Re: RDF-ISSUE-110 (g-box): A proper term for the concept formerly known as ?g-box?? [RDF Concepts]

Tom, the reason we can't give you a *definition* of the concept is that it is a natural kind term. We didn't invent this idea; it isn't ours to define.  We just observed that, out there in the real world, there were by now a lot of things that stored RDF or emitted RDF representations when suitably poked, and were identified by IRIs, or at any rate thought of as having an enduring identity, and yet were labile, in that sense that the actual RDF they contained/emitted/whatever might vary with time. Because they are labile, they can't be called RDF graphs (even though a lot of people were calling them that.) Because they are real things that people identify and use, we want to be able to refer to them, so we need a single overarching name for the general concept. So, just as with "resource" and "document" and "HTTP endpoint" and many other terms of art, we don't actually give a *definition* of them, because once anyone gives a definition they fix the meaning in stone and thereby exclude things that one might want to include but had not thought of, or had not encountered, when the definition was written. We thought of a number of evocative names (g-box, RDF document, rdf surface) but they all were seen as too limiting, as tending to exclude some useful cases (eg 'document' suggests something essentially textual, but an RDF Source could be a process which scours RDFa from Web pages on demand) and so the deliberately bland, uniformative term was chosen. Again, this is a frequently used strategy in such circumstances. (The W3C could have save itself a lot of grief if it had used the bland word "thing" instead of the jazzier term "resource".)

Seems to me that the phrasing used by Richard, viz, "persistent but mutable" resources which emit RDF representations, is right on the mark. If you want a definition, that is the best you can ever get. So, if you have something that emits RDF when poked, is going to last for a while (so you might want to identify it with a URI) , and is liable to emit different RDF at different times (or even if not, if it is for other reasons the same *kind* of thing as one that could be mutable in this sense), then whatever else it happens to be, and however you implement it, it is an RDF Source. 


On Nov 20, 2012, at 7:47 AM, Thomas Baker wrote:

> On Tue, Nov 20, 2012 at 09:52:03AM +0000, Richard Cyganiak wrote:
>> As stated in the issue description, the term will be used in the informative
>> introduction to RDF Concepts, specifically in sections 1.5 and 1.6:
>> It will not be *normatively* defined, and will not be used anywhere in
>> normative text in any of the documents.
> Understood.
>> The bar for some informative text in the introduction is considerably lower
>> than the bar for a normative concept that becomes part of the formally
>> defined RDF data model. This informative use does not preclude any future WGs
>> from normatively defining the same concept, or a similar concept, using a
>> different name.
> Also understood.  But any term introduced in RDF Concepts, however
> "informatively", will surely be picked up elsewhere.  People will want to know
> more about RDF Sources than just their name and that they are "persistent but
> mutable" resources that may change their state.
>>> For example, can we
>>> just substitute "space" with "source" from the following [1]?
>>>   An RDF space [=> source] is anything that can reasonably be said to
>>>   explicitly contain zero or more RDF triples and has an identity distinct
>>>   from the triples it contains. 
>> No. ?RDF source? has no strict definition, 
> Understood.  But if the concept did not have at least a rough, informal
> definition, why would we even give it a name?
>> and questions of containment or identity are not addressed.
> But the text, as it currently stands [1], is not entirely silent on these issues.
> It says they can be named with IRIs, and the phrase "source or container", in
> the absence of further explanation, hints (ambiguously) that "source" may be
> synonymous with "container".
> [1]
>>> I am curious how we ended up with "source"
>> Go to the tracker, find ISSUE-110, read the thread.
> That thread is pretty short and does not capture the preceding discussion.  In
> my reading, the ISSUE-110 thread says _that_ the group ended up with "source",
> but not how.
>>> it was not one of the
>>> half-dozen or so options bandied about a few months ago [1]:
>> That document was a personal draft produced by the four people named at the
>> top of the document, and reflects their personal opinions, preferences and
>> biases. It is in no way a complete or even particularly representative
>> reflection of what was discussed in the WG.
> I have no doubt.  But it would be great if the WG could say a bit more about
> this issue, somewhere other than in RDF Concepts, if only a paragraph or two
> drawing out a few key points from that long and, at times, insightful
> discussion.  If RDF Concepts has called them RDF Sources, those two paragraphs
> somewhere should say something about this choice of terms.  Was it just a coin
> toss?  Even if the group cannot agree on anything more than a name, the reasons
> why this was so would surely be of interest -- more, at any rate, than just
> "for more information, see public-rdf-wg".
> Tom
> -- 
> Tom Baker <>

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

Received on Wednesday, 21 November 2012 02:36:26 UTC