7 Conclusions

UML class diagrams are commonly used in an early phase of software development as a conceptual representation of the application (domain) for which a software system is required. Because of the seamless nature of OO software development processes, the class diagram reflects both the structure of the problem and the structure of the software solution as it provides the foundation for the design of the software system. To guarantee the efficacy of the development processes, it is therefore crucial to control the quality of this early software development artefact. There are several factors which affect the quality of class diagrams such as, size, structural complexity, cohesion, coupling, etc. Quality assurance of early development artefacts requires the use of objective instruments to assess the level of quality in the artefact. These instruments are known as software metrics and quality prediction models.

In this chapter, first, a set of metrics is presented to measure the structural complexity of UML class diagrams. Although several metrics suites that can be used for measuring UML class diagrams have been proposed in the literature, a review of this work revealed that the structural complexity caused by the use of aggregation, association and dependency relationships was not addressed. Moreover, most of the previously proposed metrics suites have not been properly validated. Above all, there is a lack of theoretical validation (i.e. demonstrating that the metrics really measure structural complexity) and thoroughly executed and sufficiently replicated empirical validation (i.e. demonstrating the usefulness of the metrics in...