A new design defects classification: Marrying detection and correction

Abstract

Previous work classify design defects based on symptoms (long methods, large classes, long parameter lists, etc.), and treat separately detection and correction steps. This paper introduces a new classification of defects using correction possibilities. Thus, correcting different code fragments appending to specific defect category need, approximately, the same refactoring operations to apply. To this end, we use genetic programming to generate new form of classification rules combining detection and correction steps. We report the results of our validation using different open-source systems. Our proposal achieved high precision and recall correction scores. © 2012 Springer-Verlag Berlin Heidelberg.