W3C

– DRAFT –
Data Shapes WG

23 March 2026

Attendees

Present
ajnelson-nist, AndyS, bergos, edmond, HolgerK, ieben, mgberg, nicholascar, Robert, simonstey, TallTed, YoucTagh
Regrets
-
Chair
-
Scribe
simonstey

Meeting minutes

phase1 deliverables

elianaP: haven't seen any updates on the review issue on github
… or have I missed anything?

nicholascar: my plan is that by the end of this month, we should have all deliverables ticking over
… we really need to get some sort of summary
… we should be hearing from the TF exactly how close they are to putting up something for wide-review

elianaP: there was a plan that we should at least have an outline and some form of commitment to what would be in that outline
… so that we know what we can expect

open pull requests

elianaP: ajnelson-nist, have you managed to address some of them?

ajnelson-nist: slowly but surely
… I'll have to check which I addressed last week

core, sparql, nodeexpr TF

HolgerK: went through all tickets and marked those that I believe should be closed as suggesting to be closed, some marked as risk
… core, not everything is covered but with sparql and nodeexpr every ticket has an owner
… the latter two are basically finished

profiling TF

nicholascar: there are 22 open issues for profiling, have to go through all of them and check which are already addressed and which arent
… assign them to owners, propose to close them if fitting, etc.

ajnelson-nist: most of our work last week was on issue 209 (?)
… issue 623 was also discussed
… waiting on input from the rules TF

<ajnelson-nist> w3c/data-shapes#576

<ajnelson-nist> w3c/data-shapes#830

<nicholascar> Discussion about conformsTo in w3c/data-shapes#825

nicholascar: dcterms:conformsTo is used by other stuff in W3C, but in only has weak semantics
… not an object property, you can't/shouldn't really use it in property chain axioms or similar

nicholascar: And so I've put the link in the chat, but there's a draft PR that talks about: do we need a property which is some SHACL-namespace conformsTo? It's the statement that you make when something validates
… this thing validates according to that thing, therefore this conforms to that. So it's the statement you can make, and then you could use it elsewhere.
… So we want to do it so we can validate something and then say: yes, that thing does conform to this thing, we can assert it. I think that is of general interest. We don't see a million use cases outside of the Profiling work yet for this, but I think if we end up doing what we think Profiling will do, we will have people asserting this more

often, and so we want to make sure that the statement reads correctly in relation to SHACL Core.
… coming back to your question eliana, we are on track

UI

<edmond> w3c/data-shapes#332

edmond: renewed interest in the above issue
… UI vs non-UI property shapes

<edmond> w3c/data-shapes#835

edmond: and we were wondering whether this issue should be more generalized so one can categorize shapes for different purposes
… #835 lists what we plan on having in our document
… somethings need to be written down yet

AndyS: wanted to ask ajnelson-nist what he was expecting from rules

ajnelson-nist: remark from HolgerK, it's tied so the sh:entailment property
… HolgerK suggested that there is a way that rules might satisfy custom entailment regimes
… not actually seeing how that's differnt from my original ask like "can something in SHACL at large" handle custom entailment regimes

HolgerK: originally, I thought this was the use case of sub-propertyof, e.g. if someone wants to activate subpropertyof for inferencing
… without bringing everything of OWL

ajnelson-nist: The subproperty-of use case is the one that is loudest in issue #219, but the bigger issue of #219 is custom entailment regimes. Like, if I want OWL entailment, but I don't want everything about OWL entailment. then... how could I say that those in an entailment regime that's completely custom, but I don't want a SHACL engine to fail

if it doesn't report that it's not satisfying this entailment regime?
… But say that... if a SHACL engine brings in a graph and asserts an entailment triple for all OWL inferencing ( OWL DL, let's say) then, if the entailment and the SHACL engine knows it can do that entailment

nicholascar: I was hoping this being something that could be done by SHACL profiling

[nicholascar and ajnelson-nist discussing how SHACL profiling could maybe used for this]

AndyS: so I'm not seeing any specific rules requirements yet
… workflows not being part of the rules scope
… I would ask how you're going to know you're pointing to a rule set without dereferencing it and finding out what its MIME type is.

HolgerK: if sh:entailment points at the URI I mean, we could at least say in the core spec
… how those should be interpreted

AndyS: That's the flexibility provided here. We provide you with the hammer. It's for you to decide whether that's a screw or a crinkle cut nail to apply that to.

rules TF

<AndyS> w3c/data-shapes#829

<nicholascar> Apologies everyone, I have to leave early - now! See the profiling people on Thursday and the rest of you next Monday.

AndyS: we have a meeting this week, the main topic being the issue above
… particularly about bnodes appearing in the head template

<HolgerK> I also need to leave sorry

AndyS: there is a proposal in the issue outlining how this could be addressed
… but this will have implications on how things are used.
… it needs to be more controlled and you can just write construct style blank nodes. there needs to be some rule ordering so that you know that each rule gets only executed once

<bergos> I also need to leave sorry

AndyS: document has sections, (some of which are empty)

AndyS: assignment / bnodes issue is probably the one big vertical issue that we have to have a solution for
… If we don't have that, then we have a real problem. It's not a feature you can drop. It's fairly essential in the core of the system.
… Please do the reading. It is a form of stratification around assignment. so a rule with assignment or template instantiation can be separated out and run at the appropriate moment and run only once which means you will get one set of blank nodes from the head template, and not an indeterminate number
… The attraction of the approach, as Oggy described it, is a classic approach to this class of problem. We have discussed some more ad hoc local solutions to the problem, but I'm somewhat cautious about getting too clever in this area unless we can find proofs that it is safe to do so

compact syntax

elianaP: no update from the editors

6th april

elianaP: only 2 people would attend, for the rest it's public holiday
… will skip it

caribou: EU is changing to summertime next week

[timezone discussions]

Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Succeeded: s/to/of/

Maybe present: caribou, elianaP

All speakers: ajnelson-nist, AndyS, caribou, edmond, elianaP, HolgerK, nicholascar

Active on IRC: ajnelson-nist, AndyS, bergos, edmond, elianaP, HolgerK, ieben, mgberg, nicholascar, Robert, simonstey, TallTed, YoucTagh