[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Schema design and attribute for ontology identification

• From: "Alexander Johannesen" <alexander.johannesen@g...>
• To: "XML Developers List" <xml-dev@l...>
• Date: Fri, 9 Nov 2007 15:50:15 +1100

Hi all,

I'm creating a very simple pipeline schema as a basis for various
SOA/ROA development projects, and I've been pondering a number of
things in this process, especially what the best way is to identify
that a certain sub-tree of an XML belongs to a certain ontology (an
ontology here is loosely defined as a vocabulary, and not as a schema
per se). Now before we head down namespaces let's make it clear that
this is not about mixed content model but about identification. I've
pondered different ways to do it (In all these examples I've use URI's
as the identifier, but they might be uid's of any kind, of course) ;

 <container xmlns="http://some.ontology"> ... </container>

Namespaces here feels wrong to me as the content of the <container>
may not even have a schema (yet). It's up to applications to deal with
content they know about, and I'm not sure I want to a) imply namespace
handling in all our tools pr. default, and b) most of our tools will
use simple but application specific XML and I don't see the benefits
to forcing everyone to also do schema work for mostly non-validating
reasons.

 <container id="http://some.ontology"> ... </container>
 <container oid="http://some.ontology"> ... </container>

The plain @id clashes with xml:id for me, and a great number of tools
will assume that's what they are. And that's not necessarily wrong.
The @oid seems quite non-descript yet workable. Perhaps call the
attribute @ontology or something similar?

Further to this is the notion that a container of XML contains data of
some kind, ala ;

 <container type="http://some.ontology"> ... </container>

Or one can pull an RDF (introducing namespaces where we might not want them) ;

 <container rdf:about="http://some.ontology"> ... </container>

Or Topic Maps (again, introducing namespaces where we might not want them) ;

 <container xtm:psi="http://some.ontology"> ... </container>

Thoughts? What do you people do these days to identify that a sub-tree
belongs to some notion that isn't necessarily another schema?


Alex
-- 
---------------------------------------------------------------------------
 Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------