On May 16, 2007, at 12:06 PM, Pat Hayes wrote:
What you say ABOUT the statement is up to you, and requires an
ontology of statement-making or belief or responsibility. For one
suggested approach to all this (which avoids the rather clunky RDF
reification mechanism) see
http://www.hpl.hp.com/techreports/2004/HPL-2004-57R1.html
I had a couple of questions regarding this proposal.
1) I'm trying to understand what it means to refer to a graph within
itself. The following would seem to be paradoxical - if marketing is
supposed to not use any statements in :G1, then it can't use the
disallowedUsage statement, which means it can (assuming that they can
legally use them unless otherwise noted)
:G1
{ _:Monica ex:name "Monica Murphy" .
_:Monica ex:email <mailto:[EMAIL PROTECTED]> .
:G1 pr:disallowedUsage pr:Marketing }
2) I want to nest graphs, for example (the si are sets of triples)
:as_published_in_paper_10020102
{ :citing_paper_30301002
{ s1 }
:citing_paper_44357422
{ s2 }
:claim1
{ s3 }
...
}
It would seem this is not directly supported, or did I misread.
If I wrote it as
:as_published_in_paper_10020102
{ s1. s2. s3 }
:citing_paper_30301002
{ s1 }
:citing_paper_44357422
{ s2 }
:claim1
{ s3 }
:citing_paper_30301002 rdfg:subGraphOf :as_published_in_paper_10020102
:citing_paper_44357422 rdfg:subGraphOf :as_published_in_paper_10020102
:claim1 rdfg:subGraphOf :as_published_in_paper_10020102
Would I be accurately expressing what I'm trying to, namely that the
same set of triples s1 is in both the
graphs :as_published_in_paper_10020102 and :citing_paper_30301002 ?
If not, what is the recommendation for accomplishing the same sort of
thing? (or shouldn't I want to do this)
3) There is continued talk about the lack of semantics of reification
being a problem and I'm trying to understand why this would be any
better (aside from possibly being less verbose). The kind of useful
thing I have seen cwm's log:implies, that can have a whole graph
included or not included as part of reasoning. But this is outside
the abilities of RDF and OWL. Given that I don't see why some
mechanism build on reification wouldn't work just as well (e.g. bag
of rdf:Statements).
Thanks,
-Alan
