Data on the Web Best Practices Working Group Teleconference

<BernadetteLoscio> https://rawgit.com/gkellogg/36b51a2681e1d6a0a9146041fd6564d5/raw/0b4af28c82074c3936e62645e2f011ed301247e0/json-ld-api-best-practices.html

<hadleybeeman> https://gist.github.com/gkellogg/36b51a2681e1d6a0a9146041fd6564d5

<hadleybeeman> scribe: annette_g

PROPOSED: approve last week's minutes

<hadleybeeman> http://www.w3.org/2016/05/27-dwbp-minutes

<BernadetteLoscio> +1

<ericstephan> +1

<antoine> +1

+1

RESOLUTION: approve last week's minutes

hadleybeeman: Antoine, let's start with your vocabulary

antoine: we still have a number of items to do with this vocabulary. We received some feedback this week, an exchange on the list this morning.
... at the moment, we don't need much input from the working group.

chiming in about issues is welcome, of course. We don't have a lot of issues that are difficult.

hadleybeeman: do you feel okay with the timetable?

antoine: yes, for the moment

<laufer> hi all

hadleybeeman: let's move on to the other vocabulary

ericstephan: I haven't seen any comments about the DUV on the public comments.

ericstephan: there were some editorial comments from BernadetteLoscio that need to get incorporated into the doc

<antoine> https://www.w3.org/2013/dwbp/track/actions/227

antoine: I realized I have an item that I need input from the group on.

<hadleybeeman> action-227?

<trackbot> action-227 -- Antoine Isaac to Work with eric s on writing section on evolution of duv wrt reuse of namespaces etc. -- due 2016-04-01 -- OPEN

<trackbot> http://www.w3.org/2013/dwbp/track/actions/227

<antoine> https://lists.w3.org/Archives/Public/public-dwbp-wg/2016May/0101.html

antoine: I'm asking input from ericstephan and BP editors
... the idea was to put a small bit of text about this in the BP doc. Now I have doubts on whether we should do it.

<BernadetteLoscio> the text was: "The Data Quality [[VOCAB-DQV]] and Data Usage vocabularies [[VOCAB-DUV]] created by the W3C Working Group publishing this document have also sought to minimize the number of formal axioms involved in their definition. For instance, the property dqv:hasQualityMeasurement has no formal domain in the RDFS/OWL sense, even though it is expected to be most often used with resources that are of type dcat:Dataset or dcat:Distributio[CUT]

<BernadetteLoscio> designers to employ it for other types of entities, for which quality measurements would also be relevant but that were not in the focus of the design process for DQV."

antoine: this is the text that was removed. My suggestion is to leave things as they are now.

ericstephan: I've seen several comments on twitter, and this has been on my mind. Do we put something like this in the vocabulary docs? I've seen a number of opinions, including at a European semantic web meeting this week. I'm not sure whether this belongs in the BP doc or maybe summarized as a note. hadleybeeman's point about whether it helps describe the vocabulary is a good one.

<hadleybeeman> Original discussion for this ACTION: https://lists.w3.org/Archives/Public/public-dwbp-wg/2016Feb/0013.html

<ericstephan> thank you!

ericstephan: I've thought the reuse of different namespaces was a strength, but some people see it differently

BernadetteLoscio: I was looking at BP16. The BP is long. I'm not sure it's worth including the text here, but maybe we can include in the vocabularies a mention of the best practices. We do the opposite in the BP doc.

hadleybeeman: other thoughts? apparently not.

ericstephan: One thing I don't want to do is come up defensively in the vocabularies. The DUV is what it is. I would be against trying to rework the best practice, so I agree with what BernadetteLoscio said.
... I do have a section in the DUV that talks about the use of other vocabularies. I haven't looked at DQV in a while, but I could look it over and check that we have similar language, but I don't think we should spend more time than that.

antoine: that sounds alright.
... we may be tempted to create concrete examples. We can try it.

hadleybeeman: in the process of explaining a vocab, it can help to explain things to refer to the other docs. It also enhances coherence within the working group, which may not be a priority.

