Fake Comment Detection Using Machine Learning Algorithm: A Comparative Approach

The issue of fake comments on social media platforms, particularly Facebook, has raised significant concerns regarding misinformation, user manipulation, and content trustworthiness. This study addresses the challenge of identifying deceptive comments by leveraging machine learning techniques to build a reliable classification system. The proposed methodology involves a comprehensive pipeline including data collection from Facebook, text pre-processing (tokenization, stop word removal, and handling of missing values), and feature extraction using Term Frequency-Inverse Document Frequency (TF-IDF) vectorization. Two supervised learning models—Naïve Bayes and Logistic Regression—were trained and evaluated using a dataset split in an 80:20 ratio for training and testing, respectively. Experimental results revealed that while both models performed reasonably well, Logistic Regression outperformed Naïve Bayes with an accuracy of 92% compared to 87%. Performance metrics such as precision, recall, and F1-score further supported this finding. Confusion matrix analysis also indicated fewer misclassifications with the Logistic Regression model. These findings highlight the viability of machine learning approaches in automating the detection of fake user-generated content. The outcomes of this study contribute to the broader field of social media content moderation and provide a foundational approach that can be expanded upon using more complex models or multilingual datasets in future research.