FMIT: Feature Model Integration Techniques

TL;DR

FMIT is a new technique for feature model integration that improves accuracy and reduces effort by leveraging contemporary strategies and a prototype to support semi-automatic or automatic integration, addressing existing challenges.

Contribution

This work introduces FMIT, a novel feature model integration technique that enhances accuracy and efficiency through strategic, automated, and semi-automated processes, validated by experimental studies.

Findings

FMIT increased integration accuracy by 43%.

FMIT reduced effort by 70%.

Participants found FMIT more effective than manual methods.

Abstract

Although feature models are widely used in practice, for example, representing variability in software product lines, their integration is still a challenge. Many integration techniques have been proposed, although none of these have proven to be fully effective. Integrating feature models becomes a difficult, costly, error-prone task. Since their transition occurs in a generalized and automated way, the techniques applied to compose the models end up giving rise to a final model, in many cases undesired, without taking into account the specific needs arising from the requirements determined by the analysts and developers. Therefore, this work proposes FMIT, a technique for integrating feature models. The FMIT is based on contemporary model integration strategies to increase the accuracy and quality of the integrated feature model. In this way, it will be possible to identify the degree of similarity between composite feature diagrams, to verify their accuracy, as well as to identify conflicts. In addition, this work proposes the development of a prototype based on the set of strategies, used to take decisions according to the requirements established during the integration of feature models, whether this is semi-automatic or automatic. To evaluate FMIT, experimental studies were conducted with 10 participants, including students and professionals. Participants performed 12 integration scenarios, 6 using the FMIT and 6 manually. The results suggest that FMIT improved accuracy by 43\% of the cases, as well as reduced the effort by 70\% to perform the integrations.