Object naming analysis for reverse-engineered sequence diagrams

UML sequence diagrams are commonly used to represent object interactions in software systems. This paper considers the problem of extracting UML sequence diagrams from existing code for the purposes of software understanding and testing. A static analysis for such reverse engineering needs to map the interacting objects from the code to sequence diagram objects. We propose an interprocedural dataflow analysis algorithm that determines precisely which objects are the receivers of certain messages, and assigns the appropriate diagram objects to represent them. Our experiments indicate that the majority of message receivers can be determined exactly, resulting in highly-precise object naming for reverse-engineered sequence diagrams.