– DRAFT –
DXWG DCAT subgroup teleconference

<SimonCox> Are you all joining webex?

<SimonCox> Need to turn on your audio!

<SimonCox> I can hear you stijn

<SimonCox> do you not hear me?

confirm agenda

<SimonCox> https://‌www.w3.org/‌2017/‌dxwg/‌wiki/‌Meetings:DCAT-Telecon2018.02.21

<SimonCox> will you join audio riccardoAlbertoni ?

<NicholasCar> Apologies from me folks: kids need readying for school now

<SimonCox> https://‌www.w3.org/‌2017/‌dxwg/‌wiki/‌Meetings:DCAT-Telecon2018.02.21

SimonCox: going through agenda items
… any better suggestions for agenda items?

approve minutes from last meeting

SimonCox: agenda confirmed

<SimonCox> https://‌www.w3.org/‌2018/‌02/‌14-dxwgdcat-minutes

<SimonCox> headings 5. and 6. need to be switched

SimonCox: there was some problem with topic headings, needed to switch headings 5 and 6
… discussion was largely about constraint axioms
… made a request to Dave to change those

AndreaPerego: yes, AFAIK Dave has to do it

<Stijn_Goedertier_AIV> +1

<AndreaPerego> +1

<SimonCox> otherwise +1

<DaveBrowning> +1

<riccardoAlbertoni> +1

SimonCox: approve the minutes modulo resolving those issues?

+1

+1 by PWinstanley

Resolved: last meeting minutes approved, modulo switching headings 5 and 6 (Dave)

fpwd dcat 1.x

<SimonCox> here is the artefact - https://‌w3c.github.io/‌dxwg/‌dcat/

SimonCox: reminder that the artifact we are working on is DCAT 1.1

<NicholasCar> prexent+

SimonCox: in the text of the editor draft is 99% taken from DCAT 1.0
… you can see a bunch of notes and issues
… the issues are linked to github issues (I think in all cases)
… in one case I put a discussion about what the issue is

SimonCox: given that we haven't made a substantial decision on issues, so we will make very obvious the areas where changes have been considered
… the ones we have dealt with so far as the easy ones
… it'd be good if more people would take a look at that
… modify the document in the standard way: create a branch, commit changes to the branch, send a pull request

<riccardoAlbertoni> +1 to the approach followed for preparing the FPWD, It makes perfectly sense to me

DaveBrowning: simple procedural question, when we merge, does the document gets rebuilt automatically?

SimonCox: yes

SimonCox: named editors have been working on a workflow on how the document will be managed in github
… in the latest merged PR (pull request), I made the changes, Alejandra reviewed them and DaveBrowning merged them

SimonCox: we have a transparent process with proper review in place
… requirement to issue this FPWD by 1st quarter of this year
… so we've got a month and a half
… it'd be great to resolve a few more issues, but it is not a problem
… most important things are transparency and communication

packaging dcat

SimonCox: sharing screen to describe packaging
… a number of discussions about axiomatization and changes to DCAT
… domain and ranges
… looking ath PR #94

https://‌github.com/‌w3c/‌dxwg/‌pull/‌94

<PWinstanley_> https://‌www.w3.org/‌TR/‌vocab-ssn/#Modularization

... group Spatial Data on the Web WG and SSN vocabulary

... discussed modularization

... each block in the diagram represents a separate RDF graph

... describing aspects of the SSN vocabulary

... in the centre SOSA module

... SOSA lists the key classes and properties in the SSN vocabulary

... without axiomatization

... no domain and ranges

... it does include annotation properties

... goal was to provide a vocabulary usable in a schema.org context, very general context

... schema.org annotations are hints, no entailment

... it was deemed as useful

... for the web community in general, not just SW community

... SSN module: added the axiomatization, imports SOSA and adds subclass relationships and constraints

... a user can load the SOSA graph with week axiomatization

... and load the SSN graph if they are interested in the full axiomatization

... other relevant feature is PROV alignment module

... importing SOSA and PROV-O, providing a mapping between both

... this approach addresses the issue of allowing profiles

... not directly and issue of this team but it would help with the connection with the profiles

... we can split the different issues into separate RDF graphs

PWinstanley_: could you explain the impact of the context of the type of information that you've been dealing in the design?
… it seems we've got two completely different types of data and I was wondering if it has to do with the nature of the data

