W3C

RDF-star WG biweekly focused meeting

12 June 2025

Attendees

Present
AndyS, AZ, doerthe, Dominik_T, gkellogg, gtw, james, ktk, niklasl, olaf, ora, pchampin, pfps, Souri, TallTed, tl, Tpt
Regrets
-
Chair
ora
Scribe
Tpt

Meeting minutes

extract the "basic-encoding" (formerly unstar) algorithm from RDF-Concepts into a Note 1

pchampin: There are some ongoing discussion on the unstar mapping. It is a non-normative part of RDF concepts that takes a significant amount of space. It might make sense to move it out of the spec in a dedicated note. This way it does not block RDF concepts to move into CR

pchampin: The main point is to move it out of RDF concepts to not block RDF concepts moving into a CR

<pfps> It's fine by me to move this to a note.

tl: Having unstar is pretty essential to connect new concepts with the existing ones

gkellogg: If it an informative note, how can it be used by RDF canonicalization? Would they have to do a new version of it that is normative?

<Zakim> pchampin, you wanted to respond re. c14n

pfps: I disagree with tl, I don't see how the basic encoding helps to explain the relation between old and new concepts

pchampin: +1 to pfps, it is important that this mapping does not give the impression it is *the* relation between triple terms and statements

<pfps> the mapping is not in any way saying how triple terms relate to rdf:Statement, at least so far

pchampin: I agree that other specs might want to rely on that and we have a problem

<pfps> as far as I remember there was very little support for making triple terms relate to the RDF reification vocabulary in any way

niklasl: Formally there is no difference between a note and having it being a part of RDF concepts

<pfps> +1 to niklas

<pchampin> niklasl, the difference is that we *could* decide to make it normative after all if it remains in Concepts; this is not an option if we move it to a Note

niklasl: It can't be done as part of RDF schema semantic, we need to do it elsewhere

tl: The semantics of triple terms is nearly indistinguishable with RDF standard reification, we could just merge the two

<niklasl> Not *only* though.

<pfps> the semantics of triple terms are different from the semantics of standard reification - the difference is important

tl: We could consider publishing as a note and in the WG maintainance mode move it to a normative element or integrate it into RDF concepts

ora: A note like this would be more than this "unstar" mapping. It would have some advises to people

ora: for me it makes sense to have it as note

<pchampin> +1 pfps

niklasl: I agree with that, perhaps we need two notes, a compatibility note and one with advices

tl: Reification in RDF 1.0 has an informal semantic.

<niklasl> Define semantics? It is based on RDF semantics, but adds none?

ora: I would like a decision

<pfps> the semantics of standard reification is just the semantics ascribed to properties and types - there is nothing more!

<niklasl> +1 to pfps

ora: We can start as "non normative" and make it normative if there is a reason

AndyS: What the barrier to having an other normative document?

pchampin: Adding a new REC track deliverable might be an heavier process

<tl> pfps: what more does it need?

pchampin: I would like to agree to separate it from RDF concepts and then differ the decision to make it a note or a recommandation

<pfps> what more does what need?

<tl> the definition of the semantics

<pfps> the semantics of what?

<tl> pfps: "the semantics of standard reification is just the semantics ascribed to properties and types - there is nothing more!" that semantics

ora: If we have a note, its scope can be larger than unstar and contain more content like advice. If it is normative, it might make more sense to have it in concepts

<pfps> I don't think that anything more is "needed". One might want to have more, but the current semantics is a (weak) stable point.

pchampin: For me the urgency is to free RDF concepts to go into CR

pchampin: then we can have further discussion on REC vs NOTE and on the document scope

pchampin: I don't want to give the impression that all RDF 1.2 implementations needs to support unstar even if it's only for specific use cases

ora: If I understand what you mean, you want to say "if you want to have this kind of processing, do it this way" and not "to implement RDF you need to be able to do that"

<pchampin> different "bridges": full-basic, basic-1.1

AndyS: If you put it into a best practice document, it would weird to have some normative sections

AndyS: We can decide now to have a separate document (initially a NOTE)

ora: For me the easiest route would be to make it a separate NOTE and then make the decision to make it a REC if we want so.

