Many-objective optimization of non-functional attributes based on refactoring of software models

Software quality estimation is a challenging and time-consuming activity, models are crucial to face the complexity of such activity on modern software applications. In this context, refactoring within development life-cycles where requirements functionalities rapidly evolve. One main challenge that improvement distinctive attributes may require contrasting actions software, as for trade-off between performance reliability (or other non-functional attributes). cases, multi-objective optimization can provide designer with wider view these trade-offs and, consequently, lead identify suitable take into account independent or even competing objectives. paper, we present an approach exploits NSGA-II genetic algorithm search optimal Pareto frontiers while considering many We consider variations model alternative respect initial model, amount antipatterns detected alternative, architectural distance, which quantifies effort obtain from one. applied our two case studies: Train Ticket Booking Service, CoCoME. observed able improve (by up 42%) preserving improving 32%) generated alternatives. also there exists order preference among Based analysis, state confirmed their ability subject in context many-objective optimization. addition, metric adopted distance seems be estimating effort.