– DRAFT –
DXWG plenary

I can scribe

scribenick annette_g

admin

proposed: accept last week's minutes

<PWinstanley> https://‌www.w3.org/‌2019/‌06/‌11-dxwg-minutes

<kcoyle> +1

<TomB> +1

<AndreaPerego> +1

<PWinstanley> +1

+1

<riccardoAlbertoni> +0 ( i was not there)

<roba> +1

Resolved: accept last week's minutes

PWinstanley: open action items

<PWinstanley> https://‌www.w3.org/‌2017/‌dxwg/‌track/‌actions/‌open

PWinstanley: 337 for Rob, about contacting Phillipe
… there is an issue about the system that Karen and I are following up on

roba: that's actually mislabeled, a conneg scope issue

PWinstanley: the email thread wasn't getting caught up in the aggregator

how to handle profiles across docs

PWinstanley: we've had a to-ing and fro-ing of emails over the past day or so. Who wants to start the conversation?

TomB: my comments are linked to in the agenda, but when I looked at conneg and at the IETF draft, I see a reasonable definition of profile as a doc that expresses the semantic constraints of other docs, then I note in DCAT a reference to what the dublin core community calls application profiles.

<AndreaPerego> ^^ https://‌lists.w3.org/‌Archives/‌Public/‌public-dxwg-wg/‌2019Jun/‌0035.html

PWinstanley: that's a note in a green box, not the main definition

TomB: it also offers another definition. My point is that I don't see any dependence on the way it's defined in profiles. The definitions in DCAT and conneg shouldn't cite the profiles vocabulary, especially since it's no longer on rec track.

PWinstanley: the point about the green box and the fact that other docs have different definitions has been pointed out by others, too (ODRL).

<alejandra> the prof vocabulary would be a formalisation of the profiles, but the definition is independent of the vocabulary

roba: There's no hard dependency on the profiles vocabulary. Neither does there need to be. We have a definition of profile that's narrower than the general community because we have a context. Every community has a definition that's dependent on their own context. The question is whether they are consistent. We have a definition that's general enough to work for everyone. I'm open to improving the definition to make it easier to interpret. I don't th[CUT]

have a broken definition. We have a requirement to have a general definition that supports conneg.

PWinstanley: asks Rob if he's satisfied with the one Tom recommended

roba: I don't think we have a problem with the intent, the question is whether we can improve it to make it easier to interpret.

kcoyle: I've been trying to contact Lars and Herbert and Ruben to find out what their needs are for conneg. Apparently Herbert and Ruben are going to write the next version of conneg, so I'm waiting to find out the relationship between our and the IETF documents.

PWinstanley: I want to make sure that we have links to all the appropriate emails and examples, important correspondence. Besides it being in the aggregator, can we pop the URL into the wiki main page?

kcoyle: there was also the page of profile definitions

<Zakim> TomB, you wanted to say that the definition in PROF seems more specific than profiles in the community

PWinstanley: I've included that as well. I want to have all these things close at hand for reference.

TomB: because I support the idea of having a broad definition of profile, but I think the profile vocab has a narrower definition, because it says it has named constraints. I don't think one can reduce it to sets of constraints. I vote for a more inclusive definition.

<roba> tom - please bring that to the profiles vocabulary sub group we will reconvene, we will address that in detail - its pretty much all we need to do at this stage :-)

PWinstanley: so you're saying the profile vocab has a profile of a profile definition ;)

PWinstanley: we need to have somewhere an appropriate, broad enough definition that recognizes that we can contstrain that for specific uses.

ncar: I would have a problem with broadening the definition. I've interpreted the definition of profiles to be expandable. We could extend it.

PWinstanley: but this is exactly what's done within profiles, for example an application profile is an augmentation.

ncar: yes, I'm saying that's okay.

roba: I sent an email around, we need to reconvene a time for revisiting updates to the profiles vocabulary. There's one functional requirement that drives all this stuff. When you extend something, it still needs to be conformant. What you can't do in a profile is say "we think author is related to the type of pet someone has." Let's take this to the profiles ontology discussion.

PWinstanley: the challenge is that DCAT is wandering off saying that it's taking a Dublin Core approach, and we have conneg which is slightly different, and we need to bring about harmony between our documents.

<roba> ncar sent the email btw.

PWinstanley: in previous meetings, we were thinking about having a sprint or something like that, and Antoine is working on something.

<roba> there is no problem IMHO with the DCAT concept of profiles, and its not a hard dependency.

kcoyle: My concerns here are managerial. We are getting close to there with DCAT, not quite as far with conneg, so having these docs at different stages, but somehow the definition has to be in place by the time DCAT is ready for CR.

PWinstanley: it's a very tiny fragment to remove from DCAT that little mention of the Dublin Core idea. We need to be able to get some sort of harmony before we head into CR. We have to think what we would lose by removing the green box.

<riccardoAlbertoni> +1 to karen

kcoyle: I'm not convince we actually need to have it consistent.

kcoyle: all along, we've had these three deliverables, but we've found they have a fair amount of independence. DCAT doesn't need to define profiles unless the DCAT folks want to go into that. Conneg needs its definition. I think the conneg definition is very different from the DCAT one.

