Re: Property example in "Self-describing web" draft

I thought I had given you an alternative, but I can't find it on
record, so I'll just try again.

The offending text is:

'a standard, machine-processable means of describing relationships
between RDF statements,
e.g. that two seemingly differing predicates are the "owl:sameAs" each other.'

which I had said steps gratuitously outside OWL-DL, because sameAs can
only be asserted between individuals in OWL-DL, not between properties
or classes. There are countless ways to fix this, since it's just an
example. Try one of the following:
 ... that two seemingly differing properties are
"owl:equivalentProperty" to one another.
 ... that two seemingly differing properties are logically equivalent
("owl:equivalentProperty").
 ... that one property is an rdfs:subPropertyOf another.
 ... that one property is owl:equivalentProperty to another.
 ... that one class is owl:equivalentClass to another.
 ... that one class is rdfs:subClassOf another.

according to your editorial taste.

RDF and OWL don't use the word "predicate". (A one-place predicate is
called a "class" and a two-place predicate is called a "property".)
I've therefore changed "predicate" to "property". I tend to use
"relationship" synonymously with "property" so if you don't like
"property" maybe "relationship" will fly.

Regarding the first part "means of describing relationships between
RDF statements", I have to point out that relationships expressed in
RDF are rarely between RDF statements; in fact they can and do hold
between arbitrary kinds of things, and I think you mean to talk about
the general case, or at least relationships between properties, not
relationships between statements. Again there are many ways to fix
this:

means of describing relationships,
means of describing relationships between resources,
means of describing relationships between properties,
means of describing relationships between predicates,
means of describing relationships between relationships,
means of describing relationships between logical entities,
means of make statements about relationships,

according to your editorial taste.

Jonathan

On Mon, Jan 19, 2009 at 9:29 PM,  <noah_mendelsohn@us.ibm.com> wrote:
> Jonathan;
>
> Going over Self-Describing Web notes in preparation for our publication
> decision this week, I came across the attached.  Did I drop the ball on
> this, were you going to propose text, or did we decide to let it go?  I'm
> sympathetic to doing the change in principle, but it's late if there's
> likely to be any controversy about the results.  What do you think is best
> to do at this point, and if it's to make any last minute changes, can you
> propose specific text?  I do apologize that, with this coming in right
> before the holidays, I seem to have missed it while going over my To Do's
> on SDW.
>
> Noah
>
> --------------------------------------
> Noah Mendelsohn
> IBM Corporation
> One Rogers Street
> Cambridge, MA 02142
> 1-617-693-4036
> --------------------------------------
>
> "Jonathan Rees" <jar@creativecommons.org>
> 12/19/2008 08:58 AM
>
>        To:     "noah_mendelsohn@us.ibm.com" <noah_mendelsohn@us.ibm.com>
>        cc:     "www-tag@w3.org WG" <www-tag@w3.org>
>        Subject:        Property example in "Self-describing web" draft
>
>
> Re: http://www.w3.org/2001/tag/doc/selfDescribingDocuments-2008-12-03.html
> where it says:
>
> 'Indeed, RDF's Schema [RDFSchema] and OWL Ontology technologies [OWL]
> together offer a standard, machine-processable means of describing
> relationships between RDF statements, e.g. that two seemingly
> differing predicates are the "owl:sameAs" each other.'
>
> The reason to avoid going outside of OWL-DL (using sameAs on
> non-individuals) is not to make any kind of endorsement of DL, but
> simply because there is no reason not to use DL here, and it would be
> less distracting if the question, which is irrelevant to this
> presentation, simply had no way to come up.
>
> The easiest fix is to use owl:equivalentProperty instead of owl:sameAs.
>
> A better example might simply use a more useful, less distracting, and
> less controversial relation, for example:
>
> e.g. that one property is an "rdfs:subPropertyOf" another
>
> I would change "predicate" to "property" to follow established RDF and
> OWL usage. However I appreciate that "predicate" or "relationship"
> might be a more intuitive term (in this context they're all roughly
> synonymous), and I would defer to your editorial taste.
>
> I have a hard time with "relationships between RDF statements".
> Relationships hold between arbitrary resources, not (just) statements,
> and I don't think you really mean relationships between statements
> here.
>
> How would the following example work for you?
>
> 'offer a standard, machine-processable means of describing
> relationships that hold between resources, e.g. that the
> foaf:publications relationship holds between persons and documents.'
>
> Domain and range (what I've written informally as "holds between")
> sound to me more like an ordinary description of a property than
> rdfs:subPropertyOf or owl:equivalentProperty, both of which seem
> comparatively exotic.
>
> (You would want to spell out the full URI for foaf:publications
> (http://xmlns.com/foaf/0.1/publications).)
>
> Best
> Jonathan
>
>
>

Received on Tuesday, 20 January 2009 19:12:20 UTC