Re: webont-01 'rename schema' proposal

Dan--

I've no problem with the idea of a name change as suggested by WebONT 
(and am prepared to grit my teeth and make all the Primer changes), but 
I don't especially like the specific change being contemplated, because 
it seems to ignore some practical issues.  The following are some 
considerations I think are important in any resolution of this business:

1.  It is *mandatory* in my opinion for there to be a *short* way of 
writing (a) the name of the language itself, and (b) the things written 
in the language.  "RDF schema" passes those tests.  "RDF Vocabulary 
Description Language" fails the (a) test, and "vocabulary description" 
(or "RDF vocabulary description") fails the (b) test (not by a whole 
lot, but it does).  WebONT has "OWL" for (a) and "ontology" for (b).  We 
need something similarly short, unless we expect no one to actually use 
this language, instead jumping directly to OWL (and if what we're 
talking about substituting for the thing described isn't an "ontology", 
if only a very impoverished one, we need a clear understanding of what 
it takes to qualify as an "ontology" versus whatever it is we're 
defining).  (NB:  "VDL" has been used already for something else.)

2.  I don't really care for encouraging the OWL editors to talk about 
"vocabulary description" without some additional amplification of what 
that means.  Our Semantics document talks about a vocabulary being a 
collection of URIrefs used in RDF graphs, and it seems to me any RDF 
that provides information about those URIrefs (by using them as subjects 
in RDF statements) "describes" that vocabulary, whether it uses the RDFS 
vocabulary or not.  So "vocabulary description" in this context has a 
particular meaning that I'd like to see elaborated a bit.

3.  If we're going to change either the name of the language, or the 
thing described by the language, or both, I really have a problem 
keeping the old namespace with "rdf-schema" in it.  I know that these 
are supposed to be opaque identifiers, but that text isn't opaque enough 
for this purpose.

--Frank


Dan Brickley wrote:

> Discussing with Brian and others the proposal from WebOnt via Jim Hendler,
> http://www.w3.org/2001/sw/RDFCore/20030123-issues/#webont-01
> raised: http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0335.html
> ...I would like to get some discussion going on the costs/benefits of 
> finally moving away from the use of the name 'RDF Schema' for our 
> vocabulary description language.
> 
> The WebOnt comment is:
> [[
> i. Although this document is called RDF Schema we think that the 
> title "RDF Vocabulary Description Language" would be clearer, and 
> make the difference from XML Schema (used for validation) more 
> evident.
> ]]
> 
> After talking this through with Brian, and prior discussions with 
> Ivan Herman and others regarding the difficulty of presenting a coherent 
> picture of the SemWeb technology stack, I am 
> increasingly inclined (albeit warily) 
> towards accepting the proposal from WebOnt.
> 
> This would be a costly exercise, both for the WG and editors, but also 
> in terms of those in the wider world who have invested time, energy and 
> braincells on 'rdf schema' technology. Whatever we do will have costs. My 
> change of mind is based on the view that making the move to drop 'schema' 
> terminology now puts an immediate burden on the WG and editors (in our 
> group and in OWL), but leaves us in a state where the future might be 
> increasingly coherent as tutorials, tools, demos etc migrate away from 
> the 'schema' terminology. If we stick with calling this thing 'schema', 
> and we get to REC, we're stuck with it.
> 
> But but but... what about other wgs? other w3c specs? other communities 
> who are talking about their rdf vocabularies as 'rdf schemas'? If the 
> terminology of 'rdf schema' were to simply vanish from the W3C RDF / 
> SemWeb specs, we risk creating a whole lot more confusion, at least in the 
> next year or two. 
> 
> So, put bluntly, I don't know what best to do.
> 
> The current RDFS/DVL spec sits on the fence. We say that it is the 
> RDF Vocabulary Description Language 1.0: RDF Schema.
> 
> I wonder if the balancing act we are currently performing could be tweaked 
> slightly, to encourage folk to think more in terms of vocabularies 
> (and eventually ontologies) rather than schemas. Strategies we might consider 
> include:
> 
>  * never using 'rdf schema' in noun form, ie avoiding talk of their being 
>   things that are 'rdf schemas' (while leaving it in as a _name_ for the 
>   basic rdf vocabulary description language defined by w3c, just as OWL is 
>   the name for W3C's 2nd RDF-based VDL).
>  * Encourage OWL spec editors to use the terminology of 'vocabulary description'  as a way of providing a conceptual bridge between the work of our specs and 
>   in  the OWL specs
>  * Accompany this with outreach efforts to help RDF early adopters (esp 
>   vocab creators) explain this transition. We could do this through RDF IG
>   and other outreach efforts, ie needn't be a work item for RDF Core.
> 
> Having typed this I'm feeling happier that there is a middle path, where
> 'schema' doesn't entirely dissapear from the spec, but we accelerate its
> phase-out by more explicitly moving to 'vocab' based terminology.
> 
> Dan
> 


-- 
Frank Manola                   The MITRE Corporation
202 Burlington Road, MS A345   Bedford, MA 01730-1420
mailto:fmanola@mitre.org       voice: 781-271-8147   FAX: 781-271-875

Received on Saturday, 24 May 2003 13:10:32 UTC