<ora> PROPOSAL: Separate "unstar" mapping from Concepts, make it a Note (initially).

<pchampin> +1

<pfps> +1

<niklasl> +1

<Dominik_T> +1

<ora> +1

<gkellogg> +1

<doerthe> +1

<tl> +1

<olaf> +1

<Souri> +1

<ktk> +1

<AndyS> +1

<james> +1

<Tpt> +1

<gtw> +1

RESOLUTION: Separate "unstar" mapping from Concepts, make it a Note (initially).

vocabulary to refer to the individual nodes in a reified triple term 2

ACTION: pchampin to extract the "unstar" mapping from RDF-concepts

<gb> Created action #165

<tl> on that topic see comment w3c/rdf-star-wg#130 (comment)

<gb> Issue 130 vocabulary to refer to the individual nodes in a reified triple term (by rat10) [needs discussion]

tl: The idea is one can annotate the reified triple term but one might want to annotate a given triple in a refied triple term

tl: I presented that in the semantic task force last year to know if we could use rdf:subject/predicate/object for that or a new vocabulary rdf:onSubject/Predicate/Object

tl: If the reified triple has for type rdf:Statement then this issue is null

<pfps> the idea is actually extremely complex, if it has any bite, being able to reach into a triple term and attach new information to its pieces is something that is far from anything else that can be done in RDF. Of course, maybe there is no "there" there and this construct has no additional semantics then what is it for, and why should the properties be in the RDF namespace.

pfps: Why having properties for that in the RDF namespace, if you need to do this, use properties in your own namespace

niklasl: The problem is inheriting the notion of identity. The very informative rdf:Statement vocabulary could be used for that

niklasl: Hopefully some of the things you want would be explained in a note. In general I have to agree with pfps, it's not a pattern we could endorse in RDF

niklasl: I would prefer specific complex concepts to have dedicated vocabulary. The prov: ontology is a good example of that

ora: For tl, do you think you can write something like a vocabulary you invent that allows to do that

ora: We can push people to design their own vocabularies for what they want to do and show it can be done

ora: It does not need to be the worry of this WG

<niklasl> I specifically mean the prov:specializationOf property.

tl: I disagree with this prov: vocabulary example, provenance is not really a thing in my example

tl: I would be in favor of using the same vocabulary in all places

<tl> w3c/rdf-star-wg#130 (comment)

<gb> Issue 130 vocabulary to refer to the individual nodes in a reified triple term (by rat10) [needs discussion]

tl: ^ an other comment, the second example there is very clear, using the standard RDF reification vocabulary

tl: we can leave it like that and talk about the triple term vocabulary and see if we actually need specific vocabularies or just reuse the plain rdf:subject/predicate/object vocabulary

<Zakim> pfps, you wanted to say that this is not referring - there is already a way to refer to the subject, predicate, and object of a triple term

<Souri> Wouldn't an "avatar"-based approach work? => :r rdf:reifies <<( :Alice12 :bought :LennyPuppet )>>, <<( :Alice12 :age 12 )>>, <<( :Alice12 :avatarOf :Alice )>>, <<( :LennyPuppet rdf:type :Puppet )>> ... .

pfps: If the subject of the triple term is Alice, we can just refer to Alice directly

<james> exactly

If we cand talk about Alice in the context of age 12, you can use a named graph for that (maybe problem solved?)

niklasl: This might be true but without semantic we don't know

niklasl: Yes, you can use the classical reification vocabulary for that

niklasl: The question you can ask yourself is "would your user accept that"?

<pchampin> niklasl, it would RDFS-entail that the reifier is of type `rdf:Statement`, which may not be desired...

ora: If I understand it correctly, it seems it really pushes the boundaries of RDF. I would like to see how do we do this with the RDF we have

ora: I don't see the WG responding particularly positively, you can use the RDF we have right now to model that. I would see that as a favorable solution

<niklasl> pchampin; Yes, good point, that is an inevitable consequence. Maybe ok:ish for the users, but indeed conceptually "muddy".

