Data on the Web Best Practices Working Group Teleconference

13 Nov 2015


See also: IRC log


ericstephan, yaso, newtoncalegari, Eric_Kauz, riccardoAlbertoni, hadley, phila, PWinstanley, laufer, jerdeb, BernadetteLoscio, BartvanLeeuwen, Caroline, peter


<yaso> Topic for #dwbp is: Agenda: https://www.w3.org/2013/dwbp/wiki/Meetings:Telecon20151113

<trackbot> Date: 13 November 2015

PhilA, webex doesn't seem to have the time zones right

We had this problem with the TAG call too (PhilA)

<phila> It's not that, hadleybeeman, it's that our meeeting booking was cacnelled

<phila> we have had trolls


Okay.: (



<yaso> scribe?

<yaso> Hi Laufer

<laufer> hi Yaso

<yaso> scribe: hadleybeeman

<yaso> http://www.w3.org/2013/meeting/dwbp/2015-11-06

PROPOSED: approve last week's minutes?

<yaso> +1


<riccardoAlbertoni> +0 ( i was not present)

<ericstephan> +1

<BernadetteLoscio> +1

<phila> PROPOSED: Accept last week's minutes http://www.w3.org/2015/11/06-dwbp-minutes

<laufer> +1

<yaso> +1

<riccardoAlbertoni> +0 ( i was not present)

<phila> RESOLUTION: Accept last week's minutes http://www.w3.org/2015/11/06-dwbp-minutes

<newton> +1

<BernadetteLoscio> +1

Updates from BP document

yaso: Bernadette or Newton?

<newton> Berna

BernadetteLoscio: We are still working on the updates of the document
... We are working on the examples. And reviewing some Bps.
... I was reviewing the BP on data vocabularies, and opened some issues and are discussing them with Antoine
... I hope we can get consensus and some examples for the end of next week.
... For the Data Access section: Newton and I were discussing examples for this

<newton> The ISSUE-213

<PWinstanley_> hadleybeeman: is it your typing that's coming over the lines? can you mute pls

BernadetteLoscio: And also the BP about use of REST for APIs. Newton just opened an issue about it
... I also sent a message about the use of a specific vocab (the GTFS) to use in examples on transport data.
... I got some answers, and am considering using this. Does the group agree?
... Or are there other suggestions? Mark sent some.
... GTFS is easy to work with.

<phila> +1 to using GTFS in examples

BernadetteLoscio: This is just an example, not to say you SHOULD use this vocabulary
... Phil, have you talked with Christophe re the Data Preservation section?

<yaso> +1 to GTFS at examples

phila: Re GTFS: as an example, that's fine. (The G used to stand for "Google"; now it's "General".) Almost everyone uses it.
... Stations, bus stops, timetables, etc.
... As long as it's just in the examples and we aren't saying "you must use this"; that's fine.
... I did contact Christophe, who responded with a pull request with examples.
... I haven't yet asked him to look at the running example.
... I haven't seen what he's written yet (in that PR), but if the example doesn't fit that, I'll see if I can edit it.

BernadetteLoscio: Okay, I'll look at his PR.
... and I'll reply on email.
... So we are working on the doc, reviewing the BPs. There is still stuff to do. This week I'll send emails asking for feedback about specific stuff.
... From the whole group.
... Soon we need to prepare the doc to publish the next draft, so it's good to get feedback now.

<newton> Title: Use Content Negotiation when providing the same resource with different formats

newton: Re Conneg: I wrote a draft on this and have discussed with Bernadette
... I've changed the title

<yaso> link, newton

<yaso> ?

newton: With this title, this BP could be basian or not
... I don't have a link yet, but I will put the text on my GitHub and send to the group
... It's better to discuss it later, with more content.

<newton> ISSUE-213

<trackbot> ISSUE-213 -- Changes on BP about REST APIs -- raised

<trackbot> http://www.w3.org/2013/dwbp/track/issues/213

yaso: And the BP for REST APIs?

BernadetteLoscio: As Newton said on the issue, it's more because the idea of us saying you should use REST APIs as an approach to implementation — instead of saying this on the title of the BP
... So the title could be more generic, like "use APIs", and then on the approach implementation section
... you could say that one option would be to use RESTful APIs
... I don't think it's in the scope of the group to say how to implement a RESTful API

<phila> Not sure I agree with that but I'll need to look at it more carefully, i.e. I have an idea we should be firmer in recommending REST

BernadetteLoscio: If you're going to use an API, you should use a RESTful API, but say this is an example.

