Re: Identifying a particular object

On Oct 17, 2014 3:19 PM, "Victor Porton" <porton@narod.ru> wrote:

> (As otherwise loading an additional RDF file may turn previously valid
data contradictory) I do not merge information from several RDF files. This
way an object (URL) is completely described in one RDF file. Thus two RDF
files may have an object described by the same object URL and this is not a
contradiction, but two different objects sharing the same URL.

For purposes of discussion it may help to talk about named graphs where the
name of the graph is the IRI of the source document.

"By design, IRIs have global scope. Thus, two different appearances of
an IRI denote
<http://www.w3.org/TR/2014/REC-rdf11-concepts-20140225/#dfn-denote> the
same resource
<http://www.w3.org/TR/2014/REC-rdf11-concepts-20140225/#dfn-resource>.
Violating this principle constitutes an IRI
<http://www.w3.org/TR/webarch/#URI-collision> collision
<http://www.w3.org/TR/webarch/#URI-collision> [*WEBARCH*
<http://www.w3.org/TR/2014/REC-rdf11-concepts-20140225/#bib-WEBARCH>]. " -
RDF Concepts § 1.3.

If IRIs name, and if names rigidly designate (eg Kripke's "Groping and
Necessity") then modal logic merely buys you different properties of the
same object.

If you want to treat  Named Graphs as  collections of speech acts which you
may or may not credit, this is consistent with the minimal semantics RDF
1.1 gives for named graphs.
There are a few possible semantics given in a separate working group note;
the interpretation as lexical strings seems to be flawed because blank
nodes.

An  interpretation of named graphs that might work better would be to
consider the graph name to denote the reified form of the statements in the
graph (for the blank nodes).  You can then translate these speech acts in
to assertions such that incorrect name uses are mapped to different names.

Simon