– DRAFT –
DCAT team meeting 2018-05-24

<roba> * dang - trying to find password..

<riccardoAlbertoni> hi all!

confirm agenda

SimonCox: proposal at the end of the discussion to close the issue provisionally
… moving on with some issues, considering that all solutions are provisional
… I propose to continue the discussion on DCAT service catalogs
… I chose 3 issues almost in order of complexity

<SimonCox> alejandra: link to ttl in FPWD - needs to be to a dated version that corresponds to dcat.ttl at time of approval

<roba> +1

<DaveBrowning> +1

<SimonCox> +1

<riccardoAlbertoni> +1

<Stijn_Goedertier_AIV> +1

approve minutes of last meeting

<SimonCox> https://‌www.w3.org/‌2018/‌05/‌17-dxwgdcat-minutes

<SimonCox> +1

<roba> +1

<Stijn_Goedertier_AIV> +1

about the ttl in FPWD, I'll continue the discussion via email

<riccardoAlbertoni> +1

+1

<DaveBrowning> +1

Resolved: approve minutes of last meeting

Expanded scope of dcat:Catalog

related to issue: https://‌github.com/‌w3c/‌dxwg/‌issues/‌172

<SimonCox> example from Andrea and SimonCox https://‌github.com/‌w3c/‌dxwg/‌issues/‌237

the discussions on the model are in the wiki here: https://‌github.com/‌w3c/‌dxwg/‌wiki/‌Cataloguing-data-services

<SimonCox> https://‌github.com/‌w3c/‌dxwg/‌issues/‌237#issuecomment-390098689

SimonCox: in the example, what is new is that the :MyCatalog uses dcat:service to link to a service
… let's ignore the URI for now
… dcterms:type has two values from the inspire code list
… and there is a link to a dcat:endpoint
… that would be a new addition in DCAT to satisfy this requirement
… the rest is insignificant

<SimonCox> https://‌rawgit.com/‌w3c/‌dxwg/‌dcat-service-simon/‌dcat/‌index.html#vocabulary-overview

SimonCox: probably the biggest issue is having DiscoveryService as a subclass of Catalog and DataService

roba: 3 things, first the bigger picture
… comment by Maxk description of a profile as a resource
… question about extending dcat:Resource
… dct:conformsTo part of a DataService
… but also for Dataset
… so it probably belongs to the superclass
… Resource

<AndreaPerego> I would be in favour of considering a profile as one of the resources that can go into a catalogue - in particular, into an adms:AssetRepository.

roba: every Resource potentially has a requirement for that
… dcat:distribution property may need a super property

roba: csw in the example is actually a skos:Collection

<Zakim> SimonCox, you wanted to comment that subclasses shown are not exhaustive and to say that sub-property relationships not shown

roba: I can probably fix that, would you like me to explicitly declare it as a dct:Standard

SimonCox: subclassing is not exhaustive
… the diagram should be read as RDF not UML
… some of the classes that we put before are things we haven't discussed fully yet (in gray)
… set of subclasses of Resource is not exhaustive
… superclass of Distribution is not yet defined
… the subproperties relationships are not shown
… I'm interested in knowing if anything is broken

<riccardoAlbertoni> +1 to SimonCox ( about reading UML under the open assumption though i would consider to change the notation or put a big warning about it)

roba: there is one error on dct:conformsTo

<AndreaPerego> +1 to SimonCox about adding sub-classes/properties statements later, and when really needed (we need to be careful not to add dependencies having negative impact on not foreseen use cases)

<Zakim> alejandra, you wanted to ask about Distribution and DataServiceDistribution

<AndreaPerego> Not clear to me what the superclass of "distribution" is meant for.

<Zakim> AndreaPerego, you wanted to ask roba to explain the point about skos:Collection

alejandra: relationship between Distribution and DataDistributionService needs to be clarified in my view

alejandra: what is their difference and relationship with Dataset

<SimonCox> dcat:Distribution is intended to be 'representation' of a dataset, like a file

AndreaPerego: I'd like to discuss further what roba mentioned about skos:Collection
… but it can be at a later stage

<roba> http://‌www.opengis.net/‌def/‌serviceType/‌ogc/‌csw is currently resolving to a thing that is a skos:Collection - I am actually maintaining this - I can change it so that...

AndreaPerego: we should look at what other communities are doing
… e.g OAI-PMH

<roba> http://‌www.opengis.net/‌def/‌serviceType/‌ogc/‌csw is a skos:Concept and a dct:Standard and http://‌www.opengis.net/‌def/‌serviceType/‌ogc/‌csw/ is the "collection of versions"

AndreaPerego: we don't have evidence to define precisely the scope of the catalogue

<roba> i.e. i can fix the data to make the example correct :-0

AndreaPerego: we should refer to existing standards for what the catalogue contains

<SimonCox> alejandra: wants to think about the relationships between DataDistributionService, Distribution, Dataset

AndreaPerego: I don't understand the point of the superclass of Distribution

<Zakim> DaveBrowning, you wanted to ask what limits there are on catalogue-able dcat:Resources...

