W3C

RDF & SPARQL WG

07 August 2025

Attendees

Present
AndyS, doerthe, gkellogg, james, ktk, niklasl, ora, pchampin, Souri, TallTed, tl, william-vw
Regrets
fsasaki, olaf, pfps
Chair
ora
Scribe
TallTed, pchampin

Meeting minutes

<gtw> I'll be joining the call today via phone only from a 310 number.

Identifying issues to solve before CR 1

<AndyS> https://github.com/orgs/w3c/projects/20/views/8

ora: we have issues to solve before CR...
… preferences for order of attack?

[ suggestions made ... which provide basis for which others? ]

w3c/rdf-concepts#129

<AndyS> First paragraph "an abstract syntax (a data model) "

<AndyS> "1.1 Graph-based Data Model"

niklasl: prefer `abstract syntax`, because `data model` can mean more different things

<AndyS> and of course the title: "Concepts and Abstract Syntax"

gkellogg: it may be difficult to completely remove `data model`, but perhaps we could define `data model` in terms of `abstract syntax`

ora: I find inclusion of `syntax` leads to confusion in some people's minds

<tl> what about "abstract model"?

niklasl: `model` is frequently enough used to mean `interpretation`, leading to more confusion

pchampin: I have always understood `abstract syntax` and `data model` to be synonyms, and changing that at this stage seems problematic

gkellogg: agree these are synonyms, and removing one from the documents at this stage may not be possible

james: they're not synonyms, based on history and elements of their definitions

<AndyS> data model (14) abstract syntax (24 but 6 is title)

james: one reader's opinion

<niklasl> some of the "data model" come from the issue note though ;)

ora: not overly concerned about this, but might be good to have a paragraph we can point people to, going forward

<AndyS> I'm happy with the "abstract syntax (a data model)" in the intro and section 1

gkellogg: people coming to RDF from Property Graphs will wonder what the `RDF data model` *is*, so we shouldn't remove it entirely

<tl> "The abstract syntax is the RDF data model..."

ora: suggests glossary that defines "data model" with "see `abstract syntax`"

james: people wanting to use RDF to `represent their domain` are thinking about `data model`

AndyS: multiple audiences mean we need to cover some things (like this) early, such as in the introduction
… we're running into the problem that there aren't enough words

<Souri> Would it be correct to say that "Abstract syntax for RDF enables the RDF data model." ?

ora: proposes we say something to the effect that we think of `abstract syntax` and `data model` as more-or-less interchangeable, and the term `abstract syntax` is largely used today due to its historical use

<AndyS> https://en.wikipedia.org/wiki/Abstract_syntax ; https://en.wikipedia.org/wiki/Data_model

ACTION: niklasl to take a stab at the paragraph differentiating c at this stage

<gb> Created action #172

drat. "`c`" should have been "`abstract syntax` and `data model`"

w3c/rdf-concepts#143

<gb> Issue 143 "3.7 Replacing Blank Nodes with IRIs" very misleading (by franconi) [ms:CR] [spec:bug]

AndyS: doesn't like this section, not really sure what it's trying to do (now)

ora: franconi and pfps are not here... thoughts from doerthe?

<AndyS> https://www.w3.org/TR/rdf12-concepts/#section-skolemization

doerthe: there is discussion of skolemization elsewhere, so we should keep this, somewhere (perhaps appendix?)

<AndyS> "6. Skolemization" - RDF Semantics - https://www.w3.org/TR/rdf12-semantics/#skolemization

doerthe: have seen this in practice, but relatively rarely

ora: I've seen it in production

ktk: has seen it more in connection with SHACL

tl: doesn't canonicalization depend on skolemization?

gkellogg: dataset canonicalization does not use skolemization. skolemization is dependent on its endpoint, among other things.

doerthe: could try to make the text more acceptable to franconi

ACTION: doerthe to try to revise the text now in appendix b

<gb> Created action #173

w3c/rdf-concepts#163

<gb> Issue 163 Move datatypes definitions in RDF-Schema? (by pchampin) [ms:CR] [spec:editorial]

pchampin: to summarize, proposal was to move datatype definitions from RDF Concepts to RDF Schema, because not required for basic definitions of RDF
… also partially to help clear the path to CR

<james> scanning through the current version of "concepts and abstract syntax", of the 14, three are trivially non-distinguishing, five are in the comment about the issue, one is accurately referring to a "model" and the others should more accurately be "abstract syntax".

gkellogg: needs to be resolved before CR if we're making big changes to RDF Concepts, which this would count as
… it might be too large a refactoring to handle at this stage, with too little benefit

ora: ambivalent. any other thoughts?

niklasl: no good arguments. might be *nice* to move from Concepts to Schema, but not *necessary*

ora: seems more a matter of taste than of technicality. no passionate opinions?

