Dataset Exchange Working Group Teleconference
12 June 2017
Karen: any objections?
<Thomas> no objections here
<Jaroslav_Pullmann> no
<roba> +1
Resolved: approved last week minutes
<phila> [NOTUC]
Karen: I've sent an email about sub-groups
… are there any questions?
… specialy about the public face of groups?
… W3C requires our work to be visible.
[No reaction]
Karen: we'll set up a mechanism for sub-groups to report
… in the meanwhile if you have something you want to discuss to the main group, send an email to the main group, or put yourself on the agenda.
… with an idea of how much time you might need.
Karen's mail: http://lists.w3.org/Archives/Public/public-dxwg-wg/2017Jun/0020.html
<roba> email doesnt change agenda :-)
Karen: if we discuss DCAT we are going to discuss AP probably
… so it would make sense to discuss these two deliverables since the beginning.
… is there anyone who would take responsibility for AP?
roba: I'm interested in the architecture aspects
… DCAT as a use case for the patterns.
s/responsiblikty/responsibility
phila: which deliverable?
<phila> Guidance
karen: the guidance one.
phila: it's about the definition of what is meant by AP, and how to share them.
… if you want to publish an AP of everything, here's what you should do.
… it's the least technical of the three
karen: would it contain things that complete DCAT?
phila: charter says that we will not create AP.
phila: in terms of DCAT work, it's about deciding what is in DCAT and what is in an AP.
… the doc may refer to examples of APs published by members of the group.
roba: I agree with the scope in the charter
… not sure conneg of AP and doc can be divorced
<phila> The WG is free to merge those 2 docs if it feels it makes sense to do so
roba: versioning may not go into the core
<AndreaPerego> I'd rather consider them modules/extensions, rather than profiles.
phila: if the WG decided that the two docs are merged, it's ok.
<Jaroslav_Pullmann> yes will try
<Jaroslav_Pullmann> yes plesase
<Jaroslav_Pullmann> I'll try to fix the audio
<phila> antoine: I may be interested in helping but it depends on the timing?
<Zakim> LarsG, you wanted to talk about profile in the context of conneg
<phila> FPWD is due Q1 2018
karen: no timing for the start but we know when it needs to be finished.
<Jaroslav_Pullmann> +1
karen: what is important is to have someone taking care of noting the principles
LarsG: +1 for not divorcing
… we need to have a definition in order to start the work on conneg.
karen: are you taking care?
LarsG: ruben and myself
karen: sounds good
AndreaPerego: I agree we need to agree on what we mean by AP
… DCAT could provide a means for versioning and some other extensions could exist
annette_g: I am interested in working on AP
… is the implementation meant to be for browser developer or publishers of data?
<alejandra> +1 kcoyle
karen: publishers of data I think
Jaroslav_Pullmann: need to understand what an AP is
… to me so far an AP was a sort of extension.
roba: hopefuly AP define content rules
… what to be in data
… and it's not only about publishing
… schema is one aspect, content is another
<Zakim> LarsG, you wanted to answer annette_g 's question about implementation
roba: we shouldn't make a decision about the scope too early
… we need to look at the use cases
LarsG: what ODRL do is rather a profile, not a schema. [??]
… to answer the quesiton on publishers vs browsers. It's not so much for browsers
<phila> antoine: It may not only be be for browsers,
<phila> ... I'm a little puzzled when Jaroslav says profiles aren't extensions - I think that's what they are
<Zakim> phila, you wanted to check folks here
phila: we need to be careful about the word 'browser'
… we should use 'client' or 'user agent'.
… clients requesting info from a server.
… we should be disciplined
… about this
<chile> sure
<chile> Markus Freudenberg
AndreaPerego: in DCAT-AP we wanted to provide guidance on how to use DCAT.
… identify the classes and properties that are mandatory optional etc.
… done by involving parties, asking how they are using DCAT and what is missing.
