With the increased use of e-commerce platforms for purchasing various online products, there is a significant growth of posting reviews about the purchased products on the platforms. These reviews play an important role for new consumers to make a decision about whether to buy the product or not based on the experience of other users. However, these days people are intentionally posting deceptive/fake reviews that can influence consumers into buying a particular product and build a false reputation about it. Since customer reviews play an important role in deciding the credibility of a product and the platform, it is important to identify these forged reviews and eradicate them in order to maintain the integrity for the same. Potential fake reviews possess certain characteristic features that can help us find and remove them. Traditional supervised methods face limitations, as acquiring labeled data is resource-intensive and may not cover the diverse range of fake review characteristics. This research seeks to address these challenges by developing an unsupervised system for initial labeling, followed by a classification model for accurate detection of fake reviews.