Fake Review Detection Using Machine Learning and Deep Learning Algorithms

Abstract

With the rapid growth of e-commerce platform, the Fake Review Detection has become a popular and significant topic for both the businesses and research area in recent years. For making a best decision, online reviews play a very important role in today’s e-commerce. There are millions of reviews available regarding various product and services in different social sites and marketing websites. Customers write their opinion based on their experience and these reviews become the source of information for the other consumers because depend on those reviews, people take their decision. But sometimes, it becomes so tough to find out the genuine reviews as there is no restriction for written a review in any online platform. Anyone can write reviews according to them which raise the number of fake reviews and can give wrong information and mislead a customer. These fake or genuine reviews play a significant role for the reputation and revenue of an organization. Usually, positive reviews attract more customers and gain high profit whereas negative reviews badly affect the reputation of an organization. This situation makes us interested to distinguish the fake and genuine review. In this study, we discuss some Machine Learning algorithms (Naïve Bayes, Random Forrest, K-Nearest Neighbor, Logistic Regression, Support Vector Machine (Linear), Decision Tree) and Deep Learning algorithms (Dense Layer Architecture, LSTM, BiLSTM) and using those algorithms we try to identify whether the review is credible or not. We applied confusion matrix and analyzed the experimental results. Then, we show some trade-off between Machine Learning and Deep Learning algorithms. Also, we discuss about some challenges we faced while doing this thesis and we discuss some of our future plan regarding this thesis.