+1 to phil from me

yaso: I agree with Phil. Maybe we should talk more?

BernadetteLoscio: We just started the discussion.

yaso: I agree that we don't need to mention the REST word. But I don't know if we don't have to mention the methods related to hypermedia and stuffy
... Maybe we should carry on the discussion on email?
... Newton or Bernadette, could you send this around?

BernadetteLoscio: It's on the issue (213)

<BernadetteLoscio> thanks Hadley!

newton: We could recommend some links on it
... But I'm not sure we should get more specific

<Zakim> phila, you wanted to ask about next publication date

phila: Newton, BernadetteLoscio, when do you think we can have a new publication? (ideally before Christmas)

BernadetteLoscio: Yes

phila: End of this month?

BernadetteLoscio: Yes

Phila: put the open issues in the document, like this one.

BernadetteLoscio: We are working on it.

phila: I'm trying to encourage you to publish something even though there is more work to do. We need another snapshot out to get feedback.
... We are headed toward Candidate Rec stage in 2 months time.

<yaso> Data Usage Vocabulary

BernadetteLoscio: Okay

updates on data usage vocabulary

<ericstephan> https://www.w3.org/2013/dwbp/wiki/images/c/ce/Duv-citations-research.pdf

ericstephan: I put this link on the agenda Update on Dataset Usage https://www.w3.org/2013/dwbp/wiki/images/c/ce/Duv-citations-research.pdf
... Sorry, it's a slow download
... I'll wait just a couple minutes while it's downloading (for all of you) — and talk about the Data usage part of the vocabulary
... BernadetteLoscio and I have resolved this AM we need the working group's input
... To do that, we have to put issues into the document so you can see where we're struggling.
... And give us your ideas.
... There is another part where external folks are working in the climate community (NASA and universities) have a tool called Tool Match, which seems up our alley
... While it's focused on geospatial, it may give us some ideas on how to direct our efforts on the DUV
... Next part is the presentation. Last 2 weeks, I appreciate everyone bearing with me on Data citations.
... It's a challenging problem, how to get the citations model right.
... I was grappling with the issue that there are a lot of different solutions, challenges, ontologies, vocabularies...
... Most people think about the Info model. [slide 1]
... I'm highlighting the MLA, APA and Chicago Manual styles, the three main ones.
... I've been using the Purdue Online Writing Lab as a resource (logo on slide), because I didn't just want to study about mental models for any type of citation—
... We care about citations for data on the Web.
... [Slide 2]
... Re constructing knowledge models... [Minecraft slide]
... You have lots of tools with different emphases. I've been looking for the best solution for us.
... [Slide 3]
... Is there some level of synchronisation possible, to go from the information model to the knowledge model?
... You've got a consumer who cites, on the Web, a dataset from a journal article.
... We have to have a link to that, or we'll only be building knowledge models and not reaching out to the greater community
... We've got a publisher who creates a metadata citation for data on the Web, or
... citing a technical document associated with the data.
... [next slide]
... Re digital libraries... I've found there is a core electronic document data model.
... There are different styles that extend that model
... Around that, you have the vocabularies. SPAR ontologies, Dublin Core, etc.
... Beyond the vocabularies, you have the Force11 community, and the other non-semweb vocabs
... You have StructureID. Bib-JSON
... Because we're looking at open data on the Web and not just semantic data, that's useful.
... I know we're focusing on URIs as identifiers, but there are legacy datasets that use other identifiers.
... If they have those, we have to bring those along too.
... Outside that, whatever we put together for a data model, we include principles from Force11 and BPs...
... And then there is Interrelated content, which is what you do with the citation
... And there are ontologies for that.
... It helped me separate in my mind everything out there, and the correlation between the info and the knowledge models.
... [Next slide]
... From the digital librarians...
... This is the core citation model
... Only thing we added: the Citation creator
... There were a few that jumped out at me. Sometimes the publisher, creator and distributor might be broken out
... [next slide]
... Based on the previous slide with the basic elements, we have right now in the model DUV-CitationAct, we recommend it includes the properties that support the attributes here.
... This is up for discussion... did we pick the best properties?
... Also, if something is published on the Web, sometimes you have authors who are a foaf:agent, sometimes foaf:organisation, sometimes foaf:project
... The role of the distributor, whoever makes the data accessible on the web... we're talking about the relationship between that and the dataset.
... We're proposing we go with the basic citation model.
...Then: we're aligned with the digital librarian community. Got feedback from Purdue.
... Through this, I've learned which communities we need to get to validate this.

yaso: Thank you for the update. Comments?

phila: I do, but I know we don't have much time
... Grateful to Eric for that huge amount of work. I need to sit down and send questions on email.

