Re: Putting metadata in the "default" graph Re: Dataset Syntax - checking for consensus

On 2012-09-27, at 12:59, Andy Seaborne wrote:
> On 27/09/12 12:52, Steve Harris wrote:
>> On 2012-09-27, at 12:24, Andy Seaborne wrote:
>>> On 27/09/12 12:13, Steve Harris wrote:
>>>> On 2012-09-27, at 12:08, Andy Seaborne wrote:
>>>>> On 26/09/12 17:41, Lee Feigenbaum wrote:
>>>>>> We need to support it for compatibility, but I think it's a
>>>>>> mistake to specify that anything important be put in that
>>>>>> graph.
>>>>> There are two uses cases: you and Steve emphasis the
>>>>> complicated case of multiple graphs collected from many
>>>>> places.
>>>> If that's not the case, why are you bothering with named graphs?
>>> Because different customers have different requirements.  I come
>>> across both cases, but not necessary with the same data.
>> I can imagine that situation in theory, but what real-world usecase
>> leads you to use named graph, but where the metadata applied to the
>> whole document?
>> It seems like a corner case.
>> Database dumps are one possibility (but then you'd often have both
>> dataset-level and graph-level metadata), but those are more commonly
>> in nquads formant, from what I've seen.
> Lee's claim was that the default graph was "perhaps not widely used." and "it's a mistake to specify that anything important be put in that graph."
> I was commenting that it's not "compatibility" when there is an important use case.

OK, but I don't think that literally "metadata in the default graph" can be an important use case, as so many systems don't support default graphs. Otherwise they would support, or users would complain, or something.

Metadata in the TriG/whatever file might be an important use case - I've not seen it done, but I guess people do. That's different though.

- Steve

>>>>> The simple case is one graph.  For that, making the publisher
>>>>> go through "naming" is just overhead for them.
>>>> But that's "just" RDF, isn't it? In that case I don't see the
>>>> issue.
>>> Yes, it is, and SPARQL is an RDF query language.
>>> Requirement - publish data.
>>> The idea of a "TriG" sub-ecosystem and an "RDF" sub-ecosystem is
>>> not a happy thought.
>> Strongly agreed, though I don't think many people share this
>> concern.
>> - Steve

Steve Harris, CTO
Garlik, a part of Experian
+44 7854 417 874
Registered in England and Wales 653331 VAT # 887 1335 93
80 Victoria Street, London, SW1E 5JL

Received on Thursday, 27 September 2012 12:04:24 UTC