<BernadetteLoscio> yes!

<ericstephan> thank you for sorting through the details!

<ericstephan> +1 for keeping action open

<antoine> +1

hadleybeeman: so, Eric will take a look at it.

ericstephan: this is great. Thanks, antoine, for bringing that up.
... I think I have my working orders for the week. I don't think I need anything else from the group, other than discussions in email during the week. I'm feeling okay with the timetable.

hadleybeeman: update on the best practices…BernadetteLoscio or newton?

<BernadetteLoscio> :(

<BernadetteLoscio> hello?

<newton> Berna, we can't hear you!

<BernadetteLoscio> can you hear me?

newton: I'm trying to reach BernadetteLoscio. I was working on other projects this week.

<BernadetteLoscio> no :(

<BernadetteLoscio> i can hear you!

<BernadetteLoscio> just a minute

<BernadetteLoscio> ahahhahahaha

<BernadetteLoscio> https://www.w3.org/2013/dwbp/wiki/Status_of_comments_about_the_last_call_working_draft

BernadetteLoscio: this week we worked on the wiki table to collect commetns. We included the comments we've received so far. For some of them, the author already has a proposal. Most are not difficult to implement.
... I would like to discuss comment 6 with the group.
... It's about the need for users to register to use a dataset.

<hadleybeeman> The commenter's message: https://lists.w3.org/Archives/Public/public-dwbp-comments/2016May/0027.html

BernadetteLoscio: for other comments, I can make a proposal and send to the group. The second thing is about the implementation grid. We can have different types of evidence. How can we check or test the BP in each case? It's more clear when we have a dataset, but for guidelines and data portals, it's more general. I don't know if we can have the same form for the three types of evidence.
... let's start with comment 6.

hadleybeeman: I'm trying to separate open data from data on the web. This is the latter, so it's in scope. He wants publishers to describe reasons. That's not technical. This feels behavioral, so in my opinion it's not clear that it belongs in a W3C spec.

<hadleybeeman> s/w3c spec./w3c spec. If he were advocating using a specific vocabulary in that use case, it would feel different to me.

laufer: I agree that it's not a technical issue, but I think it's a worry that the publisher needs to have. Maybe we can put it in a paragraph somewhere. It's kind of a reverse license. The publisher is asking for data from users. It's interesting for the publisher to say what they will do with this data. We can think of it as publication of that data.

<hadleybeeman> annette_g: I agree with Laufer. I think there is a lot of other material in our BP doc that doesn't meet that threshold of technicality. We do want to help people publish data in ways that are useful. I see it in social science data, where people need to have somebody give a little info about themselves because that's part of the legality of publishing.

<hadleybeeman> ...In the harder sciences, people need to report back to a funding agency who is using this data. It's a strong incentive for people publishing to do that. If we can give them some guidance on how to do it will, it may help people publish more data.

<BernadetteLoscio> https://www.w3.org/TR/dwbp/#dataAccess

BernadetteLoscio: I agree that it's not technical, too, and that it's relevant to mention this. Maybe we can include, as Laufer proposed, in the introduction to the data access section.

hadleybeeman: If we are going to do that, we should talk about principles and ideas rather than the specifics that Andrea put. I think the specifics will vary widely from use case to use case.
... I want this to be as broadly applicable as possible. I'm still stuck on the behavioral point. other thoughts?

silence ensues.

BernadetteLoscio, is that enough for you to progress with?

BernadetteLoscio: we'll make a proposal and get back to the group.

<BernadetteLoscio> https://docs.google.com/spreadsheets/d/1a9cOGzWJTIhh2OrAemvWBR8f0rv5xqvL03pJeMrotCo/edit#gid=0

<BernadetteLoscio> https://lists.w3.org/Archives/Public/public-dwbp-wg/2016May/0106.html

BernadetteLoscio: I'm working on this proposal to collect implementation information. An evidence can be a link to a guideline. Makx gave a good example.

