Re: (A PRIORITY) Exchanging the contents of RDF stores

On 03/02/12 19:48, Pat Hayes wrote:
>
> On Feb 3, 2012, at 5:50 AM, Andy Seaborne wrote:
>
>>
>>
>> On 01/02/12 19:10, Pat Hayes wrote:
>>>
>>> On Feb 1, 2012, at 6:30 AM, Andy Seaborne wrote:
>>>
>>>> http://www.w3.org/2011/rdf-wg/wiki/TF-Graphs-UC#.28A_PRIORITY.29_Exchanging_the_contents_of_RDF_stores
>>>>
>>>>
>>>>
>>>>
>>>>
One of our prioritised UC is "Exchanging the contents of RDF stores".
>>>>
>>>> I'd like to prose that TriG, without additional semantic or
>>>> annotation assumptions, meets this UC.  The WG should resolve
>>>> to cover this usage pattern.
>>>>
>>>> This codifies the situation where no annotations of the graph
>>>> tagging relationship exists.
>>>>
>>>> It implies that quads are not an internal implementation
>>>> matter but a way systems can exchange data.
>>>
>>> I would submit that in this case, the semantics does need to be
>>> specified with more precision than it is at present. So I
>>> strongly resist this proposal as stated, with the inclusion of
>>> the phrase "without additional semantic ... assumptions". With
>>> that phrase removed, I would support it.
>>>
>>> Pat
>>
>> Pat,
>>
>> Q1/ Are you making this case for just TriG or for all quads forms?
>
> All of them, yes.
>
>>
>> Q2/ Can the semantics be specified in a different document?
>
> Oh, sure.
>
> I just meant, I think we have (the WG has) a responsibility, if we
> endorse any kind of quad-describing *interchange* notation - or the
> very idea of interchanging quads at all, in fact - then we have to
> bite the bullet and specify a semantics (or maybe several semantics)
> for quads. Alternatively, of course, if quads arent even in our RDF
> attention focus, we don't need to do that: but then we shouldnt be
> talking about them at all.
>
>>
>> We have things like the DBpedia dumps already online:
>>
>> http://dbpedia.org/Downloads37
>>
>>
>> To have TriG, the basic syntax, not being able to express basic
>> dumps in the way N-Quads can is possible but its going to place an
>> education cost on explaining that N-Quads ->  TriG can't be done
>> without additional
>>
>> I would like to see vocabulary on top of the basic syntax, not
>> requiring it's use.  Like the time example we had in the telecon,
>> sometimes, there will be data without the full modelling details.
>
> Yes, I would like this also. So that various kinds of meaning can be
> switched on or off, or selected, by (relatively simple and
> easy-to-add) "switches" like a special triple or some such. But this
> is going to take some work in the basic RDF machinery to make it
> possible.
>
> Sorry, in haste, more later.
>
> Pat
>

OK - so we're not so far apart.

I would like to see everyone published collection of graph to define the
semantics used but, practically, it's not always going to be done. A
separate document is particularly interesting because it does not modify
the original contents so checksums apply for example.  or a "system graph".

I think there will be several choices - we have several reasonable ones
even in the WG so choosing one is highly unlikely IMO.

It would be valuable to standardise quad format(s) even if we don't
entirely rely on quads.  Like Turtle or N-triples have become the de
facto syntaxes, defining quad forms just so the details are pinned down
is valuable to people.

Looking forward to hearing about the changes to the RDF machinery,

	Andy

Received on Sunday, 5 February 2012 16:24:55 UTC