Re: Minimal dataset semantics

On Aug 23, 2012, at 18:02 , Richard Cyganiak wrote:

> On 23 Aug 2012, at 16:23, Ivan Herman wrote:
>> the text says:
>> 
>> [[[
>> The interpretation of an RDF dataset is the interpretation of its default graph. The presence or absence of named graphs does not affect the truth of a dataset.
>> ]]]
> 
> So the second sentence rules out any semantic extensions that make a dataset's truth dependent on the content of some named graph, right?

No. At least that is certainly not the intention, though maybe the formalization in English is not ideal. What it is meant to say: on the 'minimal' level the content of the named graphs do not have any effect on the truth of the dataset. They are like literals in many respect; at the basic simple level the 'content' of the literals do not influence any semantics for a graph (although there are entailment regimes where they do). (I must admit that the term 'quote' came to my mind from that analogy.)

> 
>> whether the mathematical formalism reflects that or not is secondary; if it does not, than it is wrong.
>> 
>> That being said, I presume you refer to the entry in the formalism that says 
>> 
>> ∀i,j, i,j=1,…,n: if ui = uj then Gi and Gj are equal.
> 
> No, I refer to the entry that says I(ui) = Gi.
> 
> Let's say I have datasets A and B, where the graph associated with u1 in A is a, and the graph associated with u1 in B is b. Let's say that a≠b.
> 
> Now from I(ui)=Gi it follows that in any interpretation that satisfies A, I(u1)=a. It also follows that in any interpretation that satisfied B, I(u1)=b. Now, can there be any interpretation that satisfies both A and B? No, because that would require a=b.
> 
> In other words, there is no interpretation -- no possible arrangement of the universe -- that can make both A and B true. In other words, A contradicts B.
> 
>> but it also says: 
>> 
>> [[[
>> In this section the “equality” of graphs in a dataset means that they are mutually inferable through simple entailment.
>> ]]]
>> 
>> So yes, the current quoting semantics does not do your entry (b) below, but I am not sure I understand your statement about the contradiction.
> 
> Does the explanation above clear this up?

(I presume that, in your argumentation, a=b and a≠b means in the sense of mutually inferring through simple entailment.)

If all that is true, what you seem to say is that the mathematics is wrong. This is, in fact, the partial vs. complete quoting of the other thread. The formalism provides a complete quoting, and I see people have spend some time discussing it while I was asleep. 

But I repeatedly said that I do not *really* care about the mathematics; if it is wrong, then we have to throw it out... I would not want to loose our energy on that. What was important was the English prose part.

> 
>> But all that may be moot with the recent proposals of Antoine, so this issue may be overran by events.
> 
> I'm afraid I didn't manage to follow everything that went on in the thread. What was the outcome? Is there a new minimalist proposal on the table?

Yes and no. The proposal was rather to give a formal way of defining the interpretation of a dataset with a minimal one around (formalized by Antoine).

Antoine provided some sort of a short sentence of what is happening:

"All RDF graphs in an RDF dataset don't mean the same thing. To be explicit about what they mean, we provide a vocabulary that specify the semantics of each graph. We call the semantics assigned to a <name,graph> pair its entailment regime, because it determines what entailments are valid for that pair. For example ..."

and I try to summarize what is happening in:

http://www.w3.org/mid/BBEB8470-DD31-4D19-B602-34E23A169488@w3.org

(where I quoted you but it turns out that this was a mistake, the page attributed to you was Antoine's. Sorry...)

Personally, I am little bit afraid of the complexity of the thing to be honest. Hence it may not fly.

Cheers

Ivan


> 
> Best,
> Richard
> 
> 
> 
>> 
>> Ivan
>> 
>> 
>> On Aug 23, 2012, at 17:06 , Richard Cyganiak wrote:
>> 
>>> If we define a semantics for datasets at all, it should be such that:
>>> 
>>> 1. It formalizes what is denoted by graph IRIs
>>> 2. Interesting semantic extensions can be made by third parties or future WG
>>> 3. It doesn't preclude any reasonable semantic extensions
>>> 
>>> The problem with the quoting semantics is that it fails #3. The quoting semantics makes it a contradiction if dataset A and dataset B contain the same graph IRI with different associated graphs. We cannot do semantic extensions that produce useful additional entailments from a contradiction.
>>> 
>>> No other proposed semantics does have that problem. All of the other proposed semantics can be easily extended with an additional clause that requires equal graph names to be associated with equal graphs.
>>> 
>>> Therefore, the quoting semantics is *not minimal*. Quite the opposite. It is not a "weak" semantics at all, because it makes it very easy to derive contradictions, and contradictions are *very strong* semantic effects.
>>> 
>>> I also agree with Antoine that formalizing the notion of "no semantics" is pointless.
>>> 
>>> My conclusion is that our viable options are: 
>>> 
>>> a) to say nothing regarding the semantics of datasets, or
>>> b) to define a minimal version of a "truth-based"/"entailment-based" semantics (where [[ :i1 { G } ]] entails [[ :i1 { G' } ]] if graph G entails graph G').
>>> 
>>> Best,
>>> Richard
>> 
>> 
>> ----
>> Ivan Herman, W3C Semantic Web Activity Lead
>> Home: http://www.w3.org/People/Ivan/
>> mobile: +31-641044153
>> FOAF: http://www.ivan-herman.net/foaf.rdf
>> 
>> 
>> 
>> 
>> 
>> 
> 
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
FOAF: http://www.ivan-herman.net/foaf.rdf

Received on Friday, 24 August 2012 10:16:45 UTC