake review detection, opinion spam, machine learning, deep learning, sentiment analysis, text mining, behavioral analysis, feature engineering, survey, review classification.