Capturing non-functional properties through model interlinking

The Software Product Line (SPL) paradigm promotes systematic reuse-based software development and is founded on the idea of capturing commonalities and variabilities between software products of a target domain. Feature model as the main artefact of SPL development mostly captures functional and operational variability of a system. Researchers have argued that connecting intentional variability models such as goal models with feature variability models in a target domain can enrich feature models with valuable quality and non-functional information. Interrelating goal models and feature models has already been proposed in the literature for capturing non-functional properties in software product lines, however, the integration process is cumbersome and tedious. In this paper, we propose a (semi) automated approach that systematically integrates feature models and goal models through domain-specific conceptual models represented by ontologies. We first associate each element of the models with their raison-d'etre, i.e., sections of the domain documents that justify each model element. Then, using texts that are associated with the elements and through a semantics-enabled textual analysis process, the model elements will be semantically annotated with related ontological concepts. Finally, a mapping recommendation process is proposed that would connect feature model and goal model elements through measuring the semantic similarity of their annotated ontological concept. Our proposed approach not only provides the means to systematically interrelate feature models and goal models but also allows domain engineers to identify the mutual impact of features and non-functional properties.