Re: contexts, graph names and so on

On 10/11/11 18:20, Richard Cyganiak wrote:
> Andy,
>
> Summary: We're actually saying the same thing.
>
> On 10 Nov 2011, at 14:51, Andy Seaborne wrote:
>>> So you're saying:
>>>
>>>     :g --???-->   X --denotes-->   graph
>>>
>>> and there should be some way of stating what relation --???-->   is for a particular dataset.
>>>
>>> Wouldn't it make more sense to say:
>>>
>>>     :g --denotes-->   X --???-->   graph
>>>
>>> because :g is an IRI and as such probably is supposed to denote something already. Then some obvious choices for --???-->   would be sameAs, hasRepresentation, and hasPrimaryTopic.
>>
>> It makes more sense (to me at least) to fix the second part as naming the the graph value.
>
> It actually works out exactly the same. It's really like this (pardon the ASCII art):
>
>    :g ----xxx---->  {ggg}
>     |                |
> denotes          denotes
>     |                |
>     v                v
>   thing ---yyy--->  graph

Good art - I can cope with morphism diagrams.  (Brings back memories of 
my Riemann Surfaces course.)


The difference is that yyy isn't a property.

We are agreed (I think) that one step is fixed.

If we have owl:sameAs at the denotes step:

:g --xxx--> X --owl:sameAs--> graph

vs:

:g --owl:sameAs--> Y --zzz--> graph

and xxx and zzz are properties.


> The upper half of the picture is in the space of syntax, the lower half is in the space of values.
>
> Starting at the graph IRI :g, your version goes right-then-down. The implicit X in this case is really a graph literal (it denotes the graph). Hence I write it as {ggg}.
>
> Starting at the graph IRI :g, my version goes down-then-right. The implicit X in this case is really the thing that is actually denoted by the graph IRI. Hence I write it as “thing”.
>
> Now, xxx and yyy are the same relationship – only that one is in syntax land, and one in value land. Xxx is a name for the relationship, and yyy is the actual relationship. Xxx denotes yyy.
>
> If you think about the syntax of TriG files, then it makes sense to imagine some URI xxx between the<http://graphIRI>  and the {graph literal}. If you think about the semantic relationship between the referent of<http://graphIRI>  and the actual graph (set of triples), then that's yyy.

I understood you (orginally) to be saying X --???--> {graph} is a property.

>
>> I have an action to write some short use cases but these previosu messages do fit the model as I think of it:
>
> Looking forward to the writeup.
>
> Richard
>
>
>
>>
>> Subject: The "Rolling Snapshots" Pattern and Vocabulary
>> http://lists.w3.org/Archives/Public/public-rdf-wg/2011Oct/0152.html
>>   Sandro
>>
>> Subject: Time-varying g-boxes : a dataset pattern
>> http://lists.w3.org/Archives/Public/public-rdf-wg/2011Oct/0148.html
>>   Andy
>>
>> 	Andy
>>
>

Received on Friday, 11 November 2011 14:01:05 UTC