Design level quality analysis using fuzzy UML models and weighted complex networks

Analyzing the quality of software architecture is an important task in the software development. Software architecture modeled using a weighted complex network can be an effective method to aid in quality analysis as the software becomes more complex. During the early design phases there exists fuzziness in representing the complex objects. Hence, a fuzzy UML model approach is used to represent the software during early design phases. A hybrid weighting mechanism is considered here which includes the complexities of the classes, relationship and the scenarios for including context-sensitive information while calculating the weight of the link for the transformed complex network. Once the architecture is modeled as weighted complex network, the statistical metrics studied from the network gives the clear evaluation of the quality attributes of software such as maintainability and reliability.