Re: New issue - Meaning of URIs in RDF documents

pat hayes wrote:
>
> Point taken.  Suppose we said something like, think carefully about
> what you take a URI to mean, the softbots will get confused unless
> y'all choose meanings that are related in some fairly close way.

This is supposed to be the reason that we are to decide on what an HTTP URI
identifies e.g.. a description of a galaxy vs. a galaxy  yet it seems that
RDF and OWL have already solved that problem. When using RDF/OWL we don't
have a problem distinguishing between a class definition (i.e. the XML/RDF)
vs. a class extension -- this distinction is obvious and is a direct
consequence of the model theory.

Softbots won't get confused because they presumably will be programmed to
know the difference. In the case where they might get confused why can't be
just disambiguate by saying:

<http://example.com/foo/galaxy> rdf:type web:document .

or else

<http://example.com/foo/galaxy> rdf:type astro:galaxy .

Now I am all for cleaning up the language, if for no other reason than to
tone down these permathreads, but does any software actually care?

SW software already knows how to deal with such issues. Human readable web
software doesn't seem to care about such issues -- the web is merrily
humming away...

Now for this specific issue when an RDF/OWL application dereferences a
URIref e.g.

http://www.w3.org/TR/2003/WD-owl-guide-20030331/wine.owl#TableWine

it will get back:

<owl:Class rdf:ID="TableWine">
 <owl:subClassOf>
  <owl:Class>
    <owl:intersectionOf rdf:parseType="Collection">
      <owl:Class rdf:about="#Wine" />
      <owl:Restriction>
        <owl:onProperty rdf:resource="#hasSugar" />
        <owl:hasValue rdf:resource="#Dry" />
      </owl:Restriction>
    </owl:intersectionOf>
  </owl:Class>
 </owl:subClassOf>
</owl:Class>

An OWL/RDF application doesn't get confused by this, this means the class
TableWine, not a piece of XML i.e. element owl:Class[@rdf:ID="TableWine"]
because RDF/OWL have a semantics which provides for such an interpretation
of the XML syntax.

The problem is that ultimately we need to look *outside* RDF to get the
meaning of certain URIs. For example the statement:

person:pat legal:ownerOf house:id122312 .

The property legal:ownerOf presumably has a legal definition which is
defined in legalese, supposedly a variant of natural language.

I am not sure how that TAG might solve these issues -- generally the issue
of "social meaning" -- or is there a smaller issue? I think I understand
part of this issue, and this part already seems to be solved by RDF/OWL...
am I missing the middle ground between this and the not immediately
tractable issue of connecting SW definitions to natural language
definitions?

Jonathan

Received on Saturday, 26 July 2003 11:25:47 UTC