Four automated approaches to analyze the quality of UML sequence diagrams

UML sequence diagrams describe the interaction between objects of a system. In this paper, we present four approaches to analyze the quality of sequence diagrams. First, we propose formal model checking techniques and visualization to analyze the possible interaction traces and to identify ambiguities in a sequence diagram. Second, we identified patterns of common interactive behavior and we present an approach to identify these patterns. This approach is useful to identify implicit patterns and make them explicit. Third, we propose a series of checks to identify syntactic defects in sequence diagrams. The fourth approach consists of metrics describing properties such as coverage. The proposed approaches complement each other rather than being competing alternatives. We implemented these approaches in our SquAT tool and conducted a multi case study on three industrial and five research models to validate them. These models showed a large number of defects which were acknowledged by their designers.