Neutro ou negativo? Avaliação de sentimentos expressados em comentários avaliativos de serviços de hospedagem

Abstract

Why is it so rare to see people talking badly about the services offered by other people in the comments of Airbnb, Uber, Couchsurfng and other services of shared economy? In the present work a comparison is made between the sentiments of the reviews of Booking.com, Airbnb and Couchsurfng and also a new approach to how the Airbnb listing could be different if the feelings were taken into account. Through an extensive collection and data mining of the evaluative comments of each of the platforms already mentioned and the application of sentiment analysis, it was possible to compare and reach the conclusion that, in fact, in the tools of shared economy, the polarities tend to be more positive when compared to the tools of the formal economy. This tendency of positivity ends up generating a problem for the decision making of a consumer of these platforms. In many cases, it is difcult to distinguish a good location from a medium/bad one through evaluative comments. To solve this problem, the present work proposes the attribution of a score obtained through a formula that takes into account the polarity of the feeling of a place and its quantity of comments. When the database is ordered according to this note, the best places are ranked better than the average/bad sites, helping the user in a search. To prove this difculty, a survey was applied and answered by 30 people. They should classify the options from very bad to very good according to the score obtained in the process described above. In the survey it was found that really is very difcult for a consumer to make his decision based on the comments of other users. Places with the worst grades were still classifed as average or good. Evidence that ordering a search makes all the difference.