Detecting Spam Comments in Product Reviews on E-Commerce Websites

With the increasing use of e-commerce platforms, user reviews have become a major factor influencing customer decisions. However, the presence of spam comments-whether posted automatically or manually-negatively affects the reliability of these reviews. In this study, we focus on detecting spam comments in product reviews using various machine learning techniques. The data was collected from a real e-commerce platform and preprocessed through steps like stopword removal, normalization, and vectorization. Different models such as Support Vector Machine (SVM), Naïve Bayes, Random Forest, and Logistic Regression were trained and tested. Besides text-based features, behavioral indicators such as repetitive comment patterns and posting frequency were also considered. Experimental results show that combining multiple features improves classification performance, with SVM and Random Forest achieving the highest accuracy. This project aims to contribute to making online reviews more trustworthy and to support users in making more informed purchase decisions.