Re: naming properties and classes in RDF

From: Libby Miller <Libby.Miller@bristol.ac.uk>
Subject: Re: naming properties and classes in RDF
Date: Fri, 16 Nov 2001 16:32:36 +0000 (GMT)

> > 
> > I'm not sure what you consider to be redundant here.  The only things that
> > are typed twice are <rdf:type>, <_:1>, and <_:2>.
> 
> Hi Peter
> 
> Thanks for the reply. I wasn't very clear. I'm not concerned with the
> syntax, I'm bothered that designing schemas which have similar or
> identical names for certain properties and classes is confusing for
> authors, and also seems...wrong. It's not clear that you need to know
> both the sort of link between two entities and the typing information
> about the entities. The latter doesn't seem to add anything, except
> maybe in some visualizations where it may clarify what you
> are looking at. For certain queries it might enable you to make finer
> distinctions (but for most queries I've done, and certainly in this
> schema which is very flat, typing in queries is not needed).
> 
> Does this ring any bells with anyone?

Yes, but perhaps exactly in the opposite direction of what you are getting at.

I view the ability to have (or need) both properties and classes to be an
important benefit, not a serious liability.  Yes, you can get this wrong,
and it is easier to get it wrong if you use similar names, and, moreover,
RDFS will not complain if you get it wrong.  This is where good modelling
tools come into play---ones that will complain if you use a class name as
the name of a relationship.  (Note that this is NOT an RDFS error, just a
potential modelling confusion.  People who do not understand the difference
will have serious problems living in the open world of
RDF/RDFS/DAML+OIL/most logics.)

> thanks again
> 
> libby

peter

Received on Friday, 16 November 2001 12:00:24 UTC