<gb> CLOSED Action 129 write a PR on rdf-concepts for the unstar mapping (on pchampin) due 2024-10-01

tl: agreed

<Zakim> tl, you wanted to say #129

james: volunteers to draft PR replacing `data model` with `abstract syntax` (and vice versa?) to see what happens

<niklasl> Sounds good, so we can see how it reads.

ACTION: james to draft PR replacing `data model` with `abstract syntax` (and vice versa?) to see what happens

<gb> Cannot create action. Validation failed. Maybe james is not a valid user for w3c/rdf-star-wg?

ACTION: lisp to draft PR replacing `data model` with `abstract syntax` (and vice versa?) to see what happens

<gb> Created action #174

ora: returning to question of moving datatype definitions ... let's not?

pchampin: let's close that with no action

[ no objection ]

w3c/rdf-concepts#169

<gb> Issue 169 Relative IRI Reference should bind to the irelative-ref production (by gkellogg) [ms:CR] [spec:editorial]

gkellogg: what do we do with IRIs that are parsed and match the production for a resolved IRI, but aren't the same after running through resolution algorithm?
… such as, an IRI that includes leading dot segments. is that the same as the IRI that results from removing such dot segments?
… thinks not the same

TallTed: they're not the same IRI, for most meanings of "same", but whether this matters is another question

pchampin: according to RFC, dot segments are not allowed in absolute IRIs, so these are pathological anyway

gkellogg: suggests pointing to segments of RFC 3686 & 3687 and saying that behavior with such noncompliant IRIs is undefined

ora: let's pick that up when AndyS is with us

<Zakim> pchampin, you wanted to suggest line 8

pchampin: suggests w3c/rdf-concepts#183

<gb> Issue 183 Normative statement in Security Considerations (by csarven) [ms:CR] [spec:editorial]

w3c/rdf-concepts#183

TallTed: I was going to say the same; there are easy subtitutions

ACTION: pchampin to take a stab at w3c/rdf-concepts#183

<gb> Created action #175

TallTed: e.g. 'ought' instead of 'should', 'can' instead of 'may'

w3c/rdf-concepts#228

<gb> Issue 228 RDF Concepts section ordering (by afs) [ms:CR]

gkellogg: small work to be done to finish this

w3c/rdf-concepts#92

<gb> Issue 92 identity and equality of datatype values (by pfps) [ms:CR] [needs discussion] [spec:editorial]

pchampin: we should wait for pfps, but I would consider that this issue can be closed, we had a PR addressing it
… I'll ping pfps on github to check with him

two issues about IRIs

ora: is w3c/rdf-concepts#169 related to w3c/rdf-n-triples#73 ?

<gb> Issue 73 IRI resolution requirements (by gkellogg) [ms:CR] [needs discussion] [spec:substantive]

<gb> Issue 169 Relative IRI Reference should bind to the irelative-ref production (by gkellogg) [ms:CR] [spec:editorial]

ora: very productive call. just need to see those action points resolved.
… continue in this mode in 2 weeks?

<doerthe> isn't this #144?

<gb> Issue 144 [Editorial] capitalization of "SPARQL string", "SPARQL Query string", and "SPARQL Update string" (by TallTed) [documentation]

gkellogg: question of reifies vs asserts has been open for a year ... we need to resolve it somehow

<doerthe> item 12

tl: these are w3c/rdf-concepts#220 and w3c/rdf-semantics#144

<gb> Pull Request 144 No connection between propositions and facts in model-theoretic semantics (by rat10) [ms:CR] [spec:enhancement]

<gb> Pull Request 220 Annotations on assserted triples are based on operational semantics (by rat10) [ms:CR]

tl: those are prerequisites for me to finish drafting something

[ adjourned ]

Summary of action items

  1. niklasl to take a stab at the paragraph differentiating c at this stage
  2. doerthe to try to revise the text now in appendix b
  3. james to draft PR replacing `data model` with `abstract syntax` (and vice versa?) to see what happens
  4. lisp to draft PR replacing `data model` with `abstract syntax` (and vice versa?) to see what happens
  5. pchampin to take a stab at w3c/rdf-concepts#183
Minutes manually created (not a transcript), formatted by scribe.perl version 244 (Thu Feb 27 01:23:09 2025 UTC).

Diagnostics

Succeeded: i|First paragraph|subtopic: w3c/rdf-concepts#129

Succeeded: s/james/tl

Succeeded: s/james: agreed/tl: agreed

Succeeded: i|related to|subtopic: two issues about IRIs

All speakers: AndyS, doerthe, gkellogg, james, ktk, niklasl, ora, pchampin, TallTed, tl

Active on IRC: AndyS, doerthe, gkellogg, gtw, james, ktk, niklasl, ora, pchampin, Souri, TallTed, tl, william-vw