PWinstanley: perhaps there's some merit in defusing potential criticism by adding this. If we get to a point where we start defining things, we can say that there is a certain amount of fluidity in definition.

kcoyle: the people who need each of the docs are not the same people.

PWinstanley: we want to handle the inheritance correctly

kcoyle: conneg can do that.

<Zakim> TomB, you wanted to gently push back on the idea that we need to harmonize a notion of profile - agreeing with Karen on this

TomB: I think DCAT really is an application profile in the DC tradition. I think conneg needs to have a more inclusive things, because it's looking at things that are potentially quite different.

<alejandra> DCAT mentions profiles in two places: https://‌w3c.github.io/‌dxwg/‌dcat/#conformance & https://‌w3c.github.io/‌dxwg/‌dcat/#profiles

alejandra: I tend to agree with what Karen and Tom said. It's true the audiences are different. But I think that, if we could, it would be nice to have some alignment. In the DCAT doc, there are two places where we mention profiles.
… where we talk about constraints, we need to change that.

<alejandra> https://‌w3c.github.io/‌dxwg/‌profiles/#dcat-ap

alejandra: maybe we shouldn't have such a precise definition. I think we can manage that fairly easily.

<TomB> +1 annette_g

ncar: we've seen a little bit of this before in ODRL. It has a section on profiles, which is a very narrow description, very prescriptive. The discussion with them regarding profiles vocabulary shows that they are okay with the way it's defined in the profile vocabulary. They are okay with a broader definition. It's a matter of judgment how far we have to go.

roba: I keep hearing the concept that there's some dependency. It's fairly clear in the document that there is no hard dependency. DCAT is a more general use case. It already takes the alignment into account. I'm not sure that the problem really is a problem. It seems a dangerous distraction to revisit it.

<riccardoAlbertoni> In fact we have the qualifier DCAT for the dcat profiles ;)

<TomB> +1 AndreaPerego to somehow qualify the notion of profile

AndreaPerego: if we think it's not worth harmonizing, if it's okay to use different ways depending on the deliverable, I think we can't always use the word profile. I expect deliverables coming out of the same working group are using the same definition for the same word.

riccardoAlbertoni: in DCAT we are addressing a more general definition. I'm not sure it's important to be consistent.

PWinstanley: what is our path forward now? we have two sections of the agenda, one of which is DCAT and one is how we move forward with conneg. There's also an item about the profile vocabulary. How do people want to move forward with this? Also Antoine has something he may want to bring to the group in the next week

<roba> the use of profile in DCAT (and any other community) should take into account its a specific case of a more general concept. We are already aligned via Use Case so its only DCAT references that need to be "not inconsistent"

kcoyle: I would like to see a proposal for timelines for any work we're going to do.

PWinstanley: let's look at the conneg work first and where we are with the public working draft.

kcoyle: I forgot to mention that we did an action to start a google doc to talk about profiles. It was an action on Rob.

PWinstanley: some of it depends on how the conneg definition work is going to proceed.

<ncar> Apologies, I have to leave: children to get ready for school

AndreaPerego: clarification, to be sure. I think DCAT is out of this, there's nothing to be done. What we do with the other publications doesn't affect DCAT. If we change the labeling, we can just change it in DCAT.

<riccardoAlbertoni> +n to andrea

<TomB> +1 to Andrea

<alejandra> I think it is OK to make DCAT self-contained

<kcoyle> +1 from me, too. DCAT is on its own

<roba> https://‌docs.google.com/‌document/‌d/‌10i9oSb548T3EpK0aPFDhBNR8ycy7QFthiJgPx-pdi0Q/‌edit#

roba: I believe they're aligned anyway. The document is there (above)

roba: we'll address this as comments arise in the working groups.

PWinstanley: I wanted to look at the timeline for conneg for 3PWD.

roba: Nick and I spoke yesterday about how to revitalize it. We will meet with or without Lars.
… There's nothing major, other than the issue of what's a profile, what's a specification.

PWinstanley: this is where it would be helpful to have a draft definition with text and debate in the google doc.

roba: The profiles vocab will not be a subset.

PWinstanley: will you, by next meeting, be able to offer a timeline for 3PWD?

roba: I think that's a reasonable ask, we'll aim to do that.

PWinstanley: Do you want to say anything about DCAT, Alejandra?
… it's only been out for 3 weeks. I made a presentation to the privacy group, they said they would come back with something. I also put out a W3C blog post.

<alejandra> we should all contact more people to get a wider review - I've started doing that

me too

alejandra: I think we just need to push people to contact their networks so we can get more feedback.

AndreaPerego: this is just to reiterate the request to the group for evidence of implementation. We need that to move down the rec track. The more we get the better.

<alejandra> +1 to AndreaPerego

PWinstanley: anything else?

let's meet again in a week.

<riccardoAlbertoni> bye - thanks a lot

<alejandra> thanks, bye all!