MEASURING COMPLEXITY OF DOMAIN MODELS REPRESENTED BY FEATURE DIAGRAMS

Abstract

Feature models represented by Feature Diagrams (FDs) prevail in the software product line approach. The product line approach and FDs are used to manage variability and complexity of software families and to ensure higher quality and productivity of product development through higher-level feature modeling and reuse. In this paper we, first, analyze the properties of feature models. Then, combining some properties of FDs with ideas of Miller’s, Metcalfe’s and Keating’s works, we propose three FD complexity measures. The first measure gives boundaries to estimate cognitive complexity of a generic component to be derived from the feature model. The second measure describes structural complexity of the model expressed through the number of adequate sub-trees of the given model. The third measure estimates total cognitive and structural complexity of the model. To validate the introduced measures, we present a case study with three feature models of a varying complexity.