Souri: This idea of putting of context with Alice is to introduce Alice12.

<niklasl> What Souri suggests is exactly what I suggested in w3c/rdf-star-wg#130 (comment)

<gb> Issue 130 vocabulary to refer to the individual nodes in a reified triple term (by rat10) [needs discussion]

ora: You can introduce an ontology with the concept of avatar

<niklasl> (So +1 to Souri)

tl: Yes, we can close it then, without much support there is not much point

tl: you can do everything already with the RDF we have

tl: we are looking for easier ways

make progress on test suites to move to CR 3

gkellogg: There is a big MR in RDF/XML that needs multiple approvals

AndyS: Talking tests is a way to talk about "what do we need to move to CR"

AndyS: I would like to confirm there is nothing else to do out of the tests

pchampin: Good point, the other important thing is the wide review, we need to coordinate with the W3C

ora: let's confirm that wide review is before CR

AndyS: Reformulated question: is there anything to do before wide review

gkellogg: We have open PRs for VERSION declaration in NTriples and NQuads

AndyS: Are we aiming at having RDF concepts and NTriples/NQuads CR at the same time

AndyS: If we are aiming for concepts and semantic to go out as a pair, we need NTriples/NQuads out as well

<AndyS> Given needing N-Triples, putting Turtle/TriG into the mix seems natural.

pchampin: I am in favor to have VERSION announcement in NTriples/NQuads as well

ora: It seems there is not much to do

AndyS: Can we minute this statement ^

ora: There is a great deal of interest in RDF 1.2

ora: A question I got is "what does it means for ontology definition"

ora: Any last word?

ora: Thank you for your contributions

AndyS: There is a meeting tomorrow, bumped to 14:30UTC

AndyS: This is a permanent move

<pchampin> m2gbot, link issues with transcript

<m2gbot> comment created: w3c/rdf-star-wg#163 (comment)

<m2gbot> comment created: w3c/rdf-star-wg#130 (comment)

<m2gbot> comment created: w3c/rdf-star-wg#164 (comment)

<ktk> pchampin: ah was I too late?

<ktk> ah no it worked

Summary of action items

  1. pchampin to extract the "unstar" mapping from RDF-concepts

Summary of resolutions

  1. Separate "unstar" mapping from Concepts, make it a Note (initially).
Minutes manually created (not a transcript), formatted by scribe.perl version 244 (Thu Feb 27 01:23:09 2025 UTC).

Diagnostics

Succeeded: s/AndyS: We have open/gkellogg: We have open/

Succeeded: s/prsent+/present+/

Succeeded: s/It is an non-normative/It is a non-normative/

Succeeded: s/singnificant/significant/

Succeeded: s/help to explain/helps to explain/

Succeeded: s/voculary/vocabulary/

Succeeded: s/The semantic of triple term/The semantics of triple terms/

Succeeded: s/indistiguishable/indistinguishable/

Succeeded: s/advices/advice/

Succeeded: s/for me it make sense/for me it makes sense/

Succeeded: s/then what is it for and why should/then what is it for, and why should/

Succeeded: s/provenance in not really/provenance is not really/

Succeeded: s/we can live it like that an talk/we can leave it like that and talk/

Succeeded: s/an talk about/and talk about/

Succeeded: s/there is no much point/there is not much point/

Succeeded: s/Hare we aiming/Are we aiming/

Succeeded: s/bumpt to 14:30UTC/bumped to 14:30UTC/

All speakers: AndyS, gkellogg, niklasl, ora, pchampin, pfps, Souri, tl

Active on IRC: AndyS, AZ, doerthe, Dominik_T, gkellogg, gtw, james, ktk, m2gbot, niklasl, olaf, ora, pchampin, pfps, Souri, TallTed, tl, Tpt