Using Sentiment Analysis on online product reviews for determining fairness

Product reviews became one of the most relevant ways customers have to make up their mind about buying specific products. The relevance of these reviews tempts companies to either use them to attack their rivals or to oversell their products by providing misguiding information that does not fit with the real product’s characteristics. Identifying this unfair situation is complicated but, at the same time, crucial to guarantee the reliability on the customer’s choice. In this work, we aim to simplify unlawful reviews by providing a simile mechanism. Our hypothesis claims that sentiment analysis can help to red flag unfair reviews and, consequently, simplify this difficult process. For that, we measure the correlation between unfairness and sentiments to check how much emotions are manipulated to guide shopping tendencies.

On the one hand, having access to meaningful information is, in fact, essential during the decision-making process and, on the other hand, observation of unfair data can prevent its negative impact on businesses and consumer choices, therefore this project focuses on exploring and experimenting how to detect unfair online reviews through Sentiment Analysis using Machine Learning Techniques. The experiments focus on the discovery of unfairness in online product feedback through the process of establishing the accuracy of sentiment classification algorithms aims to detect existing unfairness towards the products.