W3C home > Mailing lists > Public > public-rdf-wg@w3.org > September 2012

Re: RDF-ISSUE-98 (graph-dataset-semantics-unified): Should the semantics of RDF graphs and the semantics of RDF datasets be combined into one unified semantics? [RDF Semantics]

From: Pat Hayes <phayes@ihmc.us>
Date: Fri, 14 Sep 2012 00:53:53 -0500
Cc: public-rdf-wg@w3.org
Message-Id: <470AD428-6D8D-4D39-A2A2-A203DDE86FD9@ihmc.us>
To: Antoine Zimmermann <antoine.zimmermann@emse.fr>
On Sep 13, 2012, at 8:16 AM, Antoine Zimmermann wrote:

> ISSUE-98 is not trivial.

I disagree. With your semantics proposal and the equivalence of a graph to a dataset with no named graphs, it is entirely trivial to unify the semantics. In fact, they are in effect already unified. 

> My vote at the moment is for not changing the RDF semantics document beyond simple updates and having a separate document about dataset semantics.
> If you or Pat, or anyone else, has a proposal to make a unified semantics that does not require rewriting RDF Semantics completely, then please put it in the wiki and let us discuss it.

I already made this proposal in an earlier email (http://lists.w3.org/Archives/Public/public-rdf-wg/2012Sep/0109.html). It requires no change to RDF semantics for RDF graphs. It would be trivial (and natural) to add the dataset semantics as a later section to the RDF semantics document (assuming that we adopt this semantics for datasets, of course.) There seems to be no point in having a separate document, given that the semantics document already covers simple, RDF-, RDFS- and D-interpretations. 

BTW, I have already proposed to greatly shorten the semantics document and keep it strictly concerned with describing the model theory, as well as eliminating most, if not all, of the elaborate completeness proofs and so on. In particular, the various inference rules should be put into a different document, as they are not immediately concerned with semantics. 

By the way, do you think there might be a complete set ot inference rules for database entailment? 


> AZ
> Le 13/09/2012 14:39, Richard Cyganiak a écrit :
>> On 13 Sep 2012, at 13:12, Andy Seaborne wrote:
>>> We would have to be careful to explain why we then don't have
>>> datasets-inside-datasets and "named datasets".
>> It's a somewhat arbitrary restriction in the abstract syntax that
>> makes implementations simpler while still allowing the use cases we
>> care most about to be addressed.
>>> It's not a block to the idea but keeping them separate does make it
>>> clearer where the boundary is.
>> I think that keeping them separate would make the Semantics document
>> more complicated. The Semantics document is complicated enough as it
>> is. I think making it more complicated to account for a syntactic
>> restriction is not a good idea. If writing a semantics that is more
>> general than necessary for the abstract syntax turns out to be
>> simpler, then readers are better served by the simpler thing, IMO.
>> A related case here is literals-as-subjects, which is a
>> well-motivated restriction that I never would want to remove, but I'd
>> prefer if the Semantics document would use a generalized notion of
>> RDF graphs that doesn't have the restriction, because that would
>> remove probably a page of pointless and hard-to-understand trickery
>> that is needed to work around the restriction. At least I personally
>> found that a barrier to understanding the document.
>> Best, Richard
> -- 
> 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/

IHMC                                     (850)434 8903 or (650)494 3973   
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Friday, 14 September 2012 05:54:29 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:02:07 UTC