– DRAFT –
Dataset Exchange Working Group Teleconference

proposed: accept minutes last meeting

<azaroth> +1

<PWinstanley> https://‌www.w3.org/‌2018/‌08/‌28-dxwg-minutes

<dsr> +1

+1

<DaveBrowning> +1

<SimonCox> +1

<alejandra> +1

<roba> +1

<PWinstanley> +1

<annette_g> +1

Resolved: accept minutes last meeting

Checking Open Action items

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

close: action-188

discussed actions 86 & 163 - no change to either yet

close action-188

<trackbot> Closed action-188.

close action-110

<trackbot> Closed action-110.

report on DCAT WG

DaveBrowning reports the subgroup is close to the 2PWD

report on profile guidance WG

ncar reported a series of changes had been made to the Guidance document

<PWinstanley> https://‌w3c.github.io/‌dxwg/‌ucr/#ProfilesRequirements

how profileDesc fits into overall WG activities

<PWinstanley> https://‌www.w3.org/‌2017/‌dxwg/‌wiki/‌ProfileRoundup

PWinstanley: we have some use cases, we have a'roundup' documents with different proposals

PWinstanley: we need structure of Guidance to be put out for wide discussion so the group as a whole can decide on work, e.g. whether profileDesc is a Rec, Note etc.

DaveBrowning: we should ask the group as a whole via email

PWinstanley: there is much work to be done if profileDesc is to become a Rec

PWinstanley: profileDesc would need to be supported by the work currently in GitHub & ncar has bullet pointed work in docs to indicate this & kcoyle has summarised work in wiki pages

PWinstanley: people seem hesitant to discuss document structure with people still wanting another look

PWinstanley: I think we can make headway today (i.e. not delaying)

PWinstanley: our charter says we can work on profiling but we need to make decisions about how to proceed with specifics regarding proposals

PWinstanley: profileDesc proposal needs to indicate the gap it is filling

ncar: there is a need for profileDesc - his opinion

AndreaPerego: for a Rec, we need 2 independent implementations, can we do this?

AndreaPerego: risk is if we identify gap and propose solutions, will we be able to implement in time for WG?

<roba> +1

AndreaPerego: we should go for a Note and then see if we can move to a Rec if it looks like we can meet requirements

<Zakim> azaroth, you wanted to discuss non editorial requirements of Rec such as testing

azaroth: Recs need independent implementations and a test suite testing specific features

azaroth: if we go for a Note to a Rec, we need to do this fast in order to complete process in time of WG

PWinstanley: is this (azaroth's comments) about Guidance overall or profileDesc?

azaroth: specifically profileDesc, a technical part of Guidance

PWinstanley: are you saying the rest of Guidance would be a Rec?

azaroth: I'm not certain about the process of Best Practices rather than tech docs

<roba> we dont have enough input to justify best practices about guidance

annette_g: I support profileDesc as a Note or Rec to allow us to point to the formal description (lack of which is holding us back)

<roba> +1

annette_g: it would be strange for Guidance as a Rec but profileDesc as a Note

<azaroth> I believe that a TR can't normatively refer to a Note, as that would backdoor informal documents as providing requirements

<azaroth> But can refer to a Note informatively

<alejandra> is the shacl python validator open source, ncar ?

ncar: my projects are building an implementation of profileDesc and will likely need a test suite

SimonCox: it's not necessary to have Recs refering to only other Recs since they can refer to all sorts of thing

roba: annette_g summarised his thoughts regarding the requirement to document/describe formally profiles

roba: go for a Note and see if we get to Rec - much time already lost

roba: no assumption yet that Guidance would require use of profileDesc, only that it may be a way

roba: noone on this call has proposed profileDesc alternatives, need seems assumed

kcoyle: analysis of requirements (for profile guidance) and which are mey by profileDesc

kcoyle: profileDesc seems to describe environment in which profile exists but not the profile itself

<roba> +1 we need to be careful about claims - and we can prioritise that triage of requirements in the guidance group

annette_g: should we include in profileDesc a more robust description of what a profile is?

PWinstanley: elaborate?

annette_g: languages can be use, anything missing in existing schema languages...

PWinstanley: we talk about machine readable and RDF but what about diagrams and had crafted

roba: we choose a formalism first and W3C uses ontologies

roba: profileDesc specifically includes a definition of 'profile' which will be updated

roba: profileDesc fills gap note catered for by constraint languages (SHACL, ShEx etc) which do not talk about relationships between objects relevant to profiling

PWinstanley: have you, roba, written a gap analysis for profileDesc vis a vis constraint languages

roba: a discussion would be reasonable

PWinstanley: we need this gap analysis in the doc

roba: we can include a comparison table in the doc

alejandra: we need formalism, Note/Rec can be decided later

alejandra: seems no alternative to profileDesc

ncar: an action item from this meeting to generate a gap analysis for the Guidance doc would be good

<roba> should we map requirements to profiledesc at this point then?

PWinstanley: this can be done quickly and would give us a clear steer for profileDesc arguments

Action: ncar to draft a profileDesc gap analysis for the guidance document

<trackbot> Created ACTION-210 - Draft a profiledesc gap analysis for the guidance document [on Nicholas Car - due 2018-09-18].

<roba> - we dont have counter arguments yet - only process cycles - so where shoiuld we address them - issues would seem logical

PWinstanley: a few people have indicated writeup of profileDesc as a Note

roba: issues on requirements seems to be the way to go since there are no counter proposals

PWinstanley: comparison of Dublin Core profiles needs to be included in gap analysis

PWinstanley: at least a comparison with Singapore Framework is required

roba: read and agreed with Singapore Framework, should refere to it more

<SimonCox> +1 to needing an implementable version of profile descriptions! (Singapore Framework does not provide an implementation)

https://‌github.com/‌CSIRO-enviro-informatics/‌csiro-epub-dcap

ncar: I have done a technical analysis/review of profileDesc v. Singapore Framework and put up an example of profile described using this in the link above

Action: ncar to construct profileDesc as a Note

<trackbot> Created ACTION-211 - Construct profiledesc as a note [on Nicholas Car - due 2018-09-18].

proposal: present profileDesc as a Note

<azaroth> +1

<AndreaPerego> +1

<annette_g> doesn't it have to say PROPOSED?

kcoyle: GitHub Issue 323 lists vocabularies that are Notes

<alejandra> https://‌github.com/‌w3c/‌dxwg/‌issues/‌323

proposed: present profileDesc as a Note

<annette_g> +1

+1

<kcoyle> +1

<roba> +1

<PWinstanley> +1

<SimonCox> +1

<DaveBrowning> +1

<azaroth> +1

<alejandra> +1

<AndreaPerego> +1

Resolved: present profileDesc as a Note

PWinstanley: what are the requirements from W3C for Guidance document (non-technical) as a Rec?

Conneg group report

ncar: we have positioned the IETF doc v W3C work

ncar: we have included requirements for conneg in the doc

<azaroth> Thanks all, got to run.

<alejandra> thanks all, and bye

<annette_g> bye all

<AndreaPerego> Thanks, bye bye!