Validating an Interactive Ranking Operator for NSGA-II to Support the Optimization of Software Engineering Problems

• Willian Freire UEM
• Cláudia Rosa UEM
• Aline Amaral UEM
• Thelma Colanzi UEM

Resumo

Search-Based Software Engineering (SBSE) has been beneficial for optimizing the solution of several Software Engineering (SE) problems. The incorporation of Decision Makers (DM) preferences during the search process may help the algorithms to find more adequate solutions for their profiles. Some interactive approaches allow the DM to evaluate solutions, rating them with scores during the search process. These scores represent the adequacy level of the solutions in relation to the DM preferences and should influence the evolution of the search algorithm. In previous work, we proposed an interactive ranking operator for NSGA-II to support the complete prioritization of solutions for any SE problem domain. Although this operator worked satisfactorily in an application example, its validation is required so that it can be used in real application contexts. In this sense, we instantiated the interactive ranking operator for NSGA-II presented in previous work, and we conducted an exploratory study with a twofold goal: (i) validate the impact in the ranking of solutions, and (ii) check the diversity of them. To accomplish such goals, we made statistical tests such as correlation and regression analysis using quality metrics for Product Line Architecture (PLA) Design. The results pointed out that the interactive ranking operator can properly deal with the DM preferences, giving a greater chance of surviving to those solutions with higher scores, without compromising their diversity.

Palavras-chave: Optimization for Software Engineering problems, Interactive SBSE, NSGA-II ranking operator