Fake Reviews Filtering System Using Supervised Machine Learning

As the surge in internet users is expanding prominently, the role of the online reviewing system is also rising. For companies, the legitimacy of internet evaluations is critical, as it can directly impact their reputation and profitability. It plays an indispensable role in influencing people’s perceptions of a product or service. This research projects light on the best technique to identify and filter out authentic reviews while proposing a flexible and user-friendly website. The website will have a tremendous sway on customers and will assist them in making a better judgment about a product/service. The website is deployed with the designed supervised learning model. Firstly, the user will have to enter the URL of the website where the product is located. After which, the dataset is extracted from the given URL using Python tools for Web Scraping. The data is then analyzed and dissected using Natural Language Processing techniques to extract sound features from it. Ultimately, different Machine Learning Models are further trained on the dataset. The experimental results of this research reveal that the model performs at an accuracy of 89.12% on the datasets. The major objective of this research is to provide a fake review filtering system that will provide users with more reliable review information and eliminate revenue loss of the companies at an exponential rate.