Data on the Web Best Practices Working Group Teleconference

<scribe> scribe: phila

<scribe> scribeNick: phila

Minutes from previous call

<Caroline_> Hello! Newton and I are together and he just connected WebEx :)

-> https://www.w3.org/2016/09/30-dwbp-minutes Minutes from 30 Sept

NOTUC

RESOLUTION: Accept minutes of 30 September

BP Status

BernadetteLoscio: We have included some topics on the agenda...

<Caroline_> https://www.w3.org/2013/dwbp/wiki/Meetings:Telecon20161028#Main_agenda

Can we include a data portal as an evidence or we should include a specific dataset from the portal? For example: worldbank, open data portal nasa.

<BernadetteLoscio> https://docs.google.com/spreadsheets/d/1RhMGyG0ZYb73RkteYr39Xqt7f5xi0BUQmSiDqjkHOSA/edit#gid=853876221

BernadetteLoscio: Everyone can take a look at the current status ^^
... We organised the evidence we have so far, counting evidence for each BP.
... Some more critical than others. These are the ones listed on the agenda.
... As it's more difficult to get implementations.
... Because these BPs are not used in the real world.
... What I see is that we have BPs, eg ones related to Metadata, having an API. These are easier to find evidence for - everyone does it.
... But it's more difficult becausae people don't implement and also because it's hard to see if they have implemented it or not.
... eg BP 26 avoid breaking changes to your API. How can we assess that?

<PWinstanley> +q

BernadetteLoscio: Same for data preservation.
... For example, data quality info. Everyone knows it's important, but finding it for humans and machines can be hard. usually find publisher but rare to find more
... So this is one of the questions.
... So should we implement our own datasets as evidence?
... I couldn't find evidence for this BP. But it's possible to implement them.
... We are trying to collect evidence from datasets already available.
... We're asking people to evaluation their datasets against the BPs. But we don't have a new dataset created to show that it's possible to implement.
... So should we do it ourselves?

PWinstanley: One of the things that might be helpful for BP26, for things like the 6Aika openAPI recommendations for cities. We could use the evidence of their recommendation around management of APIs.
... Also, Michael Stowe's publication from Mulesoft (?) guide to developing rest interfaces and APIs, he addresses the issue of stability cf. agile development processes.
... Maybe we could ask him directly for illustrations?

<PWinstanley> 6Aika

<PWinstanley> Vipuvoimaa

<hadleybeeman> https://www.w3.org/TR/prov-implementations/

hadleybeeman: On provenance... the implementations from when PROV was put through Rec, that might provide some useful evidence.
... BP5

BernadetteLoscio: OK, matbe I need to think differently. I thought we had to show a dataset or portal that offers data provenance info both in human and machine readable way, the latter might use prov.

hadleybeeman: All of the implementations in the Prov implementation report might provide pointers.

<PWinstanley> https://github.com/6aika , https://www.avoindata.fi/data/fi/dataset/open-api-recommendations-for-cities

annette_g: For prov, I'd suggest looking at scientific datasets, it's important in this field. But they may not use Prov.
... As for putting up our own, that prob won't be seen as legitimate, so I'd caution against that. But I like Peter's suggestion for looking for other guides.

<PWinstanley> https://www.amazon.co.uk/Undisturbed-Rest-Guide-Designing-Perfect/dp/1329115945

BernadetteLoscio: In this case, we need help. For BPs related to APIs, we don't feel comfortable making this evaluation.

<PWinstanley> https://www.mulesoft.com/lp/ebook/api/restbook

BernadetteLoscio: I don't know how to show that the BP was implemented by someone else.
... Also wrote to annette_g as she knows this field well.

annette_g: I'll do what I can.
... I worked on the building of APIs where we made sure we didn't break it but I'm not sure that we recorded it.
... There are lots of instances of people writing about API development. I'd say pointing to other people's BPs makes sense.

hadleybeeman: Thinking back to our first F2F and saying that each BP needs to be testable.
... It is a BP doc, not a spec.
... Unless we had a before and after to point to...

annette_g: You might find a an API that pinted to a previous version.

hadleybeeman: is there a way to tighten up the language in the BP to make it easier to test.

<Zakim> phila, you wanted to mention DBpedia

<hadleybeeman> phila: pointing to other people's recommendations and guidelines is fine, especially for something like this.

<hadleybeeman> ...That's one way to do it.

