An exploratory study of fake Chinese reviews: detection and analysis of non-reviews

Purpose This study explores the influence of fake reviews, with a specific focus on non-reviews, on consumer decision-making and the credibility of e-commerce platforms. Utilizing advanced natural language processing (NLP) and machine learning techniques, the research develops a detection model designed to identify and filter irrelevant reviews, thereby strengthening the reliability of online review systems and fostering consumer trust. Design/methodology/approach This study employs heterogeneous review corpora to implement NLP and machine learning techniques, including text feature analysis, corpus construction and classification models. Data were collected from Google business reviews via web scraping. The methodology encompasses data preprocessing, feature extraction and model training using support vector machine (SVM), Logistic Regression and Naive Bayes classifier, with performance evaluated through confusion matrices and F1 scores. Findings The study confirms that the proposed model effectively identifies non-reviews among fake reviews with a high degree of accuracy. Experimental results reveal that the Naive Bayes classifier achieves an F1 score as high as 0.997, with exceptional performance in hotel reviews. Moreover, the findings highlight the superiority of the bag-of-words model in capturing intricate review details and effectively detecting fake reviews. By defining opinion sentences and emphasizing detailed feature extraction, the study significantly improves detection accuracy and model robustness across diverse review types. Originality/value This research enhances fake review detection by focusing on non-reviews and using detailed feature extraction methods. Innovative NLP and machine learning techniques, such as opinion sentence and term frequency analysis, improve the identification of fake reviews. This study advances the technology for distinguishing genuine reviews and contributes to the healthy development of e-commerce. The findings are expected to improve the online shopping experience and business decision-making, benefiting both consumers and merchants.