Data on the Web Best Practices Working Group Teleconference

<Caroline_> we will be in the webex

<scribe> scribe: phila

<scribe> scribenick: phila

Preliminaries

PROPOSED: Accept last week's minutes http://www.w3.org/2015/12/11-dwbp-minutes

<hadleybeeman> +0 (wasn't here)

<ericstephan> +1

<yaso> +1

<riccardoAlbertoni> +1

<antoine> +0 (wasn't here)

+1

<deirdrelee> sc0

<ericstephan> Great turn out this week...wow

Last week approved 2 publications

<EricKauz> +1

RESOLUTION: Accept last week's minutes http://www.w3.org/2015/12/11-dwbp-minutes

<BernadetteLoscio> +1

<newtoncalegari> +1

<Caroline_> +1

<laufer> +1

2 new publications yesterday

http://www.w3.org/TR/dwbp/

BP & DQV update

http://www.w3.org/TR/vocab-dqv/

<hadleybeeman> phila: we have two new publications from last week

<hadleybeeman> ...On this occasion, the editors (Newton and Riccardo especially) put a lot of effort into getting them ready to publish

<BernadetteLoscio> hello!!!

<ericstephan> yeah for publishing!

<hadleybeeman> ...They did thinks like link checking and validation and spell checking

<hadleybeeman> ...It still took me a lot of time to get them ready for publication, but less than usual

-> http://www.w3.org/blog/data/ Data Activity Blog

<hadleybeeman> ...Yesterday was a busy day for Data Activity.

<hadleybeeman> ...The Spatial Data on the Web updated their Use Case doc.

<hadleybeeman> ...They're close to updating their BPs

<hadleybeeman> ...Also, CSV on the Web became a recommendation

<hadleybeeman> ...They're still writing a primer, but they've finished their substantive work

<hadleybeeman> ...So it was a good day

<hadleybeeman> ...Eric is very close to publishing the DUV

<hadleybeeman> ...Big thank you to everyone involved in those documents.

<hadleybeeman> deirdrelee: So, dissemination? How can we help?

<ericstephan> yes congrats Phil! Great efforts

<hadleybeeman> phila: I did the blog post above, so we can find all these things

<hadleybeeman> ...There are several examples in the BP doc where we refer to the CSV on the Web stuff.

<hadleybeeman> ...It also refers to both of the vocabs we are working on

<hadleybeeman> ...Generally speaking, our work here is about improving the ecosystem for sharing data online

<hadleybeeman> ...These are all parts of it.

<hadleybeeman> ...So, please: the usual blogs and tweets. But more importantly, using it!

<hadleybeeman> hadleybeeman: Big congratulations!

<hadleybeeman> deirdrelee: Do the editors want to say anything?

<riccardoAlbertoni> no really, just thanks all

Data usage Vocabulary

<ericstephan> http://w3c.github.io/dwbp/vocab-du.html

<BernadetteLoscio> yes! thanks all!

dataset usage Vocabulary

-> http://w3c.github.io/dwbp/vocab-du.html The DUV

ericstephan: I felt last week that we had a doc that needed work but that we had good comments that helped Berndette, Sumit and I to make progress
... I now feel comfortable putting it in front of people seeking wide review

<BernadetteLoscio> http://w3c.github.io/dwbp/vocab-du.html

<ericstephan> http://w3c.github.io/dwbp/vocab-du.html#Vocab_Overview

ericstephan: One of the things that we did was to add to the vocab overview section, we added 3 subsections
... What's misisng is links throughout the doc to make it more intuitive
... Everyone commented last week, it's important to call out which classes and properties are relevant to that
... the other thing we did was to go through and make sure that we were reusing vocabularies
... That was a good exercise in following BP on reusing vocabs

<ericstephan> http://w3c.github.io/dwbp/vocab-du.html#relationship_vocabularies

ericstephan: In addtion to the improvements to the model, in section 5 I added some links to what we were motivated by
... My concern about bringing this to the citation community was... I'm keen to see if there are better directions for us to follow.
... We should be sensitive to open data efforts
... Some of those links are to things universities are doing

BernadetteLoscio?

