Re: Help...Current status of formalization of UML



Re: Help...Current status of formalization of UML

From: Arend Rensink ^lt;rensink@cs.utwente.nl>
Date: Sun 30 Apr 2006 - 12:02:13 BST
Message-ID: <44549935.2000101@cs.utwente.nl>
Pierre-Yves Schobbens wrote:

> 1- Graph transformation is a rather complex subject studied since 30 
> years (already by Knuth, Courcelle and other now retired people)
>   and we are still far from understanding it.

That's indeed the prevailing opinion, which has done graph transformation a great disservice: the perception is that it must be a very complicated subject because many very clever people have studied very intricate theoretical problems, and only found partial answers - so stay away from it, folks! But in fact to use GT for things like defining semantics you might not need those answers: since you will be using GT as a tool rather than a field of study you needn't necessarily be bothered about questions of category theory or logic.

> 2- A transformation, by definition, cannot be a semantics.
>    Transformations are by definition executable, while semantics 
> concepts for UML are not decidable.
>    Thus this is impossible to define semantics with transformations 
> alone: we need more powerful tools.

I don't understand this statement. A natural way to use GT as a tool to define semantics would be to give a small-step operational semantics on the basis of graph transformation rules. The semantics is then the outcome of a computation consisting of a sequence of such small steps, which is as undecidable as you could ever wish it to be.

Apart from Greg's references, I think the work by Jan-Hendrik Hausmann is worth checking out. An example is the ref. below, but has now also finished his PhD thesis and he'd probably be very glad to send you a copy.

@inproceedings{DBLP:conf/se/StorrleH05,
  author    = {Harald St{\"o}rrle and
               Jan Hendrik Hausmann},
  title     = {Towards a Formal Semantics of UML 2.0 Activities.},
  booktitle = {Software Engineering},
  year      = {2005},
  pages     = {117-128},
  crossref  = {DBLP:conf/se/2005},
  bibsource = {DBLP, http://dblp.uni-trier.de}
}

@proceedings{DBLP:conf/se/2005,
  editor    = {Peter Liggesmeyer and
               Klaus Pohl and
               Michael Goedicke},
  title     = {Software Engineering 2005, Fachtagung des GI-Fachbereichs
               Softwaretechnik, 8.-11.3.2005 in Essen},
  booktitle = {Software Engineering},
  publisher = {GI},
  series    = {LNI},
  volume    = {64},
  year      = {2005},
  isbn      = {3-88579-393-8},
  bibsource = {DBLP, http://dblp.uni-trier.de}
}

-- 
Arend Rensink                         http://www.cs.utwente.nl/~rensink
Department of Computer Science             mailto:rensink@cs.utwente.nl
University of Twente                               tel: +31 53 489 4862
P.O. Box 217, NL-7500 AE Enschede, Netherlands     fax: +31 53 489 3247
Received on Sun Apr 30 12:02:12 2006