Abstract :

Nowadays, when choosing a brand, most of us look to online reviews. Unfortunately, review sites are increasingly facing the problem of opinion spam, which spreads false information with the intent of promoting or harming specific businesses through deceiving human readers or automated systems that analyse sentiment and opinion. It is for this reason that many data-driven methods for evaluating the veracity of user-generated material disseminated via social media in the shape of online reviews have been put forth in recent years. Reviewers, reviews, and the network structure that links various entities on the review-site in test are all aspects that different techniques take into account. The purpose of this article is to examine the most prominent review and reviewer-centric aspects that have been suggested in previous works as a means of identifying fraudulent reviews, with a focus on methods that make use of supervised machine learning. On the whole, these solutions o