RE: a get-together of some RDF and TopicMap folks

Hi Dan,

Dan said:
Following the energy and good will built up
at the "RDF vs. Topic Maps" session[1]
at Extreme Markup Languages in Montreal,
a few of us got together by phone/IRC this
morning (well... morning in my time zone).

The notes we managed to take are available at

We might edit them a bit to make them look nicer
or something... but most likely we won't get
around to it.

Didier replies:
Great to know that finally a bridge is under construction between the two
islands. This is with joy that I finally see a potential fruitful exchange
between these two world.

Recently, I made a presentation in Paris about merging the RDF and topic
maps. You can get the paper at the GCA site. But what basically what was
published in June is:

<your-topic xlink:type="extended" xlink:title="this is a topic">
<your-resource xlink:type="locator" xlink:title="a resource"
<prop1>a property</prop1>
<prop2>an other property</prop2>

"your-topic" is anything you want as an element name since the topic map is
an architectural form, idem for "your-resource". The above construct allows
to include properties inside a locator and thus provides meta information
about this resource. This construct has the secondary advantage to be easily
processed by XSLT. If however it is necessary to have a properties
declaration external to the locator then the classical rdf:description can
do the job.

My personal conclusions:
The topic maps can gain a lot by using the RDF to define the properties. RDF
frames are quite compact, easy to process with XSLT. However, it would be
also useful if an rdf description could be inherited by any element. The
mechanism could be similar to the xlink inheritance mechanism for instance
something similar to:

<your-resource rdf:type="description" xlink:type="locator"

The above element named "your-resource" could inherit from the xlink
extended and from the rdf description structure and behavior. This would
imply that the above construct can simultaneously be processed by an xlink
processor and an rdf processor. Quite a powerful construct for knowledge
management indeed.

Maybe I should build an other experiment in my Didier's labs and write an
article about it...This way, people could see in action what are the
implications of such construct.


Received on Tuesday, 19 September 2000 12:16:57 UTC