<BernadetteLoscio> its ok!

<BernadetteLoscio> please continue...

ericstephan: Thanks riccardoAlbertoni for sharing his recommendation
... using multiple classes in the doc can be confusing - anyone got better visual suggestions?
... We were able to squeeze everything into the model

BernadetteLoscio: Thanks Eric for the report - I think you said everything. We still have sime improvements to make.

<ericstephan> Berna and I did a lot while balancing time zones!

BernadetteLoscio: We had some nice comments from the group and we're going in a good direction.

ericstephan: We were balanced between the 5 hour time difference with out schedule - which was an added challenge

deirdrelee: So we still have 40 mins for this call - would it be helpful to go thrpough specific issues?

laufer: We'll still have the domain definitions in the doc of other vocabularies, i.e. we seem to be defining properties for otehr vocabularies? I don't agree with this

<deirdrelee> laufer: in the diagram, there are some arrows that have two ends, I don't like this

laufer: (Some missed) - but Laufer is saying that he doesn't like defining domains and ranges for other vocabs
... and some arrows are bidirectional which I don't like

<deirdrelee> ... it gives the impression one concept is inverse of the other and this isn't the case

ericstephan: In terms of the double arrow - that was our call - making the presentation more readable
... It reduces the number of arrows

<antoine> I agree with laufer about the point on double arrows

ericstephan: I thought we had taken out D&Rs on otehr vocabs

BernadetteLoscio: If we look at DCAT, they have domain and range restrictions on properties. They don't specify domain.

<BernadetteLoscio> http://www.w3.org/TR/vocab-dcat/

BernadetteLoscio: In section 5.1 they define class:Catalog and they give the range of dcterms
... We're not re-specifying the classes and properties, we're just following DCAT
... And the double arrows, they're inverse properties

laufer: No, hang on

BernadetteLoscio: One that is inverse is oa:hasTarget...
... We have only 3 double arrows in the diagram.
... oa:hasTarget and duv:gasUserFeedback are not inverse
... I agree with you laufer but the otehr two are
... We had separate arrows in the diagram but it was too crowded so we used double arrows
... I don't mind. it'snot a big thing.

laufer: In terms of D&R, I think it's wrong. If DCAT defines a range, it's the same as in the source vocab

<antoine> yes we should probably fix it for the one double arrow we have in DQV!

laufer: If we define a domain we're extending the other vocab
... for example, if we say dcterms:creator has a domain that we define that a reasoner might infer that

deirdrelee: Interesting discussion

<riccardoAlbertoni> +1 to phil

<ericstephan> I like that Phil

phila: Repeated what was in his recent e-mail

laufer: I disgaree with phila.If they are comments, OK, but if we have RDF that specifies the D&R a reasoner will concluse what we say

phila: I agree that's the theory, not sure if it's the practice

antoine: I understand laufer's point. But I think we should distinguish between where we actually do define a D&R and where we mean a commnet
... We could perhaps say "expected subject" or "expected object"
... That's how schema.org seems to be going

laufer: It's not a problem to use any classes in the D&R if they're not already defined
... I agree that we can say 'expected class' etc. but I don't like it in the RDFS

ericstephan: ... This is a question to BernadetteLoscio - I remember the note about a usage. I'm confortable with leaving just as text
... I wanted to put something in the text as well as the picture

BernadetteLoscio: I'm fine with textual expectations, remving domain from the table and using usageNote instead
... dct:creator has a range of prov:Agent
... Not sure if it's going to look nice in the doc

<laufer> we have to understand the inferences that come from these definitions

BernadetteLoscio: I couldn't find info about D&R for CiTO

antoine: If no D&R is not specified in the original vocab they probably have a reason for keeping it open.
... There might be ways to express that a property has an expected usage

<Zakim> phila, you wanted to talk about provAgent

<deirdrelee> phila: As far as I can see, prov has reinvented foaf, and they created provagent, from my understanding provAgent is reinvention of foaf:Agent

