Experimenting with Evolutionary Algorithms to Reduce Feature Model Configuration Steps

Abstract

In the software engineering world, software product lines constitute an approach for building reliable software systems. These use feature models to capture, develop, and document shared software for a base system. One of the main challenges when using feature models to derive new products configuration is a way of selecting a configuration that takes under consideration the minimum number of steps and minimum decision-making cost, taking into account resource constraints. To satisfy the challenges of optimizing the configuration selection technique, in this paper, we present an assessment approach that makes use of genetic algorithms to generate then evaluate the best product configurations from feature models. Our empirical outcomes reveal the effectiveness of using the genetic algorithm in obtaining the product configurations that meet the best level of trading-off between steps and decisions at a reasonable production time, consequently, assisting stakeholders in selecting the product configuration that fits their requirements.