xml-dev - Re: RDF? TM? (was Re: Didier's lab report)

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

> > <MyOriginElement rdf:ID="origin1">
> >   <incstyle:remotestyle>
> >     <rdf:Alt>
> >       <incstyle:sheet incstyle:case="WML" xlink:href="wml.xslt"/>
> >       <incstyle:sheet incstyle:case="IE5" xlink:href="ie.xslt"
> > incstyle:param="version=5"/>
> >       <incstyle:sheet incstyle:case="IE4" xlink:href="ie.xslt"
> > incstyle:param="version=4"/>
> >       <incstyle:sheet incstyle:case="other" xlink:href="default.xslt"/>
> >     </rdf:Alt>
> >   </incstyle:remotestyle>
> > </MyOriginElement>
> >
> > Again, I even included a few bonuses, and it's still quite simple.  Of
> > course it does illustrate another of RDF's annoyances: that you must
> > have a namespace on all attributes if you wish to them to be considered
> > abbreviated properties.
>
> OK, syntax looks nice. How about the meaning of it?

There's a land mine.  What do you mean by "meaning"?

> I would wonder to see the RDF schema behind "incstyle" namespace.
> Please agree that no graph can be built for your example without knowing the
> schema.

The schema is an important part of the graph, but I do think you might
overstate it a bit.  Anyway, you have used one of the dreaded "S" words.
By "schema", do you mean

1)  Data type constraints?
2)  Relational constraints?
3)  Processing plan?

> Also:
> How do you know that WML is "Wireless Markup Language"

???  Maybe I'm missing something.

<vocab:abbr>
  <vocab:key>WML</vocab:key>
  <vocab:full>Wireless Markup Language</vocab:full>
</vocab:addr>

> and not  "Woman Man Love"?
> And how do you know that sheet is an XSLT stylesheet and not ...
sheetrock?
> :-))

I know because I'm human and I have context to go on.  The machine knows
because I have defined a binding between the particular symbol and a
processing plan.  Where is the problem?

> > Your statement makes me wonder how you understand the role of
"triples" in
> > RDF.  I get  the impression that you might not realize that they are
> > simply a database and list-processing convenient representation of the
> > RDF abstract model, which is a graph, just like TM's.
>
> Yes, but because XTM allows n-ary associations there are less nodes in
the
> graph.

It sounds as if you are saying that because XTM allows n-ary relationships
as a first-class construct that it is more efficient?  I could possibly
buy that, except that I know that even though the RDF spec itself only
allows n-ary relationship to be expressed through indirection, that its
extensibility makes this distinction meaningless.

Efficiency is a matter of implementation.  If I am worried about
instantiating additional nodes with RDF I can enact a processing plan that
minimizes this.  For instance, I could create an anonymous resource that
forms the "hub" of the relationship.

> Well ... I must confess that I was also getting "the impression that you
> might not realize that "
> nothing is hardcoded in Topic Maps :-))

You are correct that I got the wrong impression from your statements.  I
have no problem admitting this.  I'm learning here, just as I imagine you
are.

> By the way, if it sounds that I dislike RDF for some reason, it is wrong.
> I believe that there are certain things that are better in RDF and certain
> things that are better in Topic Maps. I hope that this discussion can make
> this distinction a little clearer.

Same here.  But I never doubted that this is a syncretic discussion.


-- 
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