Análise de sentimentos em repositórios do GitHub.

Abstract

Pull-requests are suggestions for changes or improvements, for a repository of a project on the GitHub environment. These suggestions can be commented on by developers, and they can express different sentiments in their comments. In this study, comments present in pull-requests were analyzed in order to understand whether positive comments may or may not influence the acceptance of pull-request. For this, data extraction techniques, use of state-of-the-art approaches to deal with Big Data and pre-trained tools to produce this analysis were applied. The final result verified in this study showed that, yes, there is a relationship between positive comments and the successful acceptance of pull-requests. From a covariance calculation, it was understood that there is a positive correlation between the "score variable" and the "success variable". Rejecting, through a hypothesis test T-Student, the null hypothesis that the average of comments expressing positive sentiments and expressing negative sentiments for pull-requests have equal averages. It was understood that if the means between the two variables are different, this is strongly associated with different behaviors, if the comments have sentiments with different intensities.