Re: Picking the Tools

[Uche Ogbuji <uche.ogbuji@fourthought.com>]

> If one models at the level of UML, one should be able
> to create both RDF and Topic Map instances of the UML
> models.  Yes or no?
>
> Given that we have XML Schemas, Topic Maps, RDF, and
> so forth, isn't it prudent to work with an abstract
> modeling technology such as UML over any of the above?

I'm probably the wrong person to ask this.  Remember that I'm only an
occasional amateur of object-oriented development, and I am quite opposed
to its use in any modeling outside what I consider to be its strength: the
packaging of code modules.

One of the reasons I work so much with XML and RDF is that they are *more*
abstract than OO, and allow OO modeling as well as other forms, all of
which, on aggregate, are far more expressive than OO.

Therefore I personally would be the last person to prefer UML models to
RDF or Topic maps except for the design of code modules.

While I'm at it, I'll note that this is also one of the reasons Python is
my computer language of preference: it has strong OO support, in order to
take advantage of its strengths within cohesive modules, but it is
flexible enough that you can easily go far beyond polymorphism,
inheritance and encapsulation when need be.  C++ also gives you this
capability, although you have to strain mightily to get it.  Java simply
denies you the opportunity of any deviation from OO orthodoxy.

But to answer your first question, I'm sure one could almost always derive
a mechanical conversion from XMI to RDF or XTM, and therefore UML,
but as a general facility this will be only as good as any mechanical
process in modeling, i.e. not very good.


-- 
Uche Ogbuji                               Principal Consultant
uche.ogbuji@fourthought.com               +1 303 583 9900 x 101
Fourthought, Inc.                         http://Fourthought.com
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python