Detection of Online Fake Review Using Deep Learning

Abstract

An online review is a brief, user-generated assessment of a product, service, or experience posted on the Internet. These evaluations offer insights, ratings, and opinions to help others make informed decisions. Reviews can be found on various platforms, aiding consumers in choosing quality and reliable options. Positive reviews are instrumental in attracting customers and driving higher sales. However, the growing prevalence of fake or deceptive reviews poses a challenge. Detecting fake reviews is an active research field that relies not only on review features but also on reviewer behavior. The proposed research focuses on the significance of online reviews and their impact on reputation building and decision-making for end-users. To address this problem, the study presents a deep learning strategy to spot bogus reviews. For extracting reviewer behaviors along with review text, the proposed system uses a variety of feature engineering techniques. The study compares the performance of two convolutional long short-term memory (CLSTM) and recurrent neural network (RNN) is a deep learning model. The amazon website's dataset of product reviews is utilized to evaluate the effectiveness of the suggested technique. According to the project's findings, the CLSTM model had a much greater accuracy of 99.17% compared with the RNN model's 84.83% accuracy. The study anticipates that these outcomes will surpass those of other in regard with f1-score, precision, recall, and accuracy for machine learning classifiers.