Spatial Data on the Web Working Group Teleconference

<phila_train> meeting: SDW F2F Day 1

<Linda> Hi Kerry Ed is not here yet

<kerry> Hi linda!

<kerry> I have not managed to get in on webex yet eiter

<Linda> Phil is not here yet either. So I suppose we wait a little...

<kerry> ed just sent an email -- maybe 10?

<kerry> also bill roberts said 10

<phila_train> hi all. Text messages and emails flying around suggest Ed, Bill and I will arrive at Amersfoort station in one hour's time. I just pulled out of Rotterdam Centraal.

<kerry> phil had said