… for ensuring interoperability
… to be sure we include a minimal set of metadata elements for sharing and discovery
… we need to consider the practical purpose of what an AP is built
Jaroslav_Pullmann: there was a misunderstanding about extensions
… my point is that there's more than this
… e.g. ODRL profiles
<roba> schema qualifiers (cardinality and allowed sub types), extensions and content bindings/rules all seem to be necessary to achieve interoperability
Jaroslav_Pullmann: [gives examples of potential candidates for profiles]
<Luiz_Bonino_DTL> Can profiles also be used to validate metadata entries in registries during submission?
Jaroslav_Pullmann: we should have a shared understanding of granularity/scope
<Zakim> phila, you wanted to talk about additional docs
<phila> A use case and requirement document
<phila> A test suite for content negotiation by application profile
<phila> A primer (subject to the WG’s capacity)
<phila> Subject to its capacity, the working group may choose to develop additional relevant vocabularies in response to community demand.
Jaroslav_Pullmann I got you now!
Jaroslav_Pullmann sorry for the misunderstanding
phila: we're going to need to prove that we have people ready to publish and consume APs
… if we are to develop APs.
… We can create a Primer distinct from the DCAT spec
… If someone wants to create an AP, this can be done in a community group
Jaroslav_Pullmann: is somebody tracking the development of all profiles?
karen: as far as I know, no. We could have a task in the WG
<phila> antoine: The LOV tool may be useful
<phila> ... It shows vocabs being used and many of these cases are in fact profiles
<Zakim> LarsG, you wanted to talk about LOV and profiles
LarsG: I'm surprised
<phila> antoine: Extensions may be mini vocabs, hints etc
<phila> LarsG: Then maybe we could do some work on LOD stats
LarsG: maybe LODstats
karen: I'll post an email about interesting things I have learnt based on Stats
karen: development of profile shows how we can connect things together
… it's going to be difficult to have the discussion at a meeting
<Thomas> Sorry; have to leave now (interesting discussion though). See/hear you later. I'll check the meetings later this week.
karen: it would be great if people with ideas could post them on the list
karen: we should get requirements from our use cases
… I believe our set is not complete yet
<RiccardoAlbertoni> +1 to decide requirements without thinking to much about the deliverables divisions
karen: I've put 3 fairly concrete cases on agenda
karen: first case: ID6 - DCAT Distribution to describe web services
<alejandra> https://www.w3.org/2017/dxwg/wiki/Use_Case_Working_Space#DCAT_Distribution_to_describe_web_services
karen: [reads case]
<phila> phila has changed the topic to: Dataset Exchange Weekly Call
phila: this UC came at the workshop.
<phila> Open APIs
phila: it doesn't talk about API
… Open APIs is the way people talk about APIs
… Web Services apply to anything now, but it used to have a specific tech definition
… we should be disciplined about using it when we mean 'API'
… this UC is about API
… this UC is something we have to address
<alejandra> +1 to phila
karen: the editors could make the change
roba: agree
<annette_g> +1 to phila
<jrvosse> +1 to phila
roba: other aspect: generally speaking, you need to answer the structure of a dataset
… it overlaps with other UCs
… experiments using void and datacube suggest that reuse of vocabularies is indicated
<alejandra> This use case seems to be a duplicate (or at least it overlaps) with: https://www.w3.org/2017/dxwg/wiki/Use_Case_Working_Space#Modeling_service-based_data_access
<AndreaPerego> This is what I reported to the mailing list: http://lists.w3.org/Archives/Public/public-dxwg-wg/2017Jun/0023.html
<alejandra> Yes, AndreaPerego - I was going to mention that now :-)
roba: we can't put much in DCAT core about the description of this
<LarsG> +1 to roba
@roba: have I captured right what you meant?
chile: we should have a specific look at the class Distribution
… there are 2-3 properties that are relevant, we could increase it
karen: can you create new UC or add info to this one?
AndreaPerego: there are several UC that cover the same topic
… we should look at whether they can be merged.
… people expect to find data at the Distribution URL but they find an API
… they can't do anything about this
… We split the requirements in separate issues
… we realized we need a way to say that a Distribution is not linked to a download, but to an API
… and we need to describe the info about the parameters of the API.