SimonCox: some sensor data is streamed but most of the use cases were not about high volume stuff
… internet of things
… now showing Fig 23
… expectation is that there will be reasoning involved
… I don't think that the use case is what you are imagining
… now describing property alignment
… my proposal for DCAT / PROV alignment
… is to have a new file (prov.ttl) with the alignment
… it would be normative for people that want a PROV alignment
… reusing the modularization package of SSN
… the approach

alejandra: I agree with the approach and I think that following a similar modularization as SSN is very appropriate for the requirements we need to deal with
… I was cautious to merge the PR until we had this discussion

PWinstanley_: it looks like we are promoting bringing in something new
… my concern is how adequately can be tested for real
… is there any history of this new approach being rolled out in real life

SimonCox: I can point to what we've done in SSN
… a number of these modules are labelled as non-normative
… only a small number of modules are normative
… the non-normative are FYIs
… but we believe they are informative
… the non-normative haven't been tested much

AndreaPerego: I was wondering if we can also consider another option for the alignment between DCAT and PROV
… normalisation process of the data cube
… subclasses can be added to a graph containing records written in DCAT by using SPARQL queries
… instead of having a separate definition using RDF

<Zakim> SimonCox, you wanted to ask if SPARQL queries is just an alternative way to document?

AndreaPerego: when I have it in the triple store, the SPARQL can be used

SimonCox: could we have both approaches?

AndreaPerego: yes
… they are not mutually exclusive

a-

AndreaPerego: using SPARQL queries can be done pretty easily
… depending on the uses, we need other representations (OWL, SHACL, etc)

riccardoAlbertoni: some concerns related to best practices in linked data
… term coming from a vocabulary, returned by HTTP
… it is ok to split the definitions of DCAT in different RDF files

<AndreaPerego> About the normalisation algorithm I mentioned, used in RDF Data Cube: https://‌www.w3.org/‌TR/‌vocab-data-cube/#normalize-algorithm

riccardoAlbertoni: but at the same time we should try to have a system having the full semantics

<Zakim> SimonCox, you wanted to point out that the meta-issue is the idea of modularizing additional axioms or rules into different graphs/files/artefacts and to ask riccardoAlbertoni if

SimonCox: about assembly set of information
… OWL provides 'imports'
… is that enough?

riccardoAlbertoni: we need to discuss the different RDF files

<SimonCox> alejandra: the two complementary solutions may help resolve riccardoAlbertoni problem

<SimonCox> ... create the mapping on-the-fly

<SimonCox> ... by SPARQL. While the OWL axiomatization relies on owl:imports

<SimonCox> ... OWL axiomatization allows use-as required

<SimonCox> ... need to go back and look at UCR.

<AndreaPerego> +1 to alejandra to go back to UCRs

<SimonCox> ... be careful that everything is tested

SimonCox: what I'd like to get agreement on
… is to deal with the different issues by modularizing in to multiple RDF files
… we are unclear about impact of modifying dcat.ttl
… so if we have additional RDF files
… we don't need to make a normative commitment at this point

PWinstanley_: I'm not against it, but modularisation brings some complexity
… e.g. versioning
… reminds me of unit testing microservices, etc

SimonCox: yes, managing complexity by creating a choreography complexity is a risk

SimonCox: we can work in independent files as a way of discussion
… and we can decide to merge them as a later point

riccardoAlbertoni: I would second the solution proposed by AndreaPerego for mapping into third-party voc, and modularization to manage with different level of formalization but I need to analyse this issue in more depth

SimonCox: I will create a specific issue around the proposition I was just making
… as there is no consensus
… we can continue the discussion

AndreaPerego: PWinstanley_ raised a concerned that we didn't get much feedback on UCR
… I was wondering whether to bring something about the work done to the RDA community in Berlin
… too late to organise a specific session
… but maybe we can do sth

<SimonCox> NicholasCar and SimonCox are represneting the 'australian' community, in particular around ANDS.

<SimonCox> AndreaPerego, NicholasCar and SimonCox need to trigger some feedback from RDA community - opportunity at P11 plenary in Berlin

NicholasCar: discussion about interaction with other RDA groups
… this group has come up a few times

<SimonCox> AndreaPerego? can you coordinate us at RDA - I'll back you.

NicholasCar: expecting to have more feedback
… another RDA group: storage definition group

<AndreaPerego> I'll give it a try, SimonCox [scared]

NicholasCar: interested in distributions - I presented the work on distributions to them

<riccardoAlbertoni> bye

<Stijn_Goedertier_AIV> thanks all bye bye

<DaveBrowning> Bye

<NicholasCar> bye