- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Wed, 06 Feb 2002 13:00:36 -0500
- To: seth@robustai.net
- Cc: www-rdf-logic@w3.org
From: "Seth Russell" <seth@robustai.net> Subject: Re: reification test case Date: Wed, 6 Feb 2002 09:35:11 -0800 > From: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com> > > > Well one might argue that > > > > 1/ there is at most one statement with a given predicate, subject, and > object > > Ok, but can we restate that? There is at most one *triple* with a given > predicate, subject, object. > > 2/ each triple, i.e., each potential statement, has at most one > reification > > in the form of a member of rdf:Statement > > I believe that rdf:Statement is a class and not a set, so your use of the > word 'member' here is problematic. Don't forget there is normally no global > identity to a node of rdf:type statement; it is just a Bnode, and as such > there is nothing that is special about the rdf:subject, rdf:predicate, and > rdf:object properties of that Bnode - they do not confer any special kind of > identity to the node, and the current MT does not smush them together. OK, but to completely fit in with M&S something like the following is needed: 1/ There is exactly one triple with a given pred, sub, and obj, so there can be at most one element of Statements with a given pred, sub, and obj. 2/ For each triple, consisting of a given predicate, subject, and object, there is at most one resource that is the sub of four elements of Statements; one with pred rdf:type and obj rdf:Statement, one with pred rdf:predicate and obj predicate, one with pred rdf:subject and obj subject, and one with pred rdf:object and obj object. > > Why? Well precisely so that one can uniquely identify a potential statement > > and make statements about that statement. > > Now I suppose we could put some kind of identity on a reified statement that > could imply that it referred to the one and only triple in the current tidy > graph, or even perhaps to the one and only abstract triple in the sky - but > me thinks we should invent a new property arc for that purpose. Perhaps > that might be a way for us to have it both of our ways :) But why not use the current M&S stuff as it is? If you want multiple ``reifications'', then why not use some other mechanism? > >For example, one could in this > > way relate the statement to log:false via log:truthvalue, or use it in a > > log:implies construct. > > I don't see why we can't do that right now. I could write > > <rdf:description> > <rdf:type>:Statement</rdf:type> > <rdf:subject>:Gore</rdf:subject> > <rdf:predicate>:wonThe</rdf:predicate> [ <rdf:object>:election</rdf:object> ] > <log:truthValue>False</log:truthValue> > </rdf:description> > > which holds for all such statings. > > But I could also write: > > <rdf:description> > <rdf:type>:Statement</rdf:type> > <rdf:subject>:Gore</rdf:subject> > <rdf:predicate>:wonThe</rdf:predicate> > <dc:author>:Seth</dc:author> > <log:truthValue>False</log:truthValue> > </rdf:description> > > which holds for a smaller collection of statings. HUH? What is the difference here? How do you distinguish between these two uses? As far as I can see both resources have equivalent status as far as RDF is concerned. > Seth Russell Peter F. Patel-Schneider Bell Labs Research
Received on Wednesday, 6 February 2002 13:01:28 UTC