– DRAFT –
Dataset Exchange Working Group Teleconference

admin

<kcoyle> https://‌www.w3.org/‌2018/‌01/‌30-dxwg-minutes

approve minutes of last meeting

+1

<kcoyle> PROPOSED: approve last week's minutes

<roba> +1

<Jaroslav_Pullmann> +1

<Ixchel> +1

Resolved: approve last week's minutes

<annette_g> +1

UCR comments

Need to understand and respond to comments

... UCR editors are traffic cops for this

roba: sent proposal for how to respond to list
… no discussion yet
… two weeks ago

Ixchel: was that my email?

roba: can't find email now - was it sent?
… comments from 'Anna' - possible specializations about differnet kinds of profile
… metadata profile vs. profile
… unclear at this stage if that is needed

<roba> https://‌lists.w3.org/‌Archives/‌Public/‌public-dxwg-comments/‌2018Jan/‌0001.html

roba: use-cases from Norway

<annette_g> +1 to simplifying to a single concept

roba: no different unless there is functional difference in behaviours
… looks like main issues are already covered in existing use-cases
… responsibility of vocabulary providers to package any complex collections

Jaroslav_Pullmann: Norway stated which ucs are important to them and offer to supply user stories
… Ixchel has elaborated answer to this

Ixchel: onus is on the contributors to read what we have and comment on missed requirements
… are there any deadlines on submitting comments on UCR

?

Jaroslav_Pullmann: no deadline

roba: email failed to send
… sent now

Jaroslav_Pullmann: should we put a timeframe for responding to UCR comments

roba: no deadline on UCR comments and responses

Action: Ixchel to respond to Ostein

<trackbot> Created ACTION-79 - Respond to ostein [on Ixchel Faniel - due 2018-02-13].

<Jaroslav_Pullmann> https://‌lists.w3.org/‌Archives/‌Public/‌public-dxwg-comments/‌2018Jan/‌0000.html

Jaroslav_Pullmann: comment from Chris Little
… requirement for 'update' relation
… delete and overwrite

Jaroslav_Pullmann: sees problems with replacement relation, inconsistent with datacite

SimonCox: replacement relations are needed in some scenarios, its not our job to police that

roba: dynamically updated data does not 'replace' earlier data, it is a newer 'best guess'
… different applciations will have different relationships

AndreaPerego: many different scenarios on this use case, can't apply just one vocabulary of relationships
… dataset persistence will be a policy issue on the data-provider side
… various scenarios and possible labeling solutions
… also link to Data on the Web best practices
… cannot prescribe solutions for all cases, though can describe good practices

DaveBrowning: newer 'best guess' scenario should be supported in a profile, and documented in best practices
… too much variation for core

<roba> +1 - informal examples - make sure there is at least guidance

Jaroslav_Pullmann: shoudl we document examples of common cases? maybe a new UC for 'replacement?

DaveBrowning: two distinct cases - 1. new version which replaces earlier 2. newer best guess for dataset previously published

<NicholasCar> Aren't these qualified wasDerivedFroms?

Jaroslav_Pullmann: datacite relation 'new version of'

I don't think so NicholasCar

<Makx> dct:isVersionOf?

annette_g: two versioning issues here -
… version of dataset recorded in catalogue
… version of profile
… dataset case is highly variable, not possible to prescribe versioning patterns generally

<AndreaPerego> +1

kcoyle: who will reply to Chris Little?
… please run this through the dxwg list first!

definition of 'Profile'

roba: had attempt starting from ISO 10000 definition of profile, relaxing need to be based on ISO artefacts
… still thinks it is good, but lots of push back from the group :-(

kcoyle: where is definition text?

<annette_g> 2/4/18 1:07 pm

<annette_g> "A named set of constraints on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. This definitions allows for the set of constraints may be empty, and specifications may be profiles themselves, so that all statements about conformance t

<annette_g> o a specification may be made using the single concept of a profile" 

<roba> https://‌www.w3.org/‌2017/‌dxwg/‌wiki/‌ProfileContext

<annette_g> … which I rather like

annette_g: is "this definition ..." part of the text, or commentary.

roba: commentary

<roba> "A set of one or more base standards, and, where applicable, the identification of chosen classes, conforming subsets, options and parameters of those base standards necessary to accomplish a particular function." noting that the scope of "base standards" is relaxed to any standard that is identified by a canonical IRI, and thus can be identified by software agents. Profiles may act as "base standards" for more specialised profiles.

AndreaPerego: mixed discussion: 1. profile for conneg 2. profile for metadata
… which?

roba: UCR says we need common definition of profile, conneg should be consistent with this
… ruben is concerned about data model describing profile

AndreaPerego: terminological problem - conneg-by-profile allows you to point to information complementary to immediate information
… metadata records in XML which are either ISO 19139 or DCAT - which schema?
… conneg-by-profile must complement media-type
… same term used different for content profile

<Makx> type 2, profile for metadata is mostly useful for validation, e.g. expressed as SHACL file

annette_g: 'A named set ...' focuses on the set of constraints, and is better

<annette_g> +1 to Rob

roba: the two applications of 'profile' should be consistent, big problem if they are not

<Makx> +1 to roba

AndreaPerego: 'DCAT' itself is a 'profile' in this context - content model that complements serialization

<annette_g> DCAT as profile of DCAT is just the trivial case.

AndreaPerego: profile of metadata is also a profile, similar to data

PROPOSED: use this as the definition of 'profile' - "A named set of constraints on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. "

<roba> A named set of constraints on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. This definitions allows for the set of constraints may be empty, and specifications may be profiles themselves, so that all statements about conformance t [08:38] <annette_g> o a spec

<AndreaPerego> Maybe we should say: "A named set of constraints, possibly empty, [...]"

<roba> o a specification may be made using the single concept of a profile"

<roba> +1 to Andreas's clarification

annette_g: antoine also had comments -

<Makx> if a 'named set of constraints' is empty, doesn't that mean an empty set, so nothing at all?

<annette_g> let's vote!

<Makx> so, is the NULL set a profile?

<annette_g> It is a waste of time, though.

<Makx> I would delete the 'possibly empty'

PROPOSED: "A named set of constraints, possibly empty, on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. "

<AndreaPerego> Then, e.g., DCAT is not a profile.

"A named set of constraints on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. "

... + guidance

<Makx> Last one, without the empty, OK with me

roba: in the wild, some existing specifications could be merely profiles

<roba> +1

<annette_g> +1

annette_g: + roba DCAT profiles may be narrower

+1

<NicholasCar> +1

<DaveBrowning> +1

<Makx> what are we voting on?

<AndreaPerego> 0

(we can always revisit)

<Makx> what was the proposal?

<Jaroslav_Pullmann> +1

<Makx> the second one after the PROPOSED?

Proposed: to adopt this definition of 'Profile' - "A named set of constraints on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. "

<Makx> +1

<AndreaPerego> Maybe "constraints or restrictions".

Resolved: ns, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. "

Resolved: to adopt this definition of 'Profile' - "A named set of constraints on one or more identified base specifications, including the identification of any implementing subclasses of datatypes, semantic interpretations, vocabularies, options and parameters of those base specifications necessary to accomplish a particular function. "

<NicholasCar> Cheers all

<AndreaPerego> Thanks, bye!

<roba> bye

<annette_g> bye!

<Jaroslav_Pullmann> bye!

<Makx> bye