Re: Concepts (almost) ready

On Dec 16, 2013, at 9:17 AM, Markus Lanthaler <markus.lanthaler@gmx.net> wrote:

> I think Concepts is now (almost) ready to be published as a PR and if
> possible I would like to vote on that on Wednesday. The only remaining
> things to be done/discussed are:
> 
> - Reference SPARQL; this is David's ACTION-326, thus CCed :-P
> - Normative/informative reference to Sematics [1]
> - the question regarding section 5.4 I posted before [2]
> - ISSUE-148 "IRIs do *not* always denote the same resource"; see also [3]
> - ISSUE-165, datatype maps

Reading through it one final time, I noticed a few possible wording improvements. These should not prevent publication as a PR, but I would like to see the one marked [[IMPORTANT]]  fixed before final Rec publication. 

1.2  "Unlike IRIs...blank nodes do not denote specific resources."  I know we have been through this before, but this statement still makes me shiver, because blank nodes *do* denote resources, just as IRIs do, so a very large burden is being placed here on that word "specific". This would be both more accurate and read better if it said "..do not identify specific resources". The document does use "identify " in this sense in other places (eg 1.3), and it is widely used in the sense in other foundational Web documents, in particular the WebArch document which discusses IRI collisions.

1.5 "It does not deal with time, and..." I think this is better omitted. It could be misunderstood as denying the following paragraph. The rest of the sentence says what needs to be said more clearly. 

?? 1.6  Might it be helpful to put an informative reference to Antoine's semantic survey right after "There are many possible uses for RDF datasets." ?

4.  "Blank nodes MAY be shared between graphs in an RDF dataset."  Um, I now see that this can be understood in different ways. What I think (hope) is intended here is, that if the same bnodeID is used in two graph documents in the same dataset, then that means that those two graphs do share a bnode. But what it could be read as saying is that whether or not they share the bnode is optional: they might or they might not. Which would be a very unfortunate reading. 

This could be clarified by adding the following sentence: "When blank node identifiers are used in notations defining a dataset, their scope should be understood to be the entire dataset rather than a single graph." 

[[IMPORTANT]]  In the Note:  "... the graph name does not formally denote the graph."  This is wrong as stated, and kind of dangerous in a normative section as it seems to prohibit graph names from *ever* denoting graphs. Also the use of "formally" seems to suggest two kinds of denotation (formal and informal) which is misleading. Any of these alternatives would work:

the graph name need not denote the graph.
the graph name is not required to denote the graph.
RDF does not require the graph name to denote the graph.

I like the last one as it is unambiguous and leaves open the possibilty of something else requiring this :-) 

Sorry to notice all this so late in the process. IMO all these are editorial, clarifying rather than changing any meanings. 

Pat




> 
> 
> Apart from ISSUE-148 and ISSUE-165, all comment resolutions have been
> accepted by the commenters.
> 
> 
> Cheers,
> Markus
> 
> 
> [1] http://lists.w3.org/Archives/Public/public-rdf-wg/2013Dec/0202.html
> [2] http://lists.w3.org/Archives/Public/public-rdf-wg/2013Dec/0201.html
> [3] http://lists.w3.org/Archives/Public/public-rdf-wg/2013Dec/0204.html
> 
> 
> 
> --
> Markus Lanthaler
> @markuslanthaler
> 
> 
> 

------------------------------------------------------------
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 Tuesday, 17 December 2013 05:25:39 UTC