Using developers' feedback to improve code smell detection
Mário Hozano, Henrique Ferreira, Italo Silva, Baldoíno Fonseca, Evandro De Barros Costa · 2015
Several studies are focused on the study of code smells and many detection techniques have been proposed. In this scenario, the use of rules involving software-metrics has been widely used in refactoring tools as a mechanism to detect code smells automatically. However, actual approaches present two unsatisfactory aspects: they present a low agreement in its results and, they do not consider the developers' feedback. In this way, these approaches detect smells that are not relevant to the developers. In order to solve the above mentioned unsatisfactory aspects in the state-of the-art of code smells detection, we propose the Smell Platform able to recognize code smells more relevant to developers by using its feedback. In this paper we present how such platform is able to detect four well known code smells. Finally, we evaluate the Smell Platform comparing its results with traditional detection techniques.