[Sigia-l] Distributed thesaurus?

[trav at ciaheadquarters.com]

At 10:19 PM 9/22/2002 +0200, Lars Marius Garshol wrote:
> * peter at poorbuthappy.com
> | But every topic already has the URI http://domain.com/map.xml#topicid
> | Wouldn't that suffice? Topicmaps could use that as psi.
> 
> Then you could refer from a topic map into XFML, but XFML maps still
> couldn't point out of their own universe. It means that you wouldn't
> be able to use published subjects, for example, since you can't point
> to a subject indicator.

See, this is the problem I was addressing. XFML fails as a complete solution
cause it doesn't point to a subject indicator, topicmaps fail as a complete
solution cause they need subject indicators to communicate. There's not
enough abstraction in topicmaps, nor in XFML, for those formats to be the
answer to my original question.

As I read this thread and do more research, it's apparent that I shouldn't
have asked for a syntax, or even a specific kind of graph, cause that's not
really what I was after anyway. It's obvious that a massive distributed
vocabulary can be created _independent of format_ simply by defining a
relationship between my term ( http://mysite.com/vocab#myterm ) and your
term ( http://yoursite.com/thesaurus#yourterm ). The only shared standard
is HTTP URI format, and the relationship is whatever I define it to be.
(As Lars correctly points out, predefined relationships like BT/RT/etc will
eventually be insufficient.) This is all we need to make a global, reusable,
collaborative ontology of terminology.

What I really want to know is, why isn't this happening already? Why are
html pages linked into a worldwide web while vocabularies are not? Those of
you who design vocabularies or other forms of thesauri, what do you think
about adding a bunch of related-term markup that links your terms with terms
in other vocabularies (or publishing XTM/XFML to similar effect)? Anyone who
sympathizes with the difficulty of constructing a faceted thesaurus (see
http://www.eleganthack.com/archives/002780.html ), do you see value in
tapping into a vast global network of vocab terms? Do you want to do it but
feel you should wait until there's an accepted standard?

The model I'm talking about is as simplistic as you can get, which offers
the flexibility necessary for it to be a global resource. The problem you
might see with such flexibility is: My software won't be able to understand
the format of your term! It could be a totally different XML grammar. An
XFML topic might point to a Zthes term, and break everything.

I don't consider that a problem. That's something only the software
architects should design for. The information architect's task is the model.
I believe we put way too much emphasis on formats. Standards are necessary,
but we have to recognize that people will always use whatever format works
for them. What's important is not that we agree on a format, but that we
make our formats easy to import, by providing pieces of logic (software
modules) that operate on them. (this could spin off into another of my
favorite topics. you get the idea.)

I hope that's a better statement of what I was asking for. A complete
package solution like topicmapping, or the CERES project I'd found, is far
too restrictive an interface to make me happy. But why don't I see more
thesauri with external links?

trav