Seminar by T Anil Kumar
Issue-based Variability Management: Methodological and Empirical
T Anil Kumar
Harman International and TU Munich
Date: Monday, August 1st, 2011
Time: 11 AM
Venue: CS101.Abstract:
Variability management is a challenging issue in product line requirements engineering. Particular problems include the instantiation of variations for individual products, the long-term evolution of the product line, the rationale representation for variability and the support of issue-based communication between domain engineering and application engineering. To address the problems, a new methodology called issue-based variability modeling was proposed by extending variability management with rationale management activities. Rationale management is a discipline in software engineering, which deals with the tacit knowledge that leads to a system design. To be able to empirically evaluate issue-based variability modeling, the meta-model of the Questions, Options and Criteria (QOC) is extended with the meta-model of the Orthogonal Variability Modeling (OVM) to obtain a meta-model for issue-based variability modeling. Tool support for issue-based variability modeling will be covered based on object-oriented plug-ins. In particular, extensions to open source as well as commercial software engineering tools will be presented in the talk. Issue-based variability modeling was evaluated empirically using an experimental survey with 34 professionals and a quasi-experiment with 258 students. The quasi experimental design includes several methods from case-based research such as use of neutral mentors, embedded design and mixed-methods (e.g. both quantitative and qualitative). The empirical evaluation gave evidence that the methodology is applicable in requirements teams of various modeling expertise, improves representation and instantiation of variability over conventional approaches as well as enhances longevity of variability models.