– DRAFT –
DCAT group 2018-07-19

<SimonCox> regrets Lars

<alejandra> +1

+1

minutes of last meeting

<DaveBrowning> +1

<SimonCox> https://‌www.w3.org/‌2018/‌07/‌12-dxwgdcat-minutes

<arminhaller> 0 (missed last week)

<alejandra> +1

+0

alejandra: an issue cannot be in two milestones... issues will need to be moved to next WD if necessary

2PWD

Resolved: approve minutes of last meeting

<SimonCox> https://‌github.com/‌w3c/‌dxwg/‌milestone/‌13

SimonCox: shall we remove old milestones (topic related) now antoine has improved tags

alejandra: fine by me

Action: SimonCox to remove old milestones and check tags are correct for DCAT issues

<trackbot> Sorry, but no Tracker is associated with this channel.

SimonCox: do we make an announcement now or wait till next WD

PWinstanley: outreach is ongoing. we need to check logs, but send a link to the email. ask people to repeat if we get a chance

SimonCox: goal for PWD2 in September

minor issues

SimonCox: noted PRs merged
… range for dcat:mediaType - inconsistent legacy use
… range mediaTypeorExtent

<SimonCox> proposed: accept https://‌github.com/‌w3c/‌dxwg/‌pull/‌304

<SimonCox> +1

<arminhaller> it is relaxing semantics, so shouldn't be an issue

<arminhaller> +1

<DaveBrowning> +1

<alejandra> +1

+1

<Jaroslav_Pullmann> +1

<PWinstanley> +1

Resolved: accept https://‌github.com/‌w3c/‌dxwg/‌pull/‌304 to adjust range of dcat:mediaType

<arminhaller> i think we need to change that!

<SimonCox> currently it is range=rdf:Literal with recommendation to be xsd:decimal

roba: can it have a unit of measure - megabytes probably more natural

alejandra: unit of measure makes sense

SimonCox: currently a literal with recommendation is a decimal

<alejandra> https://‌github.com/‌w3c/‌dxwg/‌issues/‌125

could we relax it - and recommend a uom form,,

<alejandra> The definition is "The size of a distribution in bytes."

SimonCox: will still need to be literal

<alejandra> so there is no room for adding units of measure, unless we do another property

<SimonCox> owl datatypes https://‌www.w3.org/‌TR/‌owl2-quick-reference/#Built-in_Datatypes

Jaroslav_Pullmann: uom implicit in names - non-negative integer

alejandra: for images bytes is not so useful - perhaps a new predicate is required

SimonCox: these may be an unrecognized requirement

Jaroslav_Pullmann: specialised properties :byteSizeMB - or can we used pattern form SSN or something with qualified values

<arminhaller> +1 alejandra's proposal, size and then qualify it with UoM

alejandra: -1 to specialized predicates for different UOM

<Jaroslav_Pullmann> ;o)

<alejandra> https://‌bioportal.bioontology.org/‌ontologies/‌UO

SimonCox: UoM is horrible - everyone needs it but not perfect answer
… many implementations - none is authoritative

<arminhaller> and a proposal for an unqualified UoM: http://‌w3id.org/‌lindt/‌custom_datatypes#>

and they have different information models

Action: alejandra to create a formal issue for uom on bytesize

<trackbot> Sorry, but no Tracker is associated with this channel.

AndreaPerego: no further knowledge of history or intent

catalogues in which dataset is an undifferentiated set of files

<SimonCox> https://‌github.com/‌w3c/‌dxwg/‌issues/‌256

<SimonCox> https://‌github.com/‌w3c/‌dxwg/‌issues/‌253

<SimonCox> see https://‌github.com/‌w3c/‌dxwg/‌issues/‌253#issuecomment-405418358

SimonCox: see discussion - need to think about range constraints

<SimonCox> https://‌github.com/‌w3c/‌dxwg/‌blob/‌gh-pages/‌CONTRIBUTING.md

PWinstanley: FYI contributions text added to github - has SimonCox contacted contributers

SimonCox: contacted all and forwarded responses to PWinstanley - all were happy to accomodate this

alejandra: contribution notes relates mainly to PR

PWinstanley: will discuss with kcoyle

SimonCox: back to example - revised (long complicated explanation - see git history)

<SimonCox> In context of a dcat:Dataset dct:hasPart, isFormatOf, references should point to other dcat:Datasets not Distribution

roba: should we consider qualification of relationships

SimonCox: existing subproperties seem to provide useful granularity
… horse has bolted - rich set

<AndreaPerego> +1

AndreaPerego: looking at PROV for relationships - not clear what pattern - we have relationships between datasets and also many different roles of people
… codelist from ISO 19115 for different roles - cumbersome :-(
… what is the UC

SimonCox: would point to PROV which supports both flavours - shortcut and qualified relationships
… propose to close this with use of specific relations - leave to users to subclass these as needed
… " i.e. recommend more specialized properties"

roba: will follow discussions about dataset and agent relationships

<AndreaPerego> BTW, there's a problem using prov:hadRole broadly - see thread starting here: https://‌lists.w3.org/‌Archives/‌Public/‌public-prov-comments/‌2015Apr/‌0001.html

<PWinstanley> s/boradly/broadly/

<SimonCox> Proposed: accept recommendation to use dct:relation for the general case of undifferentiated collection; and illustrate use of sub-properties of dct:relation for stronger semantics

<AndreaPerego> +1

<SimonCox> through examples

<SimonCox> +1

<Jaroslav_Pullmann> +1

<PWinstanley> +1

<alejandra> +1 (adding guidelines )

<DaveBrowning> +1

+1

Resolved: use dct:relation for unstructured collections

Action: SimonCox - update current PR to reflect agreement here

<trackbot> Sorry, but no Tracker is associated with this channel.