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

Re: Sandro's proposal VS RDF Datasets

From: Pat Hayes <phayes@ihmc.us>
Date: Wed, 2 May 2012 14:03:01 -0500
Cc: Antoine Zimmermann <antoine.zimmermann@emse.fr>, RDF WG <public-rdf-wg@w3.org>
Message-Id: <636B24D1-2477-4640-A7EB-FC56D5757966@ihmc.us>
To: Sandro Hawke <sandro@w3.org>

On May 2, 2012, at 10:38 AM, Sandro Hawke wrote:

> On Wed, 2012-05-02 at 16:59 +0200, Antoine Zimmermann wrote:
>> My interpretation is that when one writes:
>>     :g {
>>          :c  rdfs:subClassOf  :d .
>>          :x  rdf:type  :c .
>>     }
>> it's a statement that the two triples are true "according to :g" (or
>> in 
>> "context" :g if you prefer this word, or in "graph labelled :g") and 
>> from this follows that "in graph :g" the triple ":x  rdf:type  :d ."
>> is 
>> true, 
> I don't think it follows, since I don't think there is any sense to the
> idea of triples being "true in a graph".    

Now I am on Antoine's side, sigh. 

Don't think 'true in a graph', but 'true in a context'. The "graph name" in Antoine's semantics is really a context name, and the dataset asserts each graph *in the context determined by the "graph name"*. So entailments, merges. etc.. all work provided they are all in the same context, but not necessarily across contexts. 

Think of a world in which each context determines the meanings of IRIs, so you can only know that  :a means the same as :a when they both occur inside (triples which are assserted in) the same context. I know you don't LIKE this vision, but it does make sense, and this intuition drives all of Antoine's examples. 

> Is triple (:a :b :c) true in the graph {:a :b :c; :d :e} ?    The
> question seems malformed to me.

IF the graph defines a context (a special case) then this means basically that the graph entails the triple. Which for simple entailment means that it is in it (or has an instance in it, to be very exact., so (:a :b _:x) is "in" - better to say, is *true* in - (:a :b :c; :d :e)  But this context-being-a-graph idea seems to me to be very limiting. If we are going to have contexts. then let's be more liberal than this.)


> I see some nearby concepts that do make sense to me:
>   Is triple (:a :b :c) in the graph {:a :b :c; :d :e} ?   (This is the
> normal notion of something being an element in a set.)
>   Is triple (:a :b :c) in the g-box which currently contains
> {:a :b :c; :d :e}?   (This is the notion I think is most useful for
> datasets, and what my proposals are using these days.)
>   Is triple (:a :b :c) true of some situation where {:a :b :c; :d :e}
> is true?   (This makes some sense, but I don't understand the technical
> details.)
> Of course, the answer is "yes" in each of these cases, but what about
> the triple (:a :b: :f)?   For that, in case one, the answer is No.  In
> cases two and three, it depends on information we don't have here.
>      -- Sandro

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 Wednesday, 2 May 2012 19:03:35 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:02:04 UTC