<yaso> http://www.w3.org/TR/vocab-dqv/

Data Quality Vocabulary

riccardoAlbertoni: First a little discussion on the update in the document.
... This week, Antoine changed the data document, updating it according to the decision we made in the last call.

<phila> DQV

riccardoAlbertoni: I think that the most important thing is that we now have an example to show how to associate all the high-quality certifications for a dataset.

<riccardoAlbertoni> http://w3c.github.io/dwbp/vocab-dqg.html#express-that-a-dataset-received-an-odi-certificate

riccardoAlbertoni: This is an example. http://w3c.github.io/dwbp/vocab-dqg.html#express-that-a-dataset-received-an-odi-certificate
... Beside small changes on the document, we have 2 discussions pending.
... We are waiting for people behind the scenes.
... 1) How to encode SLA in a quality policy?
... Yesterday, Nandana sent examples to the editors. We are considering them and hope to add them to the doc soon.
... 2) Re metrics dimensions categories, is this a concrete class or abstract class?
... I want to apologise. Jeremy contacted us editors last week, and we haven't had time yet to answer.
... I'd like to move on to concrete discussion about issues now, unless there are questions?

[no questions]

<riccardoAlbertoni> issue-202

<trackbot> issue-202 -- Relation between dqv, iso 19115/19157 and geodcat-ap -- open

<trackbot> http://www.w3.org/2013/dwbp/track/issues/202

riccardoAlbertoni: Let's start with issue 202
... Which was generated by Andrea Perrego.
...Issue: relationship between the vocab, an ISO standard, the INSPIRE directive and a profile of DCAT
... Perrego suggested some possible requirements to consider.
... In the DQV, we are already saying that a dataset conforms to a standard, we have a specific property in the class for doing that.
... Andrea is asking if we would like to consider also to express that a dataset is not conformant to a standard. Or that it hasn't yet been evaluated against a standard.
... Yesterday I emailed about this
... My position: we should consider to add the support in DQV to express that a dataset is not conformant to a standard.
... But I'm not sure if it makes sense to say that it has not yet been evaluated against a standard.
... So, any comments?

phila: The thing about DCAT-AP and GeoDCAT-AP is that they are in increasing use, the European Data Portal is being launched next week which demands DCAT-AP
... I get hung up on "what is the subject of the triples."
... If I have a dataset of bus stops, the dataset doesn't conform to GeoDCAT-AP, but the metadata about the data may conform.
... You'll need a property that says, "This dataset has metadata that is conformant to this standard [like GeoDCAT-AP]
... "

<phila> <dataset> ex:hasMetadataConformantTo <DCAT-AP>

riccardoAlbertoni: That is a good suggestion
... It's hard to say if this is in the scope for DQV. It could be.
... We are trying to model the quality of datasets. The fact that the dataset is compliant to DCAT-AP, which is basically a vocabulary... so it is not a standard?

phila: The European Commission wouldn't agree that it isn't a standard.

riccardoAlbertoni: I guess there are plenty of standards.
... I wonder if this kind of statement should be captured by our vocabulary or not.

phila: I think it tells you something about the process, how the data was created.
... It tells you the organisation that created this data has taken the trouble to create metadata according to this standard.

<peter> there is an RDA - related metadata to metadata link issue: http://www.rda-jsc.org/sites/all/files/6JSC-ALA-45-CCC-response.pdf

phila: It lets you trust the data further.

riccardoAlbertoni: I get that. I wonder if the property you were suggesting could be a sub-property of dc:conformsTo?
... Shall we open an issue?

phila: Was hoping to help you close an issue

riccardoAlbertoni: This is a good suggestion, but I think it's not helping to close the issue.
... I'm happy you pointed this out.

