– DRAFT –
Weekly DXWG

<PWinstanley> https://‌www.w3.org/‌2017/‌dxwg/‌wiki/‌Meetings:Telecon2018.06.12

Admin

PROPOSED: approve minutes of June 5

<alejandra> 0 (was absent)

<PWinstanley> https://‌www.w3.org/‌2018/‌06/‌05-dxwg-minutes

<PWinstanley> +1

<antoine> +1 (and yes I've double-checked them!)

<alejandra> can meeting title, etc be added?

<roba> +1

<annette_g> +1

<Jaroslav_Pullmann> +1

<LarsG> 0 (wasn't there)

Resolved: approve minutes of June 5

open action items

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

127 and 128 are done - to be closed

<phila> close 127

<phila> close action-127

<trackbot> Closed action-127.

<phila> close action-128

<trackbot> Closed action-128.

<riccardoAlbertoni> i think my action can be closed...

<phila> Annotations

roba: minutes from subgroups not getting onto meetings page

<phila> [[Category:Meeting]]

phila: needs the annotations part in the agenda and the category:meeting statement

close action-88

<trackbot> Closed action-88.

requirements for conneg and profiles

PWinstanley: there are 3 re-written requirements

PROPOSED: accept "Requirement: Profiles may provide lists of values to pick from in order to populate data elements [ID46] (5.46)"

+1

<PWinstanley> +1

<phila> +1

<antoine> +1 (unsurprisingly, perhaps)

<roba> what does "provide"; mean?

<LarsG> +1

<Jaroslav_Pullmann> +1

<riccardoAlbertoni> +1

<alejandra> +1

<DaveBrowning> +1

<Jaroslav_Pullmann> that's fine, an example of value restriction by enumeration ..?

<annette_g> 0

<roba> enumeration is very hard to get right and unchanging in practice - referencing a vocabulary is a lot more useful

<PWinstanley> • Requirement: Profiles are "named collections of properties" or metadata terms (if not RDF) [ID41] (5.41) [profile] [Accepted, 5 June] • Requirement: Profiles can have human-readable definitions of terms and input instructions [ID46] (5.46) [profile]

roba: does this mean closed lists and enumerations, or (and?) pointing to a list?

<annette_g> I think this was the pair:
… 1.
… Requirement: Profiles may provide rules governing value validity [ID41] (5.41) [profile]

<annette_g> Requirement: Profiles can have rules for data value validation, including pick lists

phila: problem of enumerations - can change

roba: practice is that inline lists are almost never used

antoine: Europeana does have enumerated lists; the word provide does not specify how the list is made
… whether by pick list or reference
… need to be careful when we enter solution space

<antoine> Profiles can have rules for data value validation, including pick lists [ID46]

antoine: to annette, original requirement included rules for data value validation
… was considered redundant with 5.41

<antoine> Profiles may provide rules governing value validity [ID41] (5.41)

antoine: this should eliminate redundancy with 5.41

<alejandra> Potential rewrite: "Requirement: Profiles may provide lists of values to pick from, whether by enumeration or reference, in order to populate data elements [ID46] (5.46)"

annette_g: do we need to strike out one of the requirements?

antoine: no, don't need to because overlap is eliminated

alejandra: suggests a rewrite to include both pick lists and enumerations

<Zakim> phila, you wanted to refer to ID41

alejandra: should include version of vocabulary or list?

phila: agree with alejandra. This came up in shacl - but wasn't accepted as a use case, because the list might change

<Zakim> LarsG, you wanted to talk about versions of subject headings

<AndreaPerego> @alejandra, I think this should be more a caveat: if you use lists by reference, ensure that a proper versioning mechanism is in place.

LarsG: In library domain, changes are made continuously, so there aren't versions

antoine: warning to alejandra and phila, but the use case does not support this distinction in requirements
… we probably need to extend the use case

roba: agrees with antoine . asking phil: is there anything in W3C that allows us to reference an external vocabulary?

phila: thinking about LarsG comment - you may not be able to validate against it?
… what does valid mean in that case?

<Jaroslav_Pullmann> .. which type of change are we talking about - do the URIs / IDs remain the same?

<alejandra> +1 to validating URI patterns as only option if the vocabulary changes daily

LarsG: validate against URI pattern?

phila: the reality is that people use these and that we cannot guarantee that the validation will hold for an arbitrary future

<Jaroslav_Pullmann> .. ist there any sort of standardized format for "lists" - RDF collections, taxonomies etc.?

AndreaPerego: distinguish between conceptual level and how this is implemented
… conceptually we should support both options - pick list and external list
… yes there is an issue; include a caveat about using lists by reference
… this is what DCAP-AP uses for its shacl definition

<roba> I was going to point out RDF-QB allows binding to ConceptSchemes - and many registration processes (eg W3C) use deprecated status and non-reuse of ids to keep validations valid..

AndreaPerego: shouldn't prevent people from using either option; need to rely on community that maintains a list
… for shacl for dcap-ap considered all options; but couldn't include a full thesaurus in the shacl document

<roba> +1

AndreaPerego: this has more to do with validation - may not be necessary for the guidance document

PWinstanley: as antoine said, the use case is simple; we've dived into something that may need a different use case

alejandra: we have values coming from branches or children of a concept in an ontology; we should mention these issues

<roba> OGC is always referencing Corrdinate Reference systems which is a large externally managed vocabulary (management is a problem of course!)

PWinstanley: two modes: this is the simple case; we need a well-described use case for the more complex situation

Action: alejandra create new use case and requirements for more complex situation

<trackbot> Created ACTION-129 - Create new use case and requirements for more complex situation [on Alejandra Gonzalez Beltran - due 2018-06-19].

