[SKOS] TR : RE : Lack of RDF/XML examples in new standards

(transfering this to the SWD list, as I think this thread can be of interest)

-------- Message d'origine--------
De: Antoine Isaac
Date: mer. 12/11/2008 14:20
À: Stephen Bounds; public-esw-thes@w3.org
Cc: SKOS
Objet : RE : Lack of RDF/XML examples in new standards
 
Hi, 

I don't like at all any approach that would let XML community think that they can deal with RDF data (and especially ingest it) just using basic XML tools. Unless you have very well designed your stylesheet I could write dozens of RDF/XML syntactic variants of a same graph that would put it on its knees (nested descriptions, attribute/value shortcuts for literal objects, etc -- by the way that's not a personal attack , Stephen, just that properly dealing with all these in a stylesheet is hellish :-)

But I do agree that we could provide some RDF/XML example in our doc (for the Primer at least) or next to it. But here I would welcome some more concrete ideas to synchronize that with our Turtle examples. Because one thing is clear: for beginners, Turtle example should be provided, and should prevail.

Cheers,

Antoine


-------- Message d'origine--------
De: public-esw-thes-request@w3.org de la part de Stephen Bounds
Date: mer. 12/11/2008 12:42
À: public-esw-thes@w3.org
Cc: SKOS
Objet : Re: Lack of RDF/XML examples in new standards
 

Jakob & Norman,

I'm going to try and have my cake and eat it too, I'm afraid.

*Yes*, RDF/XML can be a very confusing syntax, particularly if you nest 
multiple levels of assertion deep, don't do predicate striping, etc etc.

But if you limit yourself to simple subject-predicate-object assertions, 
RDF/XML provides a syntax that is very familiar to XML authors with the 
added benefits of RDF semantics.

Not coincidentally, this is exactly what the SKOS examples showed.  A 
typical SKOS document looks like this:

<rdf:RDF>
   <skos:Concept rdf:about="http://www.example.com/foo#bar">
     <skos:prefLabel>bar</skos:prefLabel>
     <skos:broader rdf:resource="http://www.example.com/foo#baz" />
   </skos:Concept>

   <skos:Concept rdf:about="http://www.example.com/foo#baz">
     ...
   </skos:Concept>

   <skos:Concept rdf:about="http://www.example.com/foo#qux">
     ...
   </skos:Concept>
</rdf:RDF>

which essentially boils down to a sequence of assertions that are either:

   Subject -> Predicate -> Literal, or
   Subject -> Predicate -> Resource

This covers 99% of the needs of basic SKOS users, and this limited form 
of RDF/XML is both correct and easily understandable by non-RDF literati.

A personal example: I found SKOS RDF/XML to be a really useful 
intermediate language for capturing thesaurus-type relationships.  The 
initial XML was created using a Perl XML builder library from a CSV source.

I then transformed the data into static web pages using an XSL transform 
tool (Saxon), and as a dynamic web application through ingest and 
manipulation via an XML database platform (eXist).

I looked at RDF tools (Jena etc), but couldn't find anything that had 
close to the same maturity or ease of use of these XML tools.

Perhaps this was just my existing familiarity with XML rather than RDF, 
but that too is kind of my point.  The pool of XML developers is far 
larger than the pool of RDF developers.

Cheers,

-- Stephen.

Jakob said:
> RDF/XML is one of the main reasons why adoption of RDF took so long.
> There are dozens of ways to encode the same graph in RDF/XML. Without a
> full specialized parser you are lost. RDF/XML might be more known and it
> *looks* more easy but it confuses more then it helps.

Norman Gray wrote:
> Good heavens!  I would _never_ show a beginner RDF/XML.  Do you really 
> do this?