Making W3C better for Web Data Standardization -- 08 Nov 2017

<dsr> see: https://www.w3.org/wiki/TPAC/2017/SessionIdeas#Web_Data_Standardization

<scribe> scribe: sandro

<scribe> chair: dsr

<scribe> meeting: TPAC Breakout: Practices and Tooling for Web Data

https://www.w3.org/wiki/TPAC/2017/SessionIdeas#Web_Data_Standardization

<fab_gandon> dsr, please post the URL of your slides.

<fab_gandon> ?q

<JohnJansen> https://www.w3.org/blog/2017/10/questionnaire-on-practices-tooling-for-web-data-standardisation/

<JohnJansen> https://docs.google.com/forms/d/e/1FAIpQLSeVyCuNO0YOqktFjSpuMW4KSOKo1-_JijQTyLYt_Hm1VtJt1A/viewform

danbri: can I share some experience from schema.org groups? topical CGs

<ericP> danbri: schema.org is usually in the top 5 CGs in W3C

danbri: like Sports, Meat Products, ...

<ericP> ... meat products is less vibrant

danbri: some created without talking to schema.org
... FIBO an initiative else, and this is a bridge
... was a google for sports schema, about 4 years ago
... "I like this group because I can talk confidently about sports info, without feeling shy about being around all the schema.org experts"

dsr: Then how do folks share expertise across groups?

dan: some folks briding
... seo folks have kind of driven out technical discussion, over to github,
... if anyone can join, how do you manage [cut off]

gkellogg: Email is a terrible tool, (public-vocabs). Github Issues for all its failures works a lot better, but it only inviting to folks with the technical expertise

fab_gandon: Reflecting on the lifecycle of vocabularies vs the lifecycle of Recommendations

dsr: Is the current process fit for vocabs?

eric: Medical and Scientific and Engineering -- alientatingly precise semantics. People exchanging the data are so specialized that our practices are not really applicable --- or maybe I'm wrong --
... the practices they're employing are so precise it;s hard to be a player in their definition process
... so we're not going to say how a radiology report looks, but here's the metadata you want to add when wputting it online

p1: It's a matter of figuring out how to make things extensible

ericP: OTOH, there are groups like DDI which are doing cross-domain specs

p1: You have to get people in the domain to understand you're making something a little more general

<ericP> ~.

dsr: People often dont state their assumptions and aren't aware of them until they bump into other communities

annette: By including them in the development

dsr: It's hard, they might be other side of the world

dan: Discussions being in public view helps, RDF stack helps sometimes
... sometimes people want to solve business problems quickly, then come back
... example around Books vs Products, which overlap a bit. Maybe in 3-5 years we'll start to rationalize
... skills for journalists is another area of overlap

gkellogg: When you talk about ways to represent data, we often go to RDF, but Tabular Data On The Web described how to use CSV files, ...
... it isn't always the publisher that's describing the data
... that's a model that might be applied to other things, like templated parts of wiki pages

p2: Does standardization require data be open?

dsr: No, I tried to cover that in an early slide

Bob Bailey: We did some work on this, allowing re-use while still subject to licensing

scribe: we reached some conclusions, Fair Dataset Principles are similar, so maybe there's a commonality

<annette_g> https://www.force11.org/fairprinciples

<ericP> http://www.datafairport.org/

sandro: post morten vs forward looking?

dsr: try to get input on what would be helpful

<ericP> p3: so if i fill in this form, it looks like it will go into a CSV.

<ericP> ... do we have a standard way of expressing that?

<ericP> sandro: why not RDF?

<ericP> ... or JSON-LD?

<ericP> gregg: or CSVW

<ericP> [continuing through questionnaire https://docs.google.com/forms/d/e/1FAIpQLSeVyCuNO0YOqktFjSpuMW4KSOKo1-_JijQTyLYt_Hm1VtJt1A/viewform]

<ericP> dsr: some stuff needs to be stored forever, but other data should be live

sandro: Do you ever want to data to change its meaning later?

ericP: I think this is about diligence, keeping around old data
... adamant people & lazy people

p3: Legal regulations in healthcare

<ericP> p3: in health care, there are regulations about, once you've recorded a record, you can only ammend it

BrianK: There's no such thing as something that's completely stable, like what a Walrus was 30 years ago might change.

sandro: I think you want observations to be stable

<ericP> sandro: you want observations to be stable

<ericP> dsr: we need to be able to scale up the way the W3C embraces more communities

<ericP> danbri: [is RDF an embrassment or a boon?]

<ericP> sandro: in my experience, JSON-LD allows us to bypass this conversation

<ericP> gregg: what role does data have in W3C?

<ericP> dsr: what does W3C need to do to improve to enable data?

<ericP> gregg: the success story is incredible. we're not just following links around.

<ericP> dsr: i think it's also the metadata

<ericP> sandro: i'm not sure rec-track is worthwhile for vocabs

<ericP> ... we need a model of vocabs that allows us to rev and branch

eric: Huge success stories in working with medical data in RDF

bkardell: I was one of those people who didnt see the value in RDF, or see it as having a place at W3C

danbri: Polarization between data folks and browser folks? what about country drop-down as an example of working together

bkardell: Berlin airlift only happened because of the ideas that became EDI

fab_gandon: Failure vs Success is a matter of against what metric, what criteria?
... RV Guha presentation as ISWC
... about Google Knowledge Graph
... Some standards are linked to Vocabs, other are linked to a Formalism
... The 99 RDF was an a-priori standard, preparing the group; vs 2004 RDF which a a-posteriori standard. So those will need different process
... Also, LOV stuff should be involved

- DRAFT -

Making W3C better for Web Data Standardization

08 Nov 2017

Attendees

Contents

Summary of Action Items

Summary of Resolutions

Scribe.perl diagnostic output