FAKE REVIEW IDENTIFICATION USING VARIOUS KINDS OF NEURAL NETWORKS – A SYSTEMATIC REVIEW

Abstract

In modern times, the foremost factors influencing the purchase decision of a customer are the reviews of other customers. People make their purchase decisions based on the reviews given by other people, thereby making these reviews a valuable source of information for any companies that wish to sell products online, and customers that wish to buy these products. The reliance of people on these customer reviews in turn may lead to a concern that people may misuse this platform and write fake reviews to untruthfully promote or devalue a certain product or commodity. These are known as Spam Reviews, or fake reviews, wherein people change, manipulate, and poison reviews for their own benefit. This review paper will focus on the latest techniques that have been developed or are in the research phase in order to deal with this research challenge. The methods previously used for this particular application are Data Mining, Natural Language Processing, Machine learning based on supervised learning architecture, etc. These methods are better developed, but have certain limitations. To provide a clearer picture on the latest methods, this paper shall also discuss methods like Bi-directional LSTM’s, Generative Adversarial Networks, and other related methods based on deep learning.