– DRAFT –
Dataset Exchange Working Group Teleconference

<riccardoAlbertoni> PROPOSED: approve last meeting minutes https://www.w3.org/2021/02/03-dxwgdcat-minutes

approve last meeting minutes

proposed: Accept minutes of last meeting

<AndreaPerego> +1

<riccardoAlbertoni> PROPOSED: approve last meeting minutes https://www.w3.org/2021/02/03-dxwgdcat-minutes

<riccardoAlbertoni> +1

<DaveBr> +1

0 - not there

<alejandra> 0 - not present

Resolution: Accept minutes of last meeting

<riccardoAlbertoni> https://www.w3.org/2017/dxwg/wiki/Meetings:Telecon2021.02.17

approve agenda

alejandra:

alejandra: can we add the issue of representing software to the agenda?

riccardoAlbertoni: perhaps after the discussion of the two PRs

PR about dataset series

riccardoAlbertoni: we got positive feedback for proposing to handle data series

<riccardoAlbertoni> PR 1292

<alejandra> https://github.com/w3c/dxwg/pull/1292

<riccardoAlbertoni> https://raw.githack.com/w3c/dxwg/dcat-dataseries-issue1272/dcat/index.html#Class:DatasetSeries

<riccardoAlbertoni> https://raw.githack.com/w3c/dxwg/dcat-dataseries-issue1272/dcat/index.html#dataset-series

riccardoAlbertoni: new classes have been added into the normative part. guidelines have been added in the non-normative - how to document datasets that are series
… looking at the normative part there are some issues, mainly editorial. A new section is added

<riccardoAlbertoni> adms:next and adms:prev.

riccardoAlbertoni: the 2 notes are discussing the need for more properties - adms:next & adms:prev

alejandra: one suggestion - can we also have github issues for these as it will make it more visible. Feedback - I've not seen it

<riccardoAlbertoni> #1272

riccardoAlbertoni: the feedback is on the issue #1272
… I would hope that we can resolve this easily, but opening an issue is a good idea

alejandra: the PR is just a draft - is it ready for review?

riccardoAlbertoni: we can discuss in this call and see if it is ready for merging
… there might need to be small amendments

<riccardoAlbertoni> ack

AndreaPerego: I think that is is worth getting more feedback. This new class might have an impact on existing implementations

<alejandra> Checking adms:next (https://www.w3.org/TR/vocab-adms/#adms-next) it refers to the next version of the Asset, I think that what we want is to point to the next dataset in the series instead

AndreaPerego: Perhaps we should analyse the implications. One issue is that there already notions of series within communities of practice, and we might be attempting to define a compromise. E.g. ISO 19115 - the dataseries is a hierarchical collection of datasets
… the parent-child relations are diverse
… we need to consider these types of requirements

alejandra: +1 to AndreaPerego . but also that we need to include data series. I liked the draft pointing to geoDCAT-AP and bringing in what was decided in those communities so that we don't conflict with their representation. But I think it should be in the next PWD
… re: adms:next - it seems to address sequences of versions rather than the next element in a series

riccardoAlbertoni: I propose to leave the PR as a draft and include comments from this meeting before considering merging. Is this ok?

<AndreaPerego> +1

<DaveBr> +1

<alejandra> +1

proposed: continue with discussion on the PR

<AndreaPerego> +1

Resolution: continue with discussion on the PR

PR on versioning

<AndreaPerego> https://github.com/w3c/dxwg/pull/1295

riccardoAlbertoni: AndreaPerego has prepared the PR
… I support equivalence with PAV

<riccardoAlbertoni> https://raw.githack.com/w3c/dxwg/dcat-versioning-v2/dcat/index.html#dataset-versions

AndreaPerego: a second draft of the versioning section. The first had just a couple of paras indicating that we were not going to take a position, but this was challenged and so we are taking one. The survey of the current landscape reviews notions of version types
… and we provide some lightweight guidance
… other sections are focusing on specific types of version - new resource; revision; correction; etc
… these sections cover version information, and also about backward compatibility. Also, we consider lifecycle. Some institutions have a formal review process that needs to be catered for.
… After drafting there was a realisation that this review isn't going to provide guidance. One aspect that people are looking into, needing more guidance, is 'version as revision'. This has been updated in the PR.
… There is now an explicit statement of what we are providing guidance on
… We are proposing using PAV equivalents in the DCAT namespace. This is mainly because PAV is not under good maintenance - so minting in the DCAT namespace is safer

<alejandra> Comment by Stian about PAV: https://github.com/w3c/dxwg/issues/1280#issuecomment-734397742

AndreaPerego: relationships - version chain/hierarchy; replacement; backward compatibility
… Issues for consideration - backward compatibility; version identifiers
… in the new proposal there is another PAV equivalent - dcat:Version
… there is also another section on complimentary approaches to versioning
… the owl version predicate is meant to be used for ontologies, and DC version predicates are too broad in their domain

<alejandra> PWinstanley: versions are used for different things

<alejandra> ... also compability of things, either doing analysis, having multiple datasets and want to be able to see that these ones are compatible with your analysis

<alejandra> ... in the latter case, you often deal with a range

<alejandra> ... has there been any kind of thought about this in the solution?

<alejandra> ... we're going to be looking at automating data processing pipelines

<alejandra> ... we don't want to have a human in the loop

<alejandra> ... compatibility ranges

<alejandra> I think this can be done combining version and dataset series

<alejandra> PWinstanley: automated pipeline

alejandra: the versioning section is looking good - I like the diags etc and I agree we need the PAV equivalents added
… One question - owl:versionInfo is mainly used for ontologies, but perhaps it is generic enough for our use. Why did we change from thinking that this existing pred could be used?

AndreaPerego: it was one of the things I was unsure about, but once the decision had been made to focus on a specific notion of version the consequence was that we needed to pay closer attention to the semantics, and so to be crystal clear the decision was to use a re-minted PAV
… but this is just a proposal

<alejandra> https://github.com/w3c/dxwg/issues/1280#issuecomment-734397742

AndreaPerego: The initial plan was not to create new properties in this space, but after changing direction it made sense to bring in a wider set. There was feedback to support this. But it is still up for discussion

alejandra: I added the link to Stian's comment.
… Perhaps distinguishing datasets from ontologies is a good reason to move, even though formally using owl:versionInfo doesn't mean that the entity is an ontology

riccardoAlbertoni: we could relate the owl:versionInfo to the newly minted pros within DCAT

<riccardoAlbertoni> ack

AndreaPerego: we have another option - using complimentary vocabs . multiple vocabs could be used
… if there is a need to migrate/upgrade, the use of e.g. DC vocabs isn't a problem. However, I don't think dcat:version should be a subprop of owl:version has implications for the interpretation (wrong interp) of PAV

riccardoAlbertoni: I would like a vocab about versioning that is not 'just another vocab about versioning'/
… we should relate to other approaches

AndreaPerego: I understand your concerns riccardoAlbertoni , but we are creating a vocab for versioning things that are in the catalogue, for resources.

+1 to AndreaPerego point about this being for versioning catalogue items
… I'm not saying that we use one or another, they can be used together, but they are not the same things

<alejandra> yes, we should continue the discussion in the PR

AOB

riccardoAlbertoni: no other business
… thanks for a good discussion

bye

<AndreaPerego> bye

<DaveBr> bye

<riccardoAlbertoni> bye