Requirements engineering looks at the science and discipline of establishing and documenting software requirements. The Unified Modeling Language(tm) (UML) is the industry-standard language for specifying, visualizing, constructing, and documenting the artifacts of software systems. When UML is incorporated into the requirements engineering process, the elicitation of detailed requirements for software product definition (sometimes referred to as "use case modeling") becomes less art and more science.

Unfortunately, real-world UML models tend to be large, complex, difficult to review, and even harder to extract requirements from. The UML provides a notation and semantics for capturing the static and dynamics of business processes as diagrams, but without any guidelines for how to create and review the diagrams. Individuals creating the models may have a "development" bias and/or not fully understand the implications of process decisions made during model construction, exacerbating the problem of extracting a complete requirement set.

This talk will focus on the process of creating semantically correct UML models from which a full set of requirements can be extracted using automation. A set of analysis model heuristics will be presented: some of the heuristics are well known and some of them are new. The heuristics, when applied to the modeling process will address concerns such as:

• How do I start?
• When am I done?
• How do I create a verifiable model?
• How do I extract my requirement set from the completed model?

Brian Berenbach brian.berenbach@scr.siemens.com is a Senior Member of Technical Staff at the Siemens Corporate Research Center in Princeton. He has a M.Sc. from Emory University and has done additional work towards his Ph.D. at the New York Polytechnic Institute. Brian has worked with many Fortune companies to develop UML Models of their businesses for re-engineering and software product definition. He has led engineering teams creating business models at AIU, AT&T, Educational Testing Service, Fleming Distributors and Siemens Medical Systems.

A pre-meeting dinner with the speaker is held at 6 p.m. at the Rusty Scupper on Alexander Road in Princeton. If you would like to attend, please RSVP with an e-mail to princetonacm@acm.org.

Princeton ACM / IEEE Computer Society meeting are open to the public. Students and their parents are welcome. There is no admission charge, and refreshments are served.