Re: rdfs:Class vs. daml:Class ?

>Ian,
>
>Ian Horrocks wrote:
>
>>  On March 18, Steven Gollery writes:
>>  > Ian,
>>  >
>>  > Thank you for the explanation. I realized after writing the 
>>original message that it was
>>  > basically only a coincidence that the majority of DAML 
>>ontologies I was working with
>>  > were using rdfs:Class: I see now that only some of the 
>>ontologies on the list at the
>>  > DAML site do this and the majority use daml:Class.
>>  >
>>  > About your paper: has there been any official response to the 
>>idea of using rdfs(fa) to
>>  > redefine the model for rdf and rdfs?
>>
>>  Not much. There is a proposal to indicate some triples as being
>>  "non-asserted", which I believe can be seen as a very weak form of
>>  layered architecture (but when I suggested this to Pat he was rather
>>  dismissive).
>>
>
>I'm sorry to hear that:

Im sorry to hear it, as well, since I have been advocating such a 
change to the RDF WG. Im not sure where Ian got that impression from, 
but in any case please do not take casual email comments about casual 
conversations as in any way indicative of any WG policy decision. I 
also want to make it easier to align RDF with other architectures, 
and think that allowing some kind of non-asserted triples would be a 
low-cost (and low-tech) interim device that would make it possible 
for others to hack freely in their 'layers' while preserving the 
semantic integrity of RDF.  But any changes to RDF must be made 
cautiously, I have found, which takes time. Even this small a change 
to the the basic RDF model has implications for the syntax, for the 
RDF graph model, for the XML syntax, etc. etc.. , and so cannot be 
done instantly. Semantically it is trivial, of course.

>  some of the work that I'm doing could conceivably benefit from an rdf
>model that could be aligned more closely with the UML layered 
>architecture. I'll just have to
>work something out.

BTW, notice that while RDF does not enforce any 'layering', it also 
does not forbid it. It is quite possible to enforce layering (or any 
other syntactic constraint) in some namespace and still be correct 
RDF, as DAML illustrates.

Pat Hayes

>I appreciate the answers you've given me on this and on other 
>questions about DAML.
>
>Steve Gollery
>
>>
>>  Ian
>>
>>  >
>>  > Thanks again,
>>  >
>>  > Steve Gollery
>>  >
>>  > Ian Horrocks wrote:
>>  >
>>  > > On March 15, David Martin writes:
>>  > > > I think some clarification of this question would be helpful 
>>to others of us as
>>  > > > well.  I was eager to read the Pan and Horrocks paper 
>>mentioned below, but the URL
>>  > > > is broken:
>>  > > >
>>  > > > Not Found
>>  > > > The requested URL 
>>/jpan/Zhilin/download/Paper/Pan-Horrocks-rdfsfa-2001.pdf was not
>>  > > > found on this server.
>>  > > > Apache/1.3.9 Server at imgcs.cs.man.ac.uk Port 80
>>  > > >
>>  > > > Can someone please post a working URL for this paper?
>>  > >
>>  > > Sorry, but they just moved to a new server and everything is still a
>>  > > bit wobbly. You can get the paper from my site on different server:
>>  > >
>>  > > http://www.cs.man.ac.uk/~horrocks/Publications/download/2001/rdfsfa.pdf
>>  > >
>>  > > I suppose that while writing I should try to answer Steven's question:
>>  > >
>>  > > >
>>  > > > Thanks,
>>  > > >
>>  > > > David
>>  > > >
>>  > > > Steven Gollery wrote:
>>  > > >
>>  > > > > Please excuse another naive newby question....
>>  > > > >
>>  > > > > In the DAML language definition, it looks like rdfs and 
>>rdf are being
>>  > > > > used as the metamodel: daml:Class, for example, is an instance of
>>  > > > > rdfs:Class. But if that is the case, I would expect that the Class
>>  > > > > definitions in a DAML ontology would be instances of daml:Class.
>>  > > > > Instead, the sample ontologies that I've seen use rdfs:Class either
>>  > > > > exclusively or (as far as I can tell) interchangeably with 
>>daml:Class.
>>  > >
>>  > > You are right that in many cases rdf is being used as the "metamodel"
>>  > > (i.e., to describe the DAML+OIL language itself), but things are a
>>  > > little confused as some parts of rdf are used directly in DAML+OIL,
>  > > > e.g., range and domain, subClassOf.
>  > > >
>>  > > > >
>>  > > > > I understand from the Pan and Horrocks paper at
>>  > > > > 
>>http://img.cs.man.ac.uk/jpan/Zhilin/download/Paper/Pan-Horrocks-rdfsfa-2001.pdf
>>  > > > > that there is a layering problem in the RDF/RDF(S) definition that
>>  > > > > prevents a clean division between successive metamodel 
>>levels. Is the
>>  > > > > relationship between rdfs:Class and daml:Class somehow connected to
>>  > > > > this?
>>  > >
>>  > > More or less. The extension of a DAML+OIL class should be a set of
>>  > > individuals (well, strictly a set of objects that are denoted by
>>  > > individual names) and not, say, a set of properties, as could be the
>>  > > case for an rdfs:Class. Because of the lack of layering in the rdf
>>  > > architecture there is no way to enforce this, so daml:Class is just a
>>  > > label given to the subset of rdfs:Classes that have the property we
>>  > > want. Note that in the daml+oil-ex.daml file, daml:Class is used
>>  > > extensively. Also note that many of the "meta" properties in the daml
>>  > > language definition have daml:Class as a range/domain so that classes
>>  > > used in daml ontology will often be implicitly of type daml:Class.
>>  > >
>>  > > > >
>>  > > > > I suppose all I'm really asking is: when would I use 
>>rdfs:Class and when
>>  > > > > would I use daml:Class? And if it doesn't matter, why are 
>>there two of
>>  > > > > them?
>>  > >
>>  > > Always use daml:Class. I hope I explained why there are two.
>>  > >
>>  > > Ian
>>  > >
>>  > > > >
>>  > > > > Thanks for your patience,
>>  > > > >
>>  > > > > Steven Gollery
>>  >


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Monday, 1 April 2002 13:16:22 UTC