Re: [Proposal] [SKOS-Core] Initial term_status values

* Charles McCathieNevile <charles@w3.org> [2004-07-29 01:27-0400]
> 
> Hmm. So essentailly we expect to be able to change any definition of any part
> of SKOS? I am not sure that this is a good idea - and in particular I suspect
> that Concept is actually stable, and probably the {broad,narrow}er*
> properties...
> 
> what do you think?

Defining 'broader' could probably chew up a bunch more time.

As for Concept, depends what you mean by stable. I've been meaning to
suggest that the rdfs:comment isn't quite right actually (sorry
Alistair!).

<rdfs:comment>A concept is any unit of thought that can be defin
ed or described.</rdfs:comment>


This is a strong cognitive science claim about the nature and structure
of (human?) thought. It also relies on notions that some but not necc
all 'units of thought' are definable or describable. 

I think "Concept" is stable in that we all somehow have the rough notion 
of what we're trying to say there. Whether the specific text of
rdfs:comment used totally captures that is another matter. 

"A Concept is the idea of something"
"A conceptualization of a thing"
"The idea of an idea"
...etc.

Regarding "broader", we currently have:
  <rdfs:comment>This property carries weak semantics.  It may be u
sed to state that the object is in some way more general in meaning than
the sub
ject.  Essentially it provides a means of organising concepts into a
hierarchy (
tree), without being restrictive about the exact semantic implications
of the hi
erarchical structure itself.  Extend this property to create properties
that car
ry stronger semantics, but may be reduced to a hierarchical structure
for simple
 visual displays.</rdfs:comment>

(and no specification of domain or range).

I would be suprised if this definition remained precisely unchanged over
the next year. For starters it is defined (textually if not formally) in
terms of Concept, so what we say about Concept affects meaning of this
property. If concepts are really 'units of thought' then that
constrains, perhaps, the ways in which they can be 'organised into
hierarchies'. 

It would be pretty awful if every little tweak to these definitions
meant that a whole new namespace had to be deployed. Dublin Core went
down that path once and folk are still, years later, writing code to
merge back together the dc 1.0 and dc 1.1 namespaces, which are for all 
practical purposes identical in content, just described slightly
differently.

Dan

Received on Thursday, 29 July 2004 04:45:19 UTC