Fake Review Detection using Evidential Classifier

Online shopping is a major part of our lives and has proven to make our lives easier and more time-saving. Rating and reviewing those online products is how we spread the word and also get to know about a product’s quality. Ratings help the e-commerce sites rank the products in a category, and that’s how we purchase the best possible product. This affects a business, as the better the reviews, the more consumers there are and the higher the profit. On the negative side, we have human spammers and machine-generated automatic spamming of reviews and ratings, which can affect any business positively or negatively and would also create a false image of a product, leading to a bad customer experience. Therefore, if we are able to find a classifier that can tell us if a review is fake or real, it can help online shopping sites determine a product’s actual ratings and thus reduce unfair competition. Hence, in this paper, we propose the classification of reviews via the Dempster-Shafer (DS) evidential classifier. This classifier has proven to be very effective in classifying images and texts. This paper mainly focuses on how certain features, when fed to the Dempster-Shafer classifier, can help us detect fake reviews.