Design complexity measurement and testing
Communications of the ACM
Software metrics (2nd ed.): a rigorous and practical approach
Software metrics (2nd ed.): a rigorous and practical approach
Systems Engineering Using SDL-92
Systems Engineering Using SDL-92
Qualitative and Quantitative Reliability Assessment
IEEE Software
Integration of well posedness analysis in software engineering
Proceedings of the 2007 ACM symposium on Applied computing
Hi-index | 0.00 |
Software engineers need information about the quality of the specification they are constructing. Computation of measures for quality and complexity could be integrated usefully in specification tools.We take SDL (Specification and Description Language) as an instance for our proposal, and show how meaningful measures can be computed purely from the SDL source specification. We make it explicit in what sense ‘quality’ is quantified: we have a global viewpoint, and capture quality as the degree of consistency between the structural and the communication models in the specification. According to Fenton's classification, the quality measures we propose can be considered as internal product attributes. Our approach is based on graph theory, and uses a probabilistic model to provide lower and upper bounds for scaling quality measures. Hence, quality of one particular specification can be judged, and various specifications can also be compared with one another.We also present a method that suggests improvements to the system structure, and increases the quality of a given specification. As a concluding example, we sketch an initial specification, compute its quality, list possible improvements, choose one, and expose the resulting improved specification.Our approach is not confined to SDL, but can be applied to other specification languages, provided they give descriptions of structure and communication.