Abstract

Software product line engineering has emerged as a prominent software engineering paradigm, as it comprises a set of core assets sharing functionality and quality attributes. Feature modelling is one of the most frequently used techniques for modelling the variability within a software product line. There are several proposals for building Feature Models which rely on semi-automated or fully automated means. Unfortunately, automatic feature model construction has been addressed from different viewpoints, so it is not easy to know which is the best approach for automating the building of variability models. In fact, there is no clarity regarding common elements, and the main differences that characterise such approaches. Additionally, the wide variety of terms used to refer to the process of building a Feature Model (e.g. synthesis, location, re-engineering, and weaving) means that approaches are varied and very heterogeneous, making them complex to understand and classify. This paper introduces FM-CF, which is a Conceptual experience-based Framework for classifying approaches for the automatic building of Feature Models. The framework considers a set of categories mainly focused on characterising some aspects, such as input sources, methods and techniques, results, and types of evaluation. A literature review of (semi-) automated Feature Model construction was performed to identify approaches for building Feature Models by (semi-)automatic means, and the main terms used by those approaches. Then the completeness of the framework was evaluated by mapping the set of dimensions and their items, and the terms extracted from the literature. The conceptual framework provides guidance to researchers for choosing the appropriate aspects with which to build Feature Models, and helps in the understanding and clarification of the proposed approaches. (C) 2019 Elsevier Inc. All rights reserved.