Authors:
Xin Zhao
and
Jeff Gray
Affiliation:
Department of Computer Science, University of Alabama, Tuscaloosa, AL and U.S.A.
Keyword(s):
Model-driven Engineering, Feature Modeling, Model Complexity, Data Mining.
Related
Ontology
Subjects/Areas/Topics:
General-Purpose Modeling Languages and Standards
;
Languages, Tools and Architectures
;
Model-Driven Software Development
;
Reasoning about Models
;
Software Engineering
Abstract:
Software Product Lines (SPLs) play an important role in the context of large-scale production of software families. Feature models (FMs) are essential in SPLs by representing all the commonalities and variabilities in a product line. Currently, several tools support automated analysis of FMs, such as checking the consistency of FMs and counting the valid configurations of a product line. Although these tools greatly reduce the complexity of FM analysis, FM design is often performed manually, thus being prone to bad design choices in the domain analysis phase. This paper reports on our work to improve FM qualities from the exploration of the relationship between FM structure and structural complexity. By performing two common operations (i.e., consistency checking and counting valid configurations on FMs with different sizes and structures), we collected the time that an automated tool needs to finish these operations. Then, we applied data mining approaches to explore the relationshi
p between FM structure and structural complexity. In addition, we provide guidelines for designing FMs based on our observations.
(More)