Re: Comments on "SPARQL 1.1 Uniform HTTP Protocol for Managing RDF Graphs"

On Fri, 2011-03-18 at 13:43 +0100, Kjetil Kjernsmo wrote:
> [ . . . ]
> What does the URI 
> identify? Apart from a foaf:PersonalProfileDocument, is it an RDF Graph or an 
> RDF Document? 

The key problem is that you are making the implicit assumption that the
URI identifies exactly the same thing in all RDF graphs.  I have been
calling this myth #1:
This is only true for *one* interpretation
of *one* graph.  (And by "interpretation" I'm talking about
interpretations as described in the RDF Semantics.)  

Unless you have explicit disjointness assertions, in one graph the URI
could ambiguously identify a foaf:PersonalProfileDocument, an RDF Graph
and an RDF Document.  But in another graph that adds a few disjointness
assertions, it may identify only a foaf:PersonalProfileDocument.  And in
a third graph that adds *different* disjointness assertions, it may
identify only an RDF Document.

This is illustrated in this section:

This does *not* exactly mean that the meaning of a URI is "context
dependent".  A URI is global in scope and (should) have a globally
scoped identity.  But the fact is that above and beyond any globally
scoped identity, different graphs may *further* constrain the URI's
identity in different ways.  That's what happens when you add more
assertions to a graph, as illustrated in Figure 10a: 

This is not a design flaw of semantic web architecture, it is an
inescapable fact of life, and we need to learn to live with.  And, aside
from outright errors, this is a key reason why joining datasets with
owl:sameAs often leads to inconsistencies.  See

[ . . . ]
> Surely, a resource can't be both an RDF Document and an RDF Graph? 

As long as the classes of RDF Documents and RDF Graphs are not declared
disjoint in your graph, then a URI certainly *can* (ambiguously) denote
both an RDF Document and an RDF Graph.

Ambiguity of reference is an inescapable fact of life.  See "In Defense
of Ambiguity" by Pat Hayes and Harry Halpin: 
It does *not* necessarily mean that anyone was sloppy or erroneous.
Rather, it means that there are limits to our ability to avoid
ambiguity, and this has far reaching consequences in semantic web
architecture -- consequences that underly issues that have plagued the
community for years, ranging from the seemingly endless debates of the
TAG's httpRange-14 issue to more recent criticisms of owl:sameAs.

David Booth, Ph.D.

Opinions expressed herein are those of the author and do not necessarily
reflect those of his employer.

Received on Friday, 18 March 2011 18:45:23 UTC