RE: Asunto: Re: [www-rdf-interest] <none>

I promised myself I'd lurk a while, but as ever I can't resist. Hole
picking welcomed, I'm still on the steep part of the learning curve and
need every bit of help I can get.

Might as well introduce myself while I'm here. I've recently started
trying to use RDF as a layer of a system I'm working on to support the
development and use of environmental simulation models. A lot of people
are working on linking existing models, on uncertainty analysis, and on
tools for QA and auditing the model development process, and more. These
tools are generally uncoupled, and I think RDF provides a good
opportunity for loosely coupled federations of tools.

Naturally this requires that the tools all make (in the areas of
overlap) the same ontological commitments; this is a simple fact of
integrating modelling, I believe, and one that the majority working on
this stuff seem to overlook. Using RDF provides some structure to the
search for a sufficiently general and flexible shared ontology (by which
I mean ontology - structured view of what exists in the world - not
schema) and the tools to encode it (as a schema).

Anyway, back to the issue in hand:

On Wed, 2004-01-14 at 08:36, Victor Lindesay wrote:
> Joshua wrote:
> > True, you can define a mapping, and that is nice to have if you get
> > stuck in the inevitable situation where two different people use
> > different terms to mean the same thing.  But it's even nicer 
> > to just use
> > a term that others are more likely to understand in the first place.
> > IOW, the ability to map between terms is a feature to be resilient in
> > the face of antisocial vocabularies rather than an excuse to be
> > antisocial.
> Can we stop using the term anti-social in this context please? It's RDF
> software which has to deal with this and software doesn't give a damn.
> I suppose using the same argument it's anti-social that people speak
> different languages and don't all speak and understand English!

To willfully coin new terms which are simply aliases is surely not the
same as speaking a different language. 

If two communities develop overlapping vocabularies in parallel (and,
like languages, they are unlikely to be isomorphic, and there will be
things which can be said in one but not the other and vice versa) then
the tools which OWL provides for declaring that two terms identify the
same resource (or sit in some more complex relationship) are an
invaluable aid to progress.

Creating a new vocabulary when what you want to say can be well
expressed in an existing one is more akin to getting together with your
neighbours and agreeing to develop and speak your own language. Or
perhaps, in less of a fantasy situation, like working with colleagues in
a new branch of sience, building a world view and a language as you go.

To walk up to someone in the street and try to communicate with them in
a language of your own invention is certainly weird, and could in some
circumstances be regarded as anti-social, but I'd hesitate to use
"anti-social" in this context. There are reasons to ignore existing
vocabularies, particularly in a world where the available languages are
at such an early stage of development. 

It nonetheless seems quite obvious to me (as a newcomer, and certainly
not an expert, so called or otherwise) that if everyone makes every
effort to use existing vocabularies *when they are capable of expressing
what needs to be expressed* the world will be a better place for it. If
everyone rolls there own, we'll need as much OWL floating around
describing aliases as there is data in the first place.

> Can we have a more 'joined up' thinking on this please?

I can't see that there is a break in the thinking. Aliases are a fact of
life, and so the OWL tools are essential, but they shouldn't be
generated willfully, and it would save a lot of time in the long run if
people spend a little time in the near term checking what exists before
creating new vocabularies. Isn't that why SchemaWeb exists?


Received on Wednesday, 14 January 2004 06:11:35 UTC