Detection and Behavioral Analysis of Suspicious Reviewer Networks in Online Reviews

Abstract

Online product review platforms have become an essential source of information for consumers when making purchasing decisions. However, the credibility of these platforms is increasingly threatened by the presence of coordinated extremist reviewer groups that intentionally manipulate product ratings and opinions. These groups often post overly positive or overly negative reviews to influence consumer perception, promote specific products, or damage competitors. Detecting and characterizing such extremist reviewer groups is therefore critical for maintaining the reliability and trustworthiness of online review systems. This study proposes a data-driven approach to identify and analyze extremist reviewer groups in online product review platforms. The proposed framework utilizes machine learning and behavioral analytics to detect abnormal reviewing patterns, sentiment extremity, temporal activity bursts, and reviewer collaboration networks. Features such as review polarity, reviewer similarity, review frequency, and rating deviation are extracted and analyzed to identify suspicious group behavior. Clustering and classification techniques are applied to distinguish extremist reviewer groups from genuine reviewers. Experimental evaluation on real-world product review datasets demonstrates that the proposed approach effectively identifies coordinated extremist reviewing activities with improved detection accuracy. The results highlight the importance of combining sentiment analysis, network analysis, and machine learning techniques for robust detection. This research contributes to enhancing the transparency and reliability of online review ecosystems by providing an effective framework for identifying and characterizing extremist reviewer groups.