Interaction in search-based PLA

The Software Product Line (SPL) approach is a reuse technique derived from other engineering disciplines to develop a family of products within a specific domain. An essential artifact of SPL is the Software Product Line Architecture (PLA), which identifies elements characterized by variation points, variability, and variants. PLA anticipates design decisions and aims to achieve critical features such as reusability and modularity. As a multiobjective optimization problem, the interactive multiobjective optimization approach for PLA design (iMOA4PLA) employs specific metrics to optimize the PLA problem. The iMOA4PLA is implemented through the OPLA-Tool v2.0, with the decision maker (DM) assuming a participatory role in the optimization process, guiding it through various strategies. In this context, machine learning (ML) algorithms assist the DM. We conducted two experiments to evaluate the proposed model - one quantitative and one qualitative. The results indicate that the approach facilitates the generation of PLA alternatives that align with the DM's expectations. Additionaly, strategies employed are highly efficient in assisting the DM with the PLA evaluation process. This research presents a model that helps decision-makers optimize the Software Product Line Architecture while enabling an interactive process that leverages machine learning algorithms. The scientific contribution of this work lies in providing a framework for the design and evaluation of the PLA that leverages the benefits of machine learning algorithms.hat utilizes machine learning algorithms.