DaveBrowning: what is the scope of those things that can be catalogued
… what standards should be, do we have a boundary
… we are calling it Data Catalogue
… so what do we include data-related

<roba> Noting the example of https://‌joinup.ec.europa.eu/‌release/‌dcat-ap-v11

DaveBrowning: this could come up on e-publication

roba: the link above showed a profile as a dataset

<DaveBrowning> +1 to squinting....

SimonCox: if one would consider adding profile in the diagram, would it be a subclass of Dataset or Resource?

roba: it is a Resource

roba: a guidelines document is not the same as a Distribution but it is a Resource

SimonCox: this is a core model, what is not stated can be added

roba: there seems to be an equivalent to Distribution as a pattern

<Zakim> AndreaPerego, you wanted to comment on DaveBrowning 's point

AndreaPerego: about what to include into the catalogue, it is related to the work of ADMS

<SimonCox> yes to roba - dct:conformsTo could be moved up to dcat:Resource

AndreaPerego: it was designed due to the need of implementing catalogues of ReusableAssets

<SimonCox> (so that is on the record ;-)

AndreaPerego: ontologies, data models, data schemas, etc
… it was specifically designed for that
… ADMS includes what was considered relevant for documenting models
… profiles is in scope as an asset for ADMS
… when ADMS was contributed in W3C, the gov linked data group, ended up considering ADMS as a specialisation of DCAT

<roba> so we can make profile a subclass of an ADMS class and transitively have it as a subclass?

AndreaPerego: I see loads of correspondences between profiles and their definition in SHACL and its relationship of ADMS

<SimonCox> +1 to considering Profiles (and standards) in scope ... following the patterns already tested in ADMS!

AndreaPerego: we could consider the inclusion of a profile within a DCAT catalogue
… or a more specific catalogue, which is more in scope of ADMS
… the implementation resource description is a type of distribution
… I don't see the need of having a superclass of Distribution

<roba> +1 as a type of distribution - happy if no subclassing needed in dcat core

riccardoAlbertoni: I agree about including the profiles in the catalogues
… I wonder if we should refer to the profile description ontology that Nick and roba have proposed

<Zakim> SimonCox, you wanted to commend the ADMS designers for getting there first! and to reopen the point from alejandra

SimonCox: thanks AndreaPerego for reminder about ADMS
… we should check ADMS details

SimonCox: going back to the point that alejandra made on the relationship between DataDistributionService, Distribution and Dataset
… pattern of DataDistributionService often tightly bound to datasets that they serve
… that's why the link to the abstract Dataset
… Distributions are usuall at the end of the chain
… inverse properties maybe? although discussions were against them

roba: I'm happy about considering ADMS and not clutter the discussion now and move forward
… not worry much about the profiles specifics at this point

AndreaPerego: relation between Distribution and Service
… in the diagrams Simon presented and considering how DCAT is used, there is an indirect relationship between the service and the distribution
… we had discussed and it was brought up again about having a property pointing to the service description
… in DCAT-AP people point from the Distribution to this service description document
… there is not necessarily a direct relationship

SimonCox: we are trying to acknowledge the requirements from what people have done
… but also recognised that there have been some hacks

<SimonCox> alejandra: DataDistributionService should have a direct relationship to Distributions as well as Datasets

SimonCox: we used to conflate Distributions and DistributionServices
… a single DistributionService could deliver multiple Distributions

AndreaPerego: relation between service and distribution is tricky
… service can offer multiple datasets and the same distribution in different formats

isn't that a different distribution?

AndreaPerego: it is not easy to map a Distribution to a Service

<roba> +1

SimonCox: should I put a proposal and accept it provisionally to move forward?

<AndreaPerego> +1

<DaveBrowning> +1

<SimonCox> Proposed: Adopt solution shown in https://‌raw.githubusercontent.com/‌w3c/‌dxwg/‌dcat-service-simon/‌dcat/‌UML/‌DCAT-summary.png (not including greyed elements) as the provisional solution to an enlarged catalog scope, to allow us to move on and consider other DCAT matters for a while. dct:conformsTo moved up to dcat:Resource N.B. The model may be revisited later.

<AndreaPerego> +1

<Stijn_Goedertier_AIV> +1

<DaveBrowning> +1 to the proposal

+1

<riccardoAlbertoni> +1

<SimonCox> +1

alejandra: are we then closing issues about this?

SimonCox: I think so, with the provision that issues can be reopened

ack

alejandra: we need to work out the defintions

<roba> +1

Resolved: accepting the proposal modulo fix of dct:conforms and definitions

<SimonCox> definitions will be refined and resolved on the lists and GitHub

<riccardoAlbertoni> ack

<riccardoAlbertoni> ack

<Zakim> AndreaPerego, you wanted to make a suggestion

AndreaPerego: it would be good to have the link to the example

<riccardoAlbertoni> +1 to AndreaPerego

<SimonCox> also name of dcat:endPoint needs resolving

AndreaPerego: so if we close issues and there are things pending, let's open specific issues

+1 to AndreaPerego

Resolved: accepting the proposal module fix of dct:conforms and definitions