… some of this could be included in DCAT
… it used to have things about this, which were dropped when it became a standard
… This is an issue for user agents not only for humans
… DCAT core could have some of it, while details about APIs could end in an AP
karen: could you add this to the UCs?
AndreaPerego: they are there
<phila> 5 April 2012 version of DCAT included dcat:WebService
AndreaPerego: but I can seek integration
<roba> ID7, ID18, ID21, ID22 at least in part
fanieli: we need relation between UCs
… can we also be more specific when we talk about 'users'?
karen: good point
karen: data consumers could also be a program
fanieli: about the template people could add more information
<alejandra> +1 fanieli
karen: can you post this request on the list?
fanieli: ok!
<LarsG> s/karn/karen/
<Jaroslav_Pullmann> sorry my audio has died
<Jaroslav_Pullmann> I'll try later on
<Zakim> LarsG, you wanted to ask if relation between DCAT and VoID is in scope
LarsG: relation between DCAT and VoID?
… especially wrt what is a dataset in the two
… is it in scope?
phila: yes
<chile> we should also make clear the relation to concepts of DataCube/Prov-O and others
<roba> +1 key guidance issue - stop people getting confused .e.g thinking they need to choose ...
Jaroslav: I agree with Andrea
… this is the same meta UC we had with tagging
… suggestion for editors: have a generic UC covering dynamic aspect to distribution for datasets
… and then have a more detailed look at how datasets are distributed
… [gives examples of APIs]
… and then we could have a profile about the specifics of dynamic distribution
karen: we could have a UC that explains this
<Zakim> phila, you wanted to make a suggestion to fanieli
karen: are we ready to vote on this UC?
<chile> +1
karen: even though it may need some revision?
<roba> can we vote on the Use Case, but asking UC editors to propose a generalisation?
karen: or does it need revision first?
… I'm going to propose that the group accepts ID6
PROPOSED: accept ID6
<AndreaPerego> +1
<annette_g> +1
<PWinstanley> +1
<Luiz_Bonino_DTL> +1
<LarsG> +1
<nandana> +1
+1
<alejandra> +1
<Jaroslav_Pullmann> +1, rrequires further editing
<jrvosse> +1
<DavidBrowning> +1
<RiccardoAlbertoni> +1
<chile> +1
<achille_zappa> 0
<mbruemmer> +1
<Caroline_> +1
<phila> +1 modulo re-wording to avoid term 'web servce'
<alejandra> I agree it requires editing and merging, but it is important as a use case
<alejandra> +1 phila
roba: we all agree we want that scope
… but the proposal should be that the content is in scope, but we ask a generalization
karen: see comments on votes
… this is a vote on the concept not exact words
<phila> PROPOSED: That the concepts expressed in ID6 are included in the UCR as seen fit by the editors
I think the comment was for a gneeralization
<roba> +1
<phila> +1
<annette_g> PROPOSED: That the concepts expressed in ID6 are included in the UCR with minor editing by the editors
karen: we can discuss the revision on the list
+1
<Jaroslav_Pullmann> +1
<annette_g> -1
<annette_g> -1
<chile> +1
<annette_g> okay, going to type instead
<phila> Oxford F2F
<roba> Can we just make sure that we discuss wording of proposal _before_ voting - saves time in the end i think
<roba> (in general - not this one)
<annette_g> If you look at Phil's proposal without knowing what was said, it reads as "the editors can choose whether to use that use case or noe"
karen: it seems we have to do it on the list now
… which will help take into account the other comments made today
roba: we shoudl take into account discussion
… I didn't hear a call for discussion
<AndreaPerego> Sorry, I have to leave. Byer
Resolved: That the concepts expressed in ID6 are included in the UCR with minor editing by the editors
<AndreaPerego> s-Byer-Bye-
<Jaroslav_Pullmann> +1
karen: what we have voted on is that we agree that this is a requirement but that the UC needs to be written in terms of APIs not Web Services and pull info from other UCs
<annette_g> I agree with that
<roba> +1