Fake Review Analytics : A Supervised Machine Learning Approach

Abstract

Online reviews have become a cornerstone of e-commerce, wielding tremendous influence over consumers' decisions. Positive reviews can significantly bolster a product's reputation and drive sales. However, the flip side of this digital coin is the proliferation of fake or deceptive reviews, strategically crafted to deceive potential buyers. Consequently, the detection of fake reviews has evolved into a dynamic and critical research domain. Effectively identifying fake reviews relies on a nuanced understanding of both the inherent traits of reviews and the behaviors of the reviewers themselves. This study underscores the potential of machine learning as a formidable tool for fake review detection. To capture the multifaceted behaviors of reviewers, this research employs a diverse set of feature engineering techniques, complementing the process of feature extraction from the reviews themselves. In the pursuit of effective fake review identification, this study conducts a series of experiments, leveraging machine learning classifiers like K-Nearest Neighbors (KNN), Naive Bayes (NB), and Logistic Regression. These classifiers are assessed using a real dataset comprised of Yelp restaurant reviews. The results are enlightening, with Logistic Regression emerging as the top performer in terms of accuracy. These findings underscore the capabilities of machine learning algorithms in distinguishing genuine from fraudulent reviews, enhancing the trustworthiness of online review platforms and bolstering consumer confidence in the face of increasingly sophisticated fake review tactics.