On Component Identification Approaches - Classification, State of the Art, and Comparison

Partitioning a design space to identify components with desired non-functional and behavioral characteristics is a crucial task in the component-based software development process. Accordingly, the issue of how to analyze design models in order to systematically derive a set of components constitutes a research question which has been investigated repeatedly. Component identification approaches that have been published in literature, however, make use of different component definitions and identification strategies. Furthermore, they vary from ad-hoc findings and general recommendations to more formalized methods and techniques which aim at an optimized partitioning. In this paper, we elaborate on the state of the art in component identification and provide a classification of approaches that highlights their respective strengths and weaknesses. To classify component identification approaches, we introduce a classification scheme that contains important distinguishing factors. It is used to compare existing approaches which have been compiled during a literature survey. In addition, it provides the basis to discuss significant differences between them and to identify remaining issues which give information about future research directions.