On 09/24/2013 01:07 PM, Jeremy J Carroll wrote:
On Sep 24, 2013, at 6:31 AM, Sandro Hawke <san...@w3.org
<mailto:san...@w3.org>> wrote:
I'm now confident that you and I (and Jeremy) agree the problem we're
trying to solve in this thread is this: people seem to want to have
different properties on one "graph" than on another, even when the
"graphs" happen to have the same triples.
I am sorry, I actually don't agree.
The phrase '''even when the "graphs" happen to have the same
triples.''' indicates that any actual example is likely to be
contrived and artificial.
The point about these contrived and artificial examples is that they
demonstrate that my use cases for graph naming (i.e. the use cases
that matter to me) tend to be about referring to the graph as a
representation of a resource, where there is one step of remove.
To totally abuse RDF we might agree a new mechanism for publishing novels.
A novel might be published as a single triple RDF graph with a blank
subject, predicate being rdf:value and the object being the text of
the novel.
The participants of this thread might each independently come up with
the following earth-shattering great novel
[ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
And once the movie of the book grosses many millions, we all end up
in court, arguing over who wrote the novel - contradicting the
penultimate sentence.
Sandro produces his dataset with
eg:Sandro dc:creator "Sandro Hawke".
eg:Sandro [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
I produce mine with
eg:Jeremy dc:creator "Jeremy Carroll".
eg:Jeremy [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
etc., so that the court is presented with a merge of datasets:
eg:Jeremy dc:creator "Jeremy Carroll".
eg:Sandro dc:creator "Sandro Hawke".
eg:Dan dc:creator "Dan Brickley".
eg:Pat dc:creator "Pat Hayes".
eg:Gregg dc:creator "Gregg Reynolds".
eg:Jeremy [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
eg:Sandro [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
eg:Pat [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
eg:Gregg [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
eg:Dan [ rdf:value
"""Once upon a time, there was a consortium, which agreed on a
semantics for graph naming. And they all lived happily ever after. The
END!"""
].
====
I really don't think there is any contradiction here, even if
dc:creator is a functional property.
The court would have to fall back onto precedence, and I would produce
this e-mail from www-archive and win the case!
http://lists.w3.org/Archives/Public/www-archive/2013Sep/0053.html
====
My point with this example is that the identity condition that Sandro
is asking for is contrived;
I'm confused, since the point I was trying to make is that the identity
conditions are contrived (ie lack real use cases) unless we bring in
change-over-time.
but my use case is of a publishing system where the data being
published is in the form of an RDF graph, and there is metadata is
data about that graph - but not really stuff about the graph itself
but more about the pair - the naming of the graph.
So let's go back to that. Give me an example that shows three things:
the triples happen to be the same, the metadata must remain distinct,
and there is no change over time. As I think about it now, I'm
beginning to think it's impossible.
-- Sandro
Jeremy