<BernadetteLoscio> http://www.europeandataportal.eu/en/content/goldbook/preparing-data

BernadetteLoscio: This is the link that shows our BPs are being used. So this is evidence, but I don't know how we can test.
... In this case, it's hard to show that a guildeline follows our best practice. How can we take this into our report.

hadleybeeman: I'm hoping to clarify. We need things to be testable so that implementers can test that they've met our guidelines. There's no halfway on whether they meet it. So, I'm wondering whether your question is leading us to reword to make things more testable, or whether you are asking what testing means.

BernadetteLoscio: It's more the second question.

<hadleybeeman> http://www.europeandataportal.eu/en/content/goldbook/preparing-data

<BernadetteLoscio> https://lists.w3.org/Archives/Public/public-dwbp-wg/2016May/0106.html

hadleybeeman: the link isn't working

<BernadetteLoscio> http://www.europeandataportal.eu/en/providing-data/goldbook/preparing-data

<ericstephan> are they following the data unavailability bp? ;-)

BernadetteLoscio: they are using the BP, and they have a link to the section of the BP doc.

hadleybeeman: we can use data that is published following these rules, not the rules themselves.

BernadetteLoscio: so evidence of implementation should be an implementation of a dataset

hadleybeeman: yes, or a data portal

BernadetteLoscio: when you say data portal, how do we test? is it the portal itself or a specific dataset on the portal?

hadleybeeman: it depends on the BP. We need to demonstrate that somebody else out there thinks they are also good.

BernadetteLoscio: we can use datasets published before and after our BP doc was published?

hadleybeeman: yes

BernadetteLoscio: can we also include guidelines to demonstrate that our guidelines are being used by others creating guidelines

hadleybeeman: In my experience, the director will be interested in what people are doing rather than what they're saying.

BernadetteLoscio: we can also share this info with others, but not to prove that the BPs have been implemented.
... so now we're going to focus on gathering info about the publication of datasets.
... for the proposal, what we're trying to do is make it easier to give answers. It doesn't mean all the BPs will be implemented in a given dataset or data portal.
... the BPs are in the column on the left

hadleybeeman: what the director needs to see is that for every line there are at least two instances of "pass". But at the top of column C, I think we can't say it can be a guideline.
... other than that, I think what you have is very very useful.

BernadetteLoscio: It would be helpful if someone could test, try to collect some evidence and fill in the form.

hadleybeeman: do we have anyone in the group who has promised to do an implementation?

newton: yes, we have

BernadetteLoscio: feel free to give us feedback even if you aren't doing an implementation.

<hadleybeeman> ACTION: annette to look at optimising the test for BP 31. Multiple tests? [recorded in http://www.w3.org/2016/06/03-dwbp-irc]

<trackbot> Created ACTION-283 - Look at optimising the test for bp 31. multiple tests? [on Annette Greiner - due 2016-06-10].

BernadetteLoscio: the data enrichment test is something maybe we should discuss with Annette. If we have a lot of checks in the same line, that can be a problem. For one best practice, you will see that there is more than one line, because we have multiple checks. A BP will be considered implemented if we pass every test. Can the data enrichment one be more concise?

annette_g: I'll take a look at it.

<newton> q_

hadleybeeman: we have 2 more minutes. anything else?

BernadetteLoscio: we're going to try to work on the comments table and see what we can resolve by email.

laufer: I think a guideline that says what we say in a BP is an evidence. If an organization agrees with us, that is a strong evidence, maybe even stronger than a single implementation.

<hadleybeeman> http://www.w3.org/2015/Process-20150901/

hadleybeeman: we should carry this on in email. Take a look at the process document where it's all laid out.

newton: I was going to ask about JSON-LD BPs.

<ericstephan> thank you all

hadleybeeman: They've asked if we want to host their document in our working group. I wanted to ask editors what they thought and open it up to the working group.

<ericstephan> bye

<laufer> thank you all. bye