W3C home > Mailing lists > Public > www-archive@w3.org > February 2004

Re: Named graphs etc

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Thu, 26 Feb 2004 13:37:58 +0200
Message-Id: <39A231EA-6850-11D8-B072-000A95EAFCEA@nokia.com>
Cc: "Pat Hayes" <phayes@ihmc.us>, <www-archive@w3.org>, <chris@bizer.de>
To: "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com>

On Feb 23, 2004, at 17:32, ext Jeremy Carroll wrote:

> I have been being beaten up further from the pragmatic wing  by Chris
> Bizer - He is beginning to convince me ...
> My understanding of his key arguments is as follows:
> - use vocab as much as possible, not syntactic mechanisms
>     : impacts graphset tag name
>               asserted attribute

Fair enough. Though we could consider the attribute value as
a short hand, which generates a second anonymous graph containing
the statement about assertion of the first graph.

We may, though, end up with an infinite recursion. I.e., we have
a graph X that is asserted. In order to say that X is asserted,
we have to have another graph X' containing a statement that
X is asserted. But if X' is also asserted, we have to have another
graph X'' with a statement saying that X' is asserted, etc., etc.


> - meaning of vocab is shared: for example log:implies is ill-formed  
> RDF,
> because to make sense of it you have to use a second interpretation  
> that
> interprets concepts differently from the first. So the liar paradox is
> resolved by saying that RDF is not a general purpose meta language for
> logic.
>    : impacts semantics, outlaws log: vocab
> - the de re/de dicto argument is hence resolved largely in favour of  
> de re,
> but to some extent is seen as a red herring. For provenance  
> information it
> is certainly helpful to regard the provenance statements as being  
> about a
> 'de dicto' graph; but this is not because we doubt that the author may  
> be
> using the URIs differently from us.
>    : impacts semantics
>   I guess one way of doing this is to say that the interpretation of  
> the
> name of a graph is the pair consisting of the graph and a set of the
> interpreted triples, and the predicate selects which member of the  
> pair is
> interesting...
> - whether we believe any graph or not is a matter for the trust layer.
> Example below. Chris's PhD (in progress) is on the trust layer.  His
> approach is very pragmatic - users have a trust policy that can take  
> into
> account facts about the contents and or use of a graph to determine  
> whether
> to believe it. Such a policy is not a logically mechanism, much more
> prgamatic and down-to-earth. This makes the distinction between  
> asserted and
> non-asserted graphs redundant.
> - the graphset in trix is merely a syntactic necessity for XML  
> documents,
> and should not convey any meaning. If we want to talk about a  
> collection in
> RDF there are plenty of mechanisms. So if we want to talk about a  
> collection
> of graphs we use one of those. Thus, the graphset tag should be  
> changed to
> something semantic-free (e.g. trix). It should be explicitly stated  
> that the
> URL used to retrieve a trix document refers to the document and not to  
> the
> contained graphs. The point here is to stop before going on the  
> slippery
> slop to graphsetsetsets

Fair enough. I have no problems changing the name of the root element
for the sake of avoiding confusion.

> His words are at:
> http://lists.w3.org/Archives/Public/www-archive/2004Feb/att-0050/ 
> namedGraphD
> iscussion009.pdf
> Example: Provenance Chains
> Peter states, that Chris said that Andy said, that Monica Murphy is a
> person.
> G1 (Monica ex:hasName "Monica Murphy".
>     Monica rdf:type ex:Person)
> G2 (G1 ex:saidby Andy.
>     G1 ex:DocumentURL Doc1.trix.
>     G1 dc:date "2/10/2004")
> G3 (G2 ex:saidby Chris.
>     G2 ex:DocumentURL Doc2.trix.
>     G2 dc:date "2/10/2004")
> G4 (G1 dc:author Peter.
>     G2 dc:author Peter.
>     G3 dc:author Peter.)
> G5 (G4 dc:author Peter.
>     G4 dc:date "2/10/2004")
> Depending on our trust policy we might believe some, all or none of the
> above triples...

Right. That's pretty much how I've thought about trust, not that
I've delved into the topic anywhere near as deeply as Chris or others.


> Jeremy


Patrick Stickler
Nokia, Finland
Received on Thursday, 26 February 2004 06:45:55 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:42:36 UTC