<hadleybeeman> ...Then — we just got some evidence from Annette_g but we didn't record it . She just said we're careful not to break stuff.

<hadleybeeman> ...We can use that as evidence

<BernadetteLoscio> +q

<hadleybeeman> ...You don't have to necessarily think of "here is an example of API v 2. And here is v1. And look! they're backwards compatible". Thats more than you need to do.

<hadleybeeman> ...Evidence for machine readable provenance info could be harder. dbPedia's provenance is all wikipedia. Is that enough? not sure.

<hadleybeeman> ...Annette_g has some scientific examples.

<hadleybeeman> ...That's the kind of approach we can use.

annette_g: I don't have much difficulty finding blog examples for API not breaking.
... Get blog posts about twitter API for example

BernadetteLoscio: What we discussed before was that we'd have a link to a dataset or a data portal to show as evidence
... So that's what we looked for.
... Makx suggested that we used Share-PSI as evidence
... because they use our BPs.

<annette_g> this is cool: https://blog.twitter.com/2016/versioning-is-coming-to-twitter-s-ads-apis

<hadleybeeman> phila: It depends on which BP we're talking about .

<hadleybeeman> ...The reason Makx brings it up is that he wrote things for the European Commission, citing our best practices. Those have become official EC publications.

<hadleybeeman> ...If we can present the director with a mixture: direct evidence for some, indirect evidence for others, and for those that are harder we have people saying they've done it.

<hadleybeeman> ...Finding blog posts describing what people did is also helpful there.

<hadleybeeman> ...What you're trying to present to the director: a body of evidence that supports each of the BPs.

<hadleybeeman> ...Some of the BPs around data enrichment will be hard. The only way will be to write to Ghisele and ask for examples.

<PWinstanley> illustration of discussion about API change; https://www.troyhunt.com/your-api-versioning-is-wrong-which-is/

BernadetteLoscio: It's clear, it's more flexible also, thanks
... We'll use other strategies to get more evidence
... We were evaluating the OD portal for NASA, Lewis hasn't answered yet
... We really need help, giving us ideas, also doing the evaluations
... I evaluated the World Bank data portal. Should we have a 2nd review of that portal of that?

<PWinstanley> another useful one to triangulate BP 26: http://www.ben-morris.com/rest-apis-dont-need-a-versioning-strategy-they-need-a-change-strategy/

BernadetteLoscio: Do we truist people who make the evaluation. If it's just the editors doing this? It would be nice to have other people doing this.
... For example, I evaluated ....

<BernadetteLoscio> http://labs.europeana.eu/api

BernadetteLoscio: I evaluated that, and I filled in the spreadsheet on Drive and write to Antoine
... I did the same on Open data impact map
... I asked Lewis for feedback on NASA, but if no feedback from him we'll just use ours.

<Zakim> phila, you wanted to talk about NASA

<hadleybeeman> phila: Although there is nothing wrong with checking your evaluation, or sending the team whose data it is your evaluation — but i wouldn't worry if you don't get a response.

<hadleybeeman> ...We will trust whoever does the evaluation.

<hadleybeeman> ...And for NASA, perhaps Jeanne Holm? She chaired the W3C eGov interest group and used to be involved with the US Data.Gov.

<hadleybeeman> ...She'll know who to contact.

<hadleybeeman> PWinstanley: I spoke to her recently. If you send me an email, bernadette, I'll forward it to her.

<hadleybeeman> phila: But those confirmations are "nice to have". Your evaluation is fine.

<hadleybeeman> ...You don't have to get your evaluation checked by someone else.

BernadetteLoscio: Thanks to Peter as he's helping a lot
... putting us in contact with a lot of people

PWinstanley: pleasure

BernadetteLoscio: When we are evaluating an OD portal, for example the World Bank, what's better to include the URL of the WB portal or the specific dataset?

hadleybeeman: That surely depends on the BP in question

BernadetteLoscio: Some of the datasets are in more than one format

hadleybeeman: I think it depends on which BP you're looking at. This is our internal record for W3C, no one's going to go into detailed checks.

BernadetteLoscio: So I think we answered most of my Qs

riccardoAlbertoni: I have a similar problem when I post my evaluation of ?? which is a collection of datasets. Only some of the datasets follow specific BPs
... So I indicated which dataset

<Zakim> phila, you wanted to talk about the form anad spreadsheet

<hadleybeeman> phila: I think this conversation points to why the spreadsheet is easier to use than the form.

