A Framework for Analyzing UML Behavioral Metrics based on Complexity Perspectives

Abstract

As software systems become more complex, software modeling is crucial. Software engineers are adopting UML behavioral diagrams to model the dynamic features of a system. These dynamic diagrams keep changing for further improvement, hence becoming more complex. In this case, there is a need to define the measurement attributes used to measure the complexity of these diagrams. Several researchers have addressed the quality of these diagrams by developing measurement frameworks. However, the existing frameworks in the literature are limited since they do not capture the perspective complexity of these diagrams. In this paper, we establish the taxonomy complexity of UML behavioral diagrams, we then modify Kaner’s and Briand’s framework to propose measurement attributes namely, element, control flow, and interaction based on the taxonomy complexity of behavioral diagrams. Finally, we test the applicability of the proposed framework using behavioral diagram metrics. Results indicate that the proposed framework represents parameters vital to evaluate and validate the complexity measures of behavioral diagrams.