Re: (proposal) was Re: defn of Named Graph

On Sep 27, 2013, at 7:45 AM, Sandro Hawke wrote:

> On 09/27/2013 03:56 AM, Andy Seaborne wrote:
>> Overall, a discussion note would be good.
>> 
>> On 26/09/13 20:02, Sandro Hawke wrote:
>>> We have two levels of specs, right?  There are the ones that are
>>> thoroughly reviewed and proven to be implentable, which we call
>>> Recommendations.
>> 
>> The terminology "named graph" is in use in RECs and RECs-to-be. Redefining terminology might be appealing in the short term but I feel only leads to confusion long term.
>> 
>> rdf-11-concepts:
>> """
>> Each named graph is a pair consisting of an IRI or a blank node (the
>> graph name), and an RDF graph.
>> """
>> 
>> and in JSON-LD (sec 7, Data Model)
>> and in SPARQL (sec 12.1.2)
>> 
> 
> Thanks for pointing this out.    I was doing a little wishful thinking and misremembering RDF 1.1 Concepts as steering clear of this definition that I've been complaining about for years.   (I see now Section 1.6 incorrectly says that the RDF Graphs in the Dataset "are called Named Graphs".   Perhaps that sentence should be changed to including the word "informally" or something, since formally speaking it's false.)
> 
> Anyway, this leaves us with some options:
> 
> (1) Try to convince ourselves and the Director that this is not a subtantive change.    In favor of this view is the fact that it wont change a line of code.   Against this view: it seems to be something that people care about a lot, so changing it after they've reviewed the document is uncool.
> 
> (2) Be silently inconsistent in our use of the term Named Graph among our publications.
> 
> (3) Make the apology/explanation in the Note somewhat bigger.
> 
> (4) Have the Note use a different term ("gbox" or "surface")
> 
> (5) Forget the whole thing and move on.

Let me (only just now surfacing from a few days off-web) suggest:

(6) Keep the idea but modify the terminology. As Andy points out, "Named Graph" already has too much baggage (and in any case is weirdly misleading in two ways). But we already have some useful terminology in the LC specs: Concepts 1.1 defines "RDF Source" as a nameable thingie that emits or embodies RDF (and might be labile). So lets take this and run with it, and talk about "fixed RDF Source" as a non-labile species of RDF source, and say that the graph name denotes that. Or even just forget about the "fixed" part:

"3. We note that RDF Datasets can be used to state which triples are in certain RDF Sources.

4. We define a class (eg rdf:BasicDataset) of the Datasets which have those semantics."

This avoids all the issues that Andy raises, and has the advantage that it has a very smooth segway into datase... sorry, things like datasets but which are allowed to be updated. 

Pat

PS. I could probably draft a Note on this if you are getting burnout. 


> I'm torn between 4 and 5, myself.   I'm not sure I could live with 2 or 3 (in terms of dying of embarassment).   In favor of 5 is the fact that I've already spent all the time I had to write this Note arguing about it the past two weeks.
> 
>        -- Sandro
> 
> 
>>    Andy
>> 
>> 
> 
> 
> 

------------------------------------------------------------
IHMC                                     (850)434 8903 home
40 South Alcaniz St.            (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile (preferred)
phayes@ihmc.us       http://www.ihmc.us/users/phayes

Received on Friday, 27 September 2013 16:36:05 UTC