Re: Equality of RDF bags

On 29.10.2011 21:40, Pat Hayes wrote:
>
> On Oct 28, 2011, at 9:42 AM, Dieter Fensel wrote:
>
>> At 11:19 AM 10/28/2011, Antoine Zimmermann wrote:
>>> Graph 2a and Graph 2b are not equivalent either in terms of formal semantics,
>>
>> Is it then fair to say that the formal RDF semantics is "broken" not reflecting the intuitive semantics of RDF?
>
> No, but it is fair to say that RDF is a descriptive/logical language rather than a specification/programming language. And of course this is what it was designed to be, and this is exactly what the RDF semantics specifies it to be.  I suspect that the "intuitive" semantics you are referring to is in fact a mistaken intuition, gotten by thinking of RDF as something like a programming language. Wrong way to think.

+1

And, to strengthen your argument, you could have pointed Dieter directly 
to Sec. 3.3.2 of the RDF Semantics spec at

     <http://www.w3.org/TR/2004/REC-rdf-mt-20040210/#Containers>

which not only introduces RDF Bags, but also has some comprehensive 
discussion on the topic discussed here, making it clear that the weak 
(or "broken") semantics for containers (and also for lists aka 
collections, and for reification, btw.) was not by accident but was 
clearly intended and must have been discussed in depth by the old RDF 
working group.

In addition, the mentioned section contains a nice example which shows 
how trying to formalize "intuitive semantics" for Bags can easily lead 
to very unintuitive results. In a slightly reformulated version: If 
intuition tells one that

   ex:bag rdf:type rdf:Bag ;
     rdf:_1 ex:alice ;
     rdf:_2 ex:bob .

should imply (due to the missing order of the members within bags):

   ex:bag rdf:type rdf:Bag ;
     rdf:_1 ex:bob ;
     rdf:_2 ex:alice .

and if intuition further tells one that "rdf:_1" and "rdf:_2" should be 
functional properties (that's at least what many people claim to be 
required for the RDF collection properties "rdf:first" and "rdf:rest"), 
then one would get (in OWL at least, where owl:sameAs is defined):

     ex:alice owl:sameAs ex:bob .

Or, in general, it would then turn out that /all/ members of a given RDF 
bag would be mutually equal resources. Not very intutitive, IMHO...

But even if this would be acceptable to some, or if there would not be 
such peculiar ramifications, then another issue would still be that 
properly capturing the "intuitive semantics" for containers in a formal 
way might easily lead to pretty complex semantics, which would possibly 
be out of scope for "light-weight" (typically datalog'ish rule-based) 
reasoners as used today for RDFS reasoning. For comparison, OWL 
semantics requires that the order of elements in enumerations does not 
play a role, so, e.g., from

   ex:E1 owl:oneOf ( ex:alice ex:bob ) .

and

   ex:E2 owl:oneOf ( ex:bob ex:alice ) .

one should be able to infer

   ex:E1 owl:equivalentClass ex:E2 .

You really /will/ get this result by applying full-featured OWL DL 
reasoners, but you won't get it from many existing "light-weight" RDF 
entailment-rule reasoners. In particular, you do not get it from the now 
popular OWL 2 RL/RDF rules, as defined at

 
<http://www.w3.org/TR/2009/REC-owl2-profiles-20091027/#Reasoning_in_OWL_2_RL_and_RDF_Graphs_using_Rules> 


So, no one should ask for such "intuitive semantics", unless he is ok 
with requiring everyone to apply heavy-weight reasoning engines to 
safely obtain the "expected intuitive" results from it. And I know 
pretty well that some people in this thread *would* be very unhappy with 
this outcome. :-)

> Pat

Cheers,
Michael

-- 
Dipl.-Inform. Michael Schneider
Research Scientist, Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: michael.schneider@fzi.de
WWW  : http://www.fzi.de/michael.schneider
==============================================================================
FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts
Stiftung Az: 14-0563.1 Regierungspräsidium Karlsruhe
Vorstand: Dipl. Wi.-Ing. Michael Flor, Prof. Dr. rer. nat. Ralf Reussner,
Prof. Dr. rer. nat. Dr. h.c. Wolffried Stucky, Prof. Dr. rer. nat. Rudi 
Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
==============================================================================

Received on Saturday, 29 October 2011 20:37:44 UTC