Identifying Deceptive AI Reviews: A Machine Learning Approach

In today's digital age, online reviews greatly influence what people buy. However fake reviews or reviews generated by Artificial Intelligent tools by using the Natural Language Processing models cause a big problem. They mislead the consumers in judging a product. This paper aims to address the issue by using machine learning to predict if a review is real or fake. A labeled dataset containing product reviews was chosen from Kaggle. It was preprocessed and then used to train three models using three different algorithms which were Support Vector Machines (SVM), Random Forest Classifiers and Logistic Regression. The best one chosen based on evaluation metrics was Support Vector Machines. This model was then used to test reviews of some top products scraped from the Amazon website. The results show the capability of the model in detecting deceptive reviews.