Re: blank nodes (once again)

I think Pat's "BTW" actually picks out a really important issue that starts to separate out things that are sometimes getting confused.
(Either that, or I am about to be told how I don't understand :-) )
We are not just talking about RDF Graphs, but also serialisations (which is, if I recall, the genesis of these questions).
So, as Alan confirms, in an RDF graph, the interpretations remain the same, whether the nodes are blank or not - if they are not, then you have a URI for them, but that does not affect the interpretation.

However, when Ivan types the following into his email:
Suppose that I have the following graph:

:joe :has [ a :dog ]

And suppose that I also have the following one as well:

:joe :has [ a :dog ]

He is not showing RDF, but serialisations.
So now I need to know about some pragmatic, even engineering issues.
For example:
Are they serialisations I am about to assert into a store?
Or are these serialisations that are the result of queries?
Were these queries to the same store/named graph?

Answers to these may allow me to infer a little more from essentially non-RDF context.

In particular, if these are the results of two queries from the same named graph, then it may be, for example, that the graph has an interpretation that :joe has two dogs, if there is an owl:differentFrom between the two [a :dog]s (again, any b-nodeness is not relevant here). In the absence of an owl:differentFrom, I think that an isolated RDF graph such as that does not tell us anything, as in an open world :joe can always have other dogs. Of course there may be other bits to the graph that tell us other things, but we can't see those.

But I can't tell any of this from the un-decorated serialisations in the email.

In the end, I think we come back to a general problem of graph serialisation.
If I take a graph of any kind and use a particular serialisation method, and communicate that somewhere else, and then de-serialise, do I end up with the same graph?
For most serialisations, the answer will be "no".
There is then possibly a second question, that is related to what the graph we are dealing with is meant to represent.
In RDF, when will the graphs that result have the same interpretations?
This is of course an interesting question, which I am sure can be answered by others.

Hope I have not just confused things further :-)
Best
Hugh

On 17 Mar 2011, at 03:59, Alan Ruttenberg wrote:

On Wed, Mar 16, 2011 at 2:14 PM, Reto Bachmann-Gmuer
<reto.bachmann@trialox.org<mailto:reto.bachmann@trialox.org>> wrote:

On Wed, Mar 16, 2011 at 5:34 PM, Pat Hayes <phayes@ihmc.us<mailto:phayes@ihmc.us>> wrote:


Quite. BTW, this would also be true if you had two graphs with URIs
instead of blank nodes, even if they were different URIs.

:joe :has [:dog1 a :dog]
:joe :has [:dog2 a :dog]

You would still not know that the two different *names* for dogs did or
did not name the same dog. Maybe (or maybe not)

:dog1 owl:sameAs :dog2

Yet you would know something more about the world, that either :joe has two
dogs or that :dog1 owl:sameAs :dog2. Of course, :joe might be a self-owned
dog also known as :dog1 and :dog2.

Reto,

How do you conclude this? As far as I can tell, given the RDF
semantics, all of these and more, were possibilities before you put
anything in a graph, and remain possible interpretations subsequently.

The whole part of this thread that tries to reason from "same
knowledge" to "same graph" seems the wrong kind of thinking. For one
thing it's hard to see this crowd seeing

G1: [a owl:Restriction; owl:onProperty :foo; owl:someValuesFrom owl:Thing]
G2: [a owl:Restriction; owl:onProperty :foo; owl:minCardinality 1]

as being the same graph, despite them having the same "knowledge"
according to the OWL semantics.

-Alan


--
Hugh Glaser,
              Intelligence, Agents, Multimedia
              School of Electronics and Computer Science,
              University of Southampton,
              Southampton SO17 1BJ
Work: +44 23 8059 3670, Fax: +44 23 8059 3045
Mobile: +44 78 9422 3822, Home: +44 23 8061 5652
http://www.ecs.soton.ac.uk/~hg/

Received on Thursday, 17 March 2011 10:35:55 UTC