<hadleybeeman> ...I tried filling in the form. I picked a dataset at random from data.io. I spend 1.5 hours and got less than half way through the form.

<hadleybeeman> ...It was time consuming, and frustrating because I wanted to explain my responses.

<hadleybeeman> ...Because this is a best practices document, it's useful to have the flexibility.

<hadleybeeman> ...I will fill it in, because I want to contribute to this — but I find the "pass/fail/partial pass" is good, but you need a text field too'

Caroline_: We also developed the Google doc sheet which is much easier. We're aware of the complexity of the form which is why it took so long to prepare. Hence the index on the left.
... But we understand the complexity.
... How can we approach other W3C members?
... I know Karen said she'd help

<BernadetteLoscio> this is the template: https://docs.google.com/spreadsheets/d/1JE5pDy9YCu9eafQv50JJ3SauK4Jq1QmagV-GCDNY24E/edit

Caroline_: And I wanted to talk about the deadline. We thought we have until Nov 10 to get this done, so we've been sending more and more e-mails, but we have 10/11 as the deadline

<hadleybeeman> phila: Re contacting w3c members: we can go through the AC list (if you are a member). otherwise you write to individual AC reps for companies.

<hadleybeeman> ...I might be able to help you choose companies.

<hadleybeeman> ...In terms of deadlines: The Working group had to be formally extended until the end of December. That's a formal process. A note went out to the membership saying that had happened.

<hadleybeeman> ...We will not get an extension beyond 31 December. So we have to get to PR by the end of November.

<scribe> ACTION: phila to contact friendly AC reps to ask for help [recorded in http://www.w3.org/2016/10/28-dwbp-minutes.html#action01]

<trackbot> Created ACTION-297 - Contact friendly ac reps to ask for help [on Phil Archer - due 2016-11-04].

Caroline_: OK, we'll keep working hard and are grateful for the help.
... We need more robust evidence

hadleybeeman: You asked me to ask the group for more help. Can i paste the list from today's agenda?

BernadetteLoscio: Sure

hadleybeeman: Anything else to cover today?

<Zakim> hadleybeeman, you wanted to ask Bernadette about the email she wants me to send

BernadetteLoscio: We need suggestions for sci data portals

<newton> ACTION: newton to update form with new possibilities of answering (partial fail and partial pass) [recorded in http://www.w3.org/2016/10/28-dwbp-minutes.html#action02]

<trackbot> Created ACTION-298 - Update form with new possibilities of answering (partial fail and partial pass) [on Newton Calegari - due 2016-11-04].

BernadetteLoscio: Feel free to send us a message with possible evidence, a dataset, a portal etc.

riccardoAlbertoni: If I understand, we're also looking for references to other BP docs etc. Are we going to use the same form to collect this evidence?

BernadetteLoscio: What's easier?

riccardoAlbertoni: if the doc counts as much as a dataset for our purposes, then ...

BernadetteLoscio: We can make this distinction when compiling the report

riccardoAlbertoni: OK, we'll use the same form and the editors will make the distinction.

BernadetteLoscio: ANd if we have Qs we'll contact you.

-> https://www.w3.org/2013/share-psi/lg/CzechRepublic/ Share-PSI example

<hadleybeeman> phila: This may be relevant... This is an example from Share-PSI

<hadleybeeman> ...That is a local guide, from the Czech Republic.

<hadleybeeman> ...It tells you exactly which of the best practices that Share-PSI adopted, that are recommended by the Czech government.

<hadleybeeman> ...There are 39 of them, one for each country .

-> https://www.w3.org/2013/share-psi/bp/ BP page

<PWinstanley> +q

https://www.w3.org/2013/share-psi/lg/

PWinstanley: Are we going to be sending this request to Share-PSI
... Some targeted e-mails would be helpful

Next meeting

<PWinstanley> hadley I can't hear you

hadleybeeman: Let's set our next call now. If our deadline for evidence is 10 Nov, that's almost 2 weeks. Do we need a call next Friday?

BernadetteLoscio: For me I don't think we need next week

hadleybeeman: So our next call is 11/11 (a date that works on both sides of t he Atlantic)

So DNM is 11/11

<riccardoAlbertoni> good weekend!

<PWinstanley> bye!