W3C home > Mailing lists > Public > public-rdf-wg@w3.org > March 2013

Re: RDF-ISSUE-120 (set-of-triples-are-graphs): Is any set of RDF triples an RDF graph? [RDF Concepts]

From: Antoine Zimmermann <antoine.zimmermann@emse.fr>
Date: Thu, 14 Mar 2013 19:17:34 +0100
Message-ID: <5142143E.1080202@emse.fr>
To: public-rdf-wg@w3.org
So, if I have a set of RDF graphs (say, in a specification based on RDF 
semantics) I do not have any concept from RDF that I can use to treat it 
as a single graph, as far as truth is concerned.
Why should this be the case in RDF 1.1? What was fundamentally wrong 
with having such a notion? What is it trying to address?

Now, what is the truth of a set of triples? RDF 1.1 Semantics currently 
do not say, cf. the semantic condition on bnode assumes this:

"""
If E is an RDF graph then I(E) = true if [I+A](E) = true for some 
mapping A from the set of blank nodes in the scope of E to IR, otherwise 
I(E)= false.
"""

What is the scope of E when E contains bnodes from different scopes?

The other thing is: what is the scope of the merge?
A complete graph has bnodes in only one scope. Therefore, the union of 
two different complete graphs is necessarily containing bnodes in 
different scopes. So merge creates non-scoped-graphs. And by the 
definitions in Section "Notation and terminology", a surface syntax can 
only serialise scopoed graph, so a surface syntax cannot serialise the 
merge.

Should I really spend all of my time writing down counter arguments? 
Don't you (or anyone) see that there *is* a problem with the proposed 
design?


AZ.


Le 14/03/2013 18:13, Peter Patel-Schneider a écrit :
> My understanding of the semantics in the current Semantics document is as
> follows:
>
> 1/ Any set of triples is an RDF graph.
> 2/ Combining RDF graphs is done by taking the union of the triples in them.
>
> That's is, as far as the semantics goes.  I don't see any wording anywhere
> in Semantics to the contrary.
>
> Note that there is no notion of scope here at all, and none of the
> semantics depends on scoping in any way.  All of the discussion on scoping
> graphs is irrelevant, and the discussion of complete graphs could/should be
> rewritten into something like saying that unions are implied by a set of
> graphs when each graph includes either all or none of the triples in the
> union that include any particular bnode.
>
>
> The Semantics document also discusses bnode scoping, which is not needed in
> the semantics per se, but is needed in surface syntaxes and implementations
> that identify bnodes using syntactic elements that can be accidentally
> repeated.   Here bnode scope is used to determine when using the same
> syntax (e.g., a bnode identifier) results in the same bnode or results in a
> different bnode.
>
> Why is this (currently) in Semantics?  Just because it (currently) isn't in
> Concepts.
>
>
>
> In my opinion, quite a bit of the discussion of scoping in Semantic need
> not survive in either Semantic or Concepts, but it might have some
> explanatory value so I'm not arguing about removing it.
>
> peter
>

-- 
Antoine Zimmermann
ISCOD / LSTI - Institut Henri Fayol
École Nationale Supérieure des Mines de Saint-Étienne
158 cours Fauriel
42023 Saint-Étienne Cedex 2
France
Tél:+33(0)4 77 42 66 03
Fax:+33(0)4 77 42 66 66
http://zimmer.aprilfoolsreview.com/
Received on Thursday, 14 March 2013 18:18:08 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:04:26 UTC