Achieving change requirements of feature models by an evolutionary approach

Abstract

Feature models are a widely used modeling notation for variability and commonality management in software product line (SPL) engineering. In order to keep an SPL and its feature model aligned, feature models must be changed by including/excluding new features and products, either because faults in the model are found or to reflect the normal evolution of the SPL. The modification of the feature model to be made to satisfy these change requirements can be complex and error-prone. In this paper, we present a method that is able to automatically update a feature model in order to satisfy a given update request. The method is based on an evolutionary algorithm that iteratively applies structure-preserving mutations to the original model, until the model is completely updated or some other termination condition occurs. Among all the possible models achieving the update request, the method privileges those structurally simpler. We evaluate the approach on real-world feature models; although it does not guarantee to completely update all the possible feature models, empirical analysis shows that, on average, around 89% of requested changes are applied.