– DRAFT –
JSON-LD WG Meeting

Announcements and Introductions

<pchampin> slava: Slava Tykhonov, head of dynamic intelligence at CODATA

I'm going to talk about Croissant mostly

Presentation from Slava Tykhonov of CODATA

bigbluehat: should we change our schedule due to AC meeting?

pchampin: I'm not going

ivan: me neither

bigbluehat: ok, let's keep our usual schedule

Slideset: https://docs.google.com/presentation/d/1dTCuiGHEeNqGThwbgUqK8oL5rzk8YVXP/edit?slide=id.g3c447b49fa9_0_527#slide=id.g3c447b49fa9_0_527 and archived PDF copy

slava: (presenting slides) CDIF is not really a new standard, more a compilation of existing standards
… Croissant ls widely supported by AI platform
… (presenting more slides)

bigbluehat: we need to move on, but I'm amazed how much ground you are covering.
… I have a JSON-LD related question
… In your Croissant examples, the context is always embedded. Is it by design?

slava: this is by design, pushed by Google. We (CODATA) have a different opinion.
… We would prefer to have contexts by reference, managed with ODRL policies.
… But Google do not like that access to some data can be removed at any time.
… We want more granularity and access control -- e.g. for use-cases such as medical data.

pchampin: it's great to see all these technologies brought together
… Slava may be aware, but the Dataset Exchange WG is being rechartered
… we're adding DDI related things
… and at some point, I suppose this will become part of CDIF

slava: it's great that these are becoming official standards

YAML-LD check-in

bigbluehat: let's skip this one

CBOR-LD PRs & Issues

bigbluehat: we have a handful of PRs, all approved and editorial

https://github.com/w3c/cbor-ld/pull/73

wes-smith: I didn't want to merge cbor-ld#73 before mentioning it here

<gb> Pull Request 73 Apply writing and grammar fixes, remove constraints on registry dictionary format. (by wes-smith)

wes-smith: anyone objecting? it is editorial

[crickets]

Pull Request 76 Prepare First Public Working Draft (by pchampin)

wes-smith: pchampin's PR to add a FPWD snapshot

pchampin: please let's merge it, so that I can move forward with the transition to actuallly publish it

Pull Request 75 add json ld links (by BigBlueHat)

bigbluehat: I'll try to have this done by next week.

Issue 46 Define registration process (by dlongley) [needs discussion]

wes-smith: my view is that registration will look like w3id registration
… people would propose PR
… the requirements for registration would depend on what you would request
… small values are precious, so the requirements on those would be higher (complete specification, proof of implementation)
… bigger values would be first-come first-served
… the current editors of the CBOR-LD spec at the time of the request would be the reviewers

pchampin: there may be a situation where there is no current editor of the CBOR-LD specification
… so we'd need a backup plan

bigbluehat: if it leans on editors for now, that's fine; we can move forward from there
… once the WG closes, we don't want wes-smith to be stuck with this for eternity
… we need a transition within the lifetime of the WG
… there are several questions? is it going to be a W3C registry? a simple github repo?
… what wes-smith described seems subjective
… (smaller values vs. bigger values)

wes-smith: yes, we could try to make it less subjective
… e.g. require a specification from an official SDO

bigbluehat: unlike the DID registry, it is not open season
… what if there are conflicting use of the same value?
… unlike DNS, there is no programmatic enforcement

wes-smith: if someone chose to use CBOR-LD payloads with an entry number that do not comply with that entry, noone will be able to consume it

dlehn: this looks similar to the way CBOR tags registration works

wes-smith: yes, that's the inspiration, and that's where the subjectivity comes
… part of it is specified, part of it is subjective

dlehn: there are technical reasons for doing that, but also resource problems

bigbluehat: what would be the break point between small and big value?

wes-smith: that would be a byte size, as defined by the CBOR format
… I'll turn that into a drafty PR

bigbluehat: we don't have time for anything else
… thanks for giving time to Slava to present; loads of W3C specifications involved, with JSON-LD all the way
… interesting use-cases