A Many-Objective Estimation Distributed Algorithm Applied to Search Based Software Refactoring

Abstract

Refactoring is a modification in the internal structure of software, in order to improve quality, understandability and maintainability, without changing its observable behavior. Search Based Software Refactoring (SBSR) deals with automatic software refactoring processes using optimization algorithms. In this context, here we investigate the problem of finding a sequence of refactorings that provides code improvement, according to software quality attributes, expressed by a combination of software metrics. There are multiple criteria to define the quality of a solution, therefore this problem is defined as a Many-Objective Combinatorial Optimization Problem. There is a lack of works that focus on Many-Objective Discrete Problems in SBSR. In this direction, this work proposes a Many-Objective Estimation Distributed Algorithm to find a sequence of refactorings on an object-oriented software. The algorithm explores archiving methods and probabilistic models. A set of experiments is performed, with the aim of investigating which is the best algorithm configuration, regarding the probabilistic model and selection procedure.