Re: use/mention and reification

At 11:15 AM 1/23/02 +0200, Patrick Stickler wrote:
>On 2002-01-23 0:45, "ext Graham Klyne" <Graham.Klyne@MIMEsweeper.com> wrote:
>
>
> > So what is the RDF behind Lois' beliefs? (slipping into N3)...
> >
> >   [ a foaf:Person ;
> >     foaf:name "Lois" ;
> >     ex:accepts
> >       { [ a foaf:Person ; foaf:name "Superman" ] es:is ex:Strong } ;
> >     ex:doesNotAccept
> >       { [ a foaf:Person ; foaf:name "Clark Kent" ] es:is ex:Strong }
> >   ]
> >
> > This way of coding the information has the things whose interpretation is
> > possibly local ("Superman", "Clark Kent") expressed as literals, not as 
> URIs.
>
>But how then do we unify the shared beliefs of different individuals
>if we use literals rather than URI refs?
>
>E.g.
>
>    [ a foaf:Person ;
>      foaf:name "Jimmy" ;
>      ex:accepts
>        { [ a foaf:Person ; foaf:name "Superman" ] es:is ex:Strong } ;
>      ex:doesNotAccept
>        { [ a foaf:Person ; foaf:name "Clark Kent" ] es:is ex:Strong }
>    ]
>
>How do we know who "Lois" or "Jimmy" really are, and that they are
>talking about the same "Superman" or "Clark Kent" if we don't use
>URI refs. Isn't that the whole point of URI refs?

How do we ever know who "Lois" or "Jimmy" really are?  (This is in danger 
of becoming philosophical which won't help this debate.)

I understood DanBri to be saying that he wasn't yet ready to assume that 
URIs have global scope:  that person:Lois in one context means the same as 
person:Lois in some completely different context.  OTOH, without allowing 
that it seems that URI-refs don't offer anything that label strings like 
"Lois" and "Jimmy" don't also offer.  The point of my comment was to 
suggest that matters relating to personal belief of identity shouldn't 
really be expressed in terms of URIs.

>I agree with Jan's view that a URI ref denotes an RDF resource, and
>whether two or more resources denote the same "thing" in some
>universe is a separate issue. It is no different than whether or not
><dc:title> and <foo:title> mean the same thing or whether "5.0"
>or "000005" or "5.000000" denote the same value.
>
>Equality of things denoted by RDF resources in a given universe of
>interpretation is a separate issue from identity and equality
>of RDF resources in an RDF graph.

I have no argument with this...

>Reification using URI refs provides consistent identity of RDF
>resources without forcing assertion, and quoting URI refs of
>resources as literals is just going to add an unnecessary
>layer of interpretation.

I'm slightly concerned that you may be muddling assertion of a statement 
with interpretation of resources.  As things stand, a URIref-labelled node 
in an RDF graph denotes some resource (per some interpretation) 
irrespective of whether or not any statement within which it appears is 
asserted.  Non-assertion of a statement doesn't somehow suspend 
interpretation of the graph nodes used to describe it.  Maybe this is what 
DanC and DanBri are arguing for?

When I look at your example (below), I find myself worrying about what 
ex:#Superman actually denotes, and the extent to which that affects the 
truth of what you are expressing.  Absent some very special treatment of 
the reification vocabulary, there's nothing in the model theory (that I can 
see) that allows the various occurrences of "#Superman" to have different 
denotations (depending on whose belief is being described), which is what I 
think DanBri is trying to allow with his RDF-through-RDF scenario.


>Much better, I think, to say
>
>--
>
><?xml version="1.0"?>
><rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>          xmlns="ex:">
>
><rdf:Description rdf:about="#Lois">
>    <accepts>
>       <rdf:Statement>
>          <rdf:subject rdf:resource="#Superman"/>
>          <rdf:predicate rdf:resource="ex:is"/>
>          <rdf:object rdf:resource="#Strong"/>
>       </rdf:Statement>
>    </accepts>
>    <doesNotAccept>
>       <rdf:Statement>
>          <rdf:subject rdf:resource="#Clark_Kent"/>
>          <rdf:predicate rdf:resource="ex:is"/>
>          <rdf:object rdf:resource="#Strong"/>
>       </rdf:Statement>
>    </doesNotAccept>
></rdf:Description>
>
><rdf:Description rdf:about="#Jimmy">
>    <accepts>
>       <rdf:Statement>
>          <rdf:subject rdf:resource="#Superman"/>
>          <rdf:predicate rdf:resource="ex:is"/>
>          <rdf:object rdf:resource="#Strong"/>
>       </rdf:Statement>
>    </accepts>
>    <doesNotAccept>
>       <rdf:Statement>
>          <rdf:subject rdf:resource="#Clark_Kent"/>
>          <rdf:predicate rdf:resource="ex:is"/>
>          <rdf:object rdf:resource="#Strong"/>
>       </rdf:Statement>
>    </doesNotAccept>
></rdf:Description>
>
><rdf:Description rdf:about="#Superman">
>    <is rdf:resource="#Strong"/>
></rdf:Description>
>
></rdf:RDF>
>
>--
>
>Now we have the means to infer that there is some agreement
>between the views of Lois and Jimmy, yet none of those views
>are actually asserted, insofar as the expression of those
>views are concerned -- and we also see that at least some
>of their shared views intersect with asserted "real world
>knowledge".

Yes (apart from possible the bit about "real world knowledge"), but that's 
not the issue that I'm chewing on.

>The intersections between quoted and asserted knowledge from
>the above example shows very nicely in the graph produced
>by the W3C RDF validator.
>
>Eh?

I don't know what you mean by "intersections between quoted and asserted 
knowledge".

#g


------------------------------------------------------------
Graham Klyne                    MIMEsweeper Group
Strategic Research              <http://www.mimesweeper.com>
<Graham.Klyne@MIMEsweeper.com>
        __
       /\ \
      /  \ \
     / /\ \ \
    / / /\ \ \
   / / /__\_\ \
  / / /________\
  \/___________/

Received on Wednesday, 23 January 2002 06:24:24 UTC