RE: Linkbases, Topic Maps,and RDF Knowledge Bases -- help meunderstand, please

[Jonathan Borden <jborden@mediaone.net>]

Michael Champion wrote:
>
>
> Well, this has been an enlightening thread!
>
> So, RDF is in some sense the most "low level," and there are
> well-respected
> mappings from XTM to RDF and XLink to RDF.  So RDF is presumably
> rich enough
> to cover the others, but we can certainly imagine XLink and/or
> XTM providing
> more conveniently useful syntax/semantics for their more specialized
> purposes.  Right?

There are alot of different issues being bundled together here. RDF has a
'model' or abstract syntax along with an XML syntax. Topic Maps have a
processing model as well as an XML syntax (XTM) (this is highly simplified -
i know).

View XLink as an alternate, attribute based XML syntax appropriate for RDF.
For example my XSLT based simplified RDF parser accepts XLink'd documents
(http://www.openhealth.org/RDF/rdfExtractify.xsl)

>
> So my next question is when one or  another is most useful/appropriate.
> Does anyone have some use cases to illustrate when it makes more sense to
> use XLink (with metadata, not as simple physical links) vs XTM vs
> raw RDF to
> solve a real world problem?

For example RDDL uses an XLink syntax along with XHTML. I wrote an XSLT that
converts a RDDL document into RDF (assuming the URI references allow
conversion into a qname) http://www.rddl.org/rddl2rdf.xsl There are
syntactic issues why one might prefer one over the other. Actually in
general a linkbase might be a better way to serialize an RDF model than the
RDF XML syntax which depends on being able to convert URI references into
QNames.


>
> Getting into dicier territory, I see that RDF and XTM, and RDF
> and XLink can
> peacefully co-exist (or coop-etate, if that were a word) with one another.
> How about XLink and XTM?

XTM uses XLink.

>
> So, does the world need all three?  Just RDF?  RDF+XLink? RDF+XTM?
>
> Given the fragmentation of the XTM effort into TopicMaps.net and
> TopicMaps.org (not to mention whatever ISO is doing), is there good reason
> to believe that the various conceptions of XTM will interoperate?
> What's the
> "executive summary" for why a real-world company would want to wade into
> that mess *if* RDF+XLink can cover the same territory?

In my view, the what TM uniquely brings to the table is the processing model
(deciding that several topics refer to the same subject). Otherwise I see no
reason why TM might not have an RDF based serialization syntax.

>
> I know that some of these are probably stupid
> questions/observations, but I
> think there are a lot of people just as confused about this stuff as I am,
> so any clarification will be appreciated.
>

This stuff is extremely confusing. Its particularly hard when things just
almost work together except for little niggley syntactical issues.

-Jonathan