<PWinstanley> Requirement: Profiles should be able to indicate what external standards are expected to be applied to the data provided. [ID42, ID43] (5.42, 5.43) [profile]

PWinstanley: kcoyle rewrote 5.42 & 5.43 into two separate requirements

proposed: Requirement: There needs to be a property in the profile where the rules for the descriptive content can be provided. This would apply to the entire profile. [ID42] (5.42)

<annette_g> +1

<alejandra> but do we add the mention of reference and enumeration?

Resolved: accept "Profiles may provide lists of values to pick from in order to populate data elements [ID46] (5.46)"

+1

<antoine> +1

roba: "property" is too solution-oriented

<alejandra> Maybe... "Requirement: There needs to be an association between a profile and the rules for the descriptive content can be provided. This would apply to the entire profile. [ID42] (5.42)"?

<Jaroslav_Pullmann> .. I am wondering where the term standard has gone?

PWinstanley: take out solution space language

<alejandra> New rewrite... "Requirement: There needs to be an association between a profile and the rules for its descriptive content. This would apply to the entire profile. [ID42] (5.42)"?

<alejandra> kcoyle: 'descriptive content' is cultural heritage speech

<alejandra> ... the rules that define how to describe things

<Jaroslav_Pullmann> .. I had sth. like following in mind: Profiles may indicate standards a constituent and its value(s) must conform t

<alejandra> I agree that 'descriptive content' is not too clear

Jaroslav_Pullmann: profiles refer to standards; this should be made explicit

<roba> a profile MAY provide rules and guidance for expression of descriptive metadata

<roba> ?

<alejandra> kcoyle: standard is not really what 'descriptive content' means

<alejandra> ... an organisation's rules are a standard?

<alejandra> ... 'standard' has a lot of baggage

<AndreaPerego> Probably better to use "specification" rather than "standard".

<roba> +1 for specification

<alejandra> +1 to specification

<Zakim> phila, you wanted to wonder whether 'style guide' might be relevant to this?

<Jaroslav_Pullmann> .. an example of how colleagues look at (industrial) standards: http://‌i40.semantic-interoperability.org/‌sto/‌index-en.html

<alejandra> New rewrite... "Requirement: There needs to be an association between a profile and a specification. This would apply to the entire profile. [ID42] (5.42)"?

<alejandra> phila: when reading the requirement I thought about 'style guide'

phila: 'style guide' - ?

<alejandra> kcoyle: yes, but it should include style + content

antoine: how about mentioning standards and style guides?

<LarsG> kcoyle: in the library domain it's generally referred to as "cataloguing rules"

antoine: descriptive content - descriptive metadata?

<roba> "a profile MAY provide rules and guidance for expression of descriptive metadata" ?

Jaroslav_Pullmann: thought it was reference to value standards

annette_g: make this more 'can' than 'must'

<Jaroslav_Pullmann> yes

<AndreaPerego> +1

PROPOSED: accept "Requirement: Profiles should be able to indicate which external standards are expected to be applied to values of individual properties. [ID43] (5.43)"

<annette_g> +1

<riccardoAlbertoni> +1

<PWinstanley> +1

<antoine> +1

<phila> +1

<LarsG> +1

+1

<roba> specifications?

<DaveBrowning> +1

<roba> or even specifications?

<Jaroslav_Pullmann> .. "to be applied" or "is compliant with"?

<annette_g> +1 to specifications

<alejandra> yes, specifications is better

<antoine> happy with roba 's suggestion

AndreaPerego: is this different from lists vs enumeration?

<roba> its broader requirement

<alejandra> I would add a note to indicate that a specification may be a standard, but doesn't need to be

<Jaroslav_Pullmann> .. the spec/standard may even refer to structure..

AndreaPerego: could be broader than a standard

<alejandra> this is listing the external specifications

AndreaPerego: is the previous one sub this one?

PWinstanley: no, they are separate

<phila> I struggle to see an essential difference between the two but won't make a fuss

antoine: not necessary one broader than the other

<annette_g> Does this fix #1? Requirement: There needs to be a way to associate a profile with a specification that would apply to the entire profile. [ID42] (5.42)

alejandra: this says that the profile could have a list of external specifications
… it would be good to cluster them to show what is different between them

Resolved: accept "accept "Requirement: Profiles should be able to indicate which external standards are expected to be applied to values of individual properties. [ID43] (5.43)" with specifications in stead of standards

<PWinstanley> PROPOSED: accept "Requirement: Profiles should be able to indicate which external specifications are expected to be applied to values of individual properties. [ID43] (5.43)"

<roba> +1

<alejandra> +1

<roba> not null ?

<AndreaPerego> Maybe we should say "content" instead of "values".

Resolved: accept "Requirement: Profiles should be able to indicate which external specifications are expected to be applied to values of individual properties. [ID43] (5.43)"

<PWinstanley> Requirement: There needs to be a property in the profile where the rules for the descriptive content can be provided. This would apply to the entire profile. [ID42] (5.42)

Action: kcoyle create github issue for ID42

<trackbot> Created ACTION-130 - Create github issue for id42 [on Karen Coyle - due 2018-06-19].

<Jaroslav_Pullmann> +1

<AndreaPerego> +1

PWinstanley: meet weekly (until we get through this?)

<riccardoAlbertoni> bye, Thanks, good night!

<PWinstanley> bye

<annette_g> :O

<alejandra> thanks, and bye!

<Jaroslav_Pullmann> bye!

<annette_g> bye all!

<roba> bye