<yaso> ops

laufer: About the issue of seeing or not if the datasets conforms to a standard:
... I want to ask, what is the model we are assuming for quality? A closed world, or an open world?
... I think we have to assert that the quality of a dataset is in a closed world.

riccardoAlbertoni: I tend to agree

<jerdeb> it should be closed world

riccardoAlbertoni: In most cases, they are working with a closed world assumption. Depends on the community.

laufer: I think it's difficult to assert all the things that a dataset is not.

hadleybeeman: That's profound

riccardoAlbertoni: You are saying there is difficulties in expressing this statement, but you would find it useful to express it?

laufer: I don't know. I think we have to define what are the assertions about quality that our vocabulary will enable.
... So, one can assert that a dataset is conforment to a standard.
... But it seems a strange thing to assert to what the dataset is not conferment.
... If you don't assert that it IS conformant, then we can conclude that it isn't conforment.

riccardoAlbertoni: The point is that Andrea is point out that these kinds of needs are required in some domains, like geographical domain.
... Where there is a huge amount of open data.
... I think we should take into consideration the possibility that saying that one dataset is not conformant to a standard. But I don't think "not evaluated yet" is useful
... And may make the management of metadata more difficult

laufer: If you don't have any assertion about a standard, we can conclude...nothing?

riccardoAlbertoni: Yes, nothing.
... Because some people will work in an open-world assumption. Others in the closed-world assumption. We can't presume.
... I'm not sure; this is an open discussion

<phila> hadleybeeman: I wanted to scroll back to talking about the metadata being conformant to a standard

<phila> ... If I'm combining datasets, that's what I want to know. That then makes it easy to know what I can do with it

<yaso> +1 to hadleybeeman

<phila> ... So +1 for that being a user need

riccardoAlbertoni: Okay, I appreciate that. I think we really need to open an issue here.
... Phil's suggestion could be a good starting point.
... I could have an open action to me to include this new relation?

<phila> [..] Riccardo thinks

<trackbot> Sorry, phila, I don't understand 'Trackbot, please decide whether we should open an issue or an action'. Please refer to <http://www.w3.org/2005/06/tracker/irc> for help.

<scribe> ACTION: Riccardo to send an email to discuss the suggestion to add the assertion that a dataset's metadata conforms to a standard (like GeoJSON) [recorded in http://www.w3.org/2015/11/13-dwbp-minutes.html#action01]

<trackbot> Created ACTION-219 - Send an email to discuss the suggestion to add the assertion that a dataset's metadata conforms to a standard (like geojson) [on Riccardo Albertoni - due 2015-11-20].

<riccardoAlbertoni> issue-188

<trackbot> issue-188 -- Section "Express the quality of a linkset" will be completed by examples coming from Riccardo's work on measuring the quality of linksets between SKOS concept schemes -- open

<trackbot> http://www.w3.org/2013/dwbp/track/issues/188

riccardoAlbertoni: We have already added an example in the doc pertaining to linkset quality.
... I want to close this issue
... As it is written, it is misleading and doesn't make sense. And doesn't require more discussion.

<phila> If it's dealt with it's dealt with IMHO

riccardoAlbertoni: I will add the missing part, as editorial note. Reminder that we have to add an example.

yaso: Is everyone okay with closing this issue?
... I don't think we need a group decision to close this.

riccardoAlbertoni: Yes. No group discussion is needed.

yaso: Thank you, riccardoAlbertoni for this update.

<Zakim> phila, you wanted to discuss 10 secs at the end

<phila> What our final docs will look like (as of Jan 2016)

Any other business

<yaso> close issue-188

<trackbot> Closed issue-188.

phila: You may have seen that we've been redesigning our document template.
... This is a link to the new template, which comes in to force in Jan.
... The final version of this group's docs will look like this.
... Looking for comments on this new TR design?

<riccardoAlbertoni> thank for the suggestions!

<ericstephan> Okay bye everyone

<laufer> bye all, nice wknd...

Yaso: [closes the meeting]

<Caroline> great design! :)

<Caroline> Bye! tks

<peter_> bye


Summary of Action Items

[NEW] ACTION: Riccardo to send an email to discuss the suggestion to add the assertion that a dataset's metadata conforms to a standard (like GeoJSON) [recorded in http://www.w3.org/2015/11/13-dwbp-minutes.html#action01]
[End of minutes]