<deirdrelee> ... and dcterms:agent

<deirdrelee> ... antoine mentioned datashapes, which is linked to wider work of application profiles, EC has been working on DCAT-AP, Andrea and JRC are working on GeoDCAT-AP

https://kwz.me/Rn

<deirdrelee> ...it would be great if datashapes could agree on an approach and move forward

<deirdrelee> ... this links leads to a large set of shape expressions (shexs) that seem to be winning the day, a common way to define an application profile

<deirdrelee> ... which is what we're talking about here, it's not up to us to redefine provAgent. We don't mind if you use foafagent or dcterms agent, it doesn't change the semantics

<deirdrelee> ... much happier to rely on work on datashapes based on this!

ericstephan: I was talking to Jim Myers this week
... In the prov group they caled it aspects
... the way you're describing things about the shapes WG is right on track with where we're trying to go
... We're trying to show uage patterns

<hadleybeeman> I'm thrilled to hear we're so connected and getting so much feedback from other related work.

ericstephan: That sounds timely
... The more we constrain models or have conflicts between D&Rs, the harder it is to reuse vocabs
... So the Shapes WG is an important part of this
... I'm wondering for now whether we should put in usage Notes
... anda then if there are opportunities to express a shape in the new year we can leverage that

<laufer> The range of duv:refersTo has two classes: dcat:Dataset, dcat:Distribution If we have this in our rdf edfinition of duv, what we are saying is that the resource O in a triple "S duv:refesTo O" is a dcat:Dataset and a dcat:Distribution. And I do not thinnk that a resource could be a dcat:Dataset and a dcat:Distribution

hadleybeeman: Is there anyone involved in both WGs

<deirdrelee> hadleybeeman: do they have a use-case document?

-> http://www.w3.org/TR/shacl-ucr/ Data Shapes UCR

antoine: I'm not folliowing the shapes work, but DCMI put in some use cases, and Karen Coyle is very active
... I don't think we'll have any use cases that aren't already in there

<annette_g> It looks like Sumit is in the data shapes WG

talked a bit about use cases and fulfilling them

<BernadetteLoscio> and changing the arrows

ericstephan: So if we make the changes that laufer was asking for, i.e. making D&R restrictions into usage notes, can we vote for a publication?

deirdrelee: I'd think we should wait as we're running out of time

<Zakim> hadleybeeman, you wanted to ask if we take out domain and range restrictions... will it affect how we judge conformance?

hadleybeeman: If we take out D&R for a term, we still need to demonstrate conformance - how do we do that?

ericstephan: We need an answer for that

antoine: I'd say it's a good question. The formal semantics of the open world, we're not really imposing anything. Even if the world's triples say that a dataset is also a car, you're Ok as long as no one says it's not a cat

<hadleybeeman> @antoine, I'm still a bit unclear on how that answers the question, but will be happy to see as the changes are made.

phila: Suggests that we aim to vote to publish during our next meeting on 8 Jan 2016
... That gives us the break to review

<yaso> +1 to vote on 8 Jan..

phila: People who will be on holiday then can register their assent/against during the holiday
... Notes to ericstephan that even if we vote now, it wouldn't be published until 5 Jan - we're talling about a 1 week delay in effect

BernadetteLoscio: I'm OK with this

ericstephan: I have snow and sleet here, so I can stay inside and work on this

deirdrelee: So with that...
... We'll close up for today. Thanks everyone for the discussion on domains and ranges and where the Data Shapes Wg is

<riccardoAlbertoni> \me merry xmas/ vacation to all

<laufer> bye all... nice weekend...

<yaso> thank all

<yaso> bye!

<annette_g> bye all, and happy holidays

BernadetteLoscio: I'd like to say I think we did a lot this year - thanks everyone

<Caroline_> bye! Thank you! Great Holidays!!

<newton> bye all, nice holidays! :-)

https://www.force11.org/meetings/force2016/forms/pre-conference-meeting-proposals