Theoretical and Practical Complexity of Unified Modeling Language: Delphi Study and Metrics Analyses

Systems have become increasingly complex, and as a result development methods have become more complex as well. The unified modeling language (UML) has been criticized for the often cited and sometimes overwhelming complexity it presents to its users, and those seeking to learn to use it. Using Rossi and Brinkkemper’s (1996) complexity metrics, Siau and Cao (2001) completed a complexity analysis of UML and 36 other modeling techniques, finding that UML is indeed more complex than other techniques. Siau, Erickson and Lee (2002) proposed that Rossi and Brinkkemper’s metrics present the theoretical maximum complexity, known as theoretical complexity. This is different from a practice-based complexity, known as practical complexity, that more accurately reflects the complexity of the language in practice. This research develops a subset of UML (a kernel) composed of the most commonly used constructs, that can be equated with the complexity that practitioners actually face when using the modeling language. A Delphi study is conducted using UML experts in identifying a use-based UML kernel. Metric analyses are then computed and analyzed.

Presentation Information

Researcher Information

	Siau, Keng L Edwin J. Faulkner Professor Expertise: Systems Analysis and Design Design Science Web 2.0, Virtual World, and 3-D Web Supply Chain Integration and Management Electronic, Mobile, and Ubiquitous Commerce Management CBA 255 P.O. Box 880491 University of Nebraska-Lincoln Lincoln, NE 68588-0491, USA Phone: (402) 472-3078 Fax: (402) 472-5855 ksiau@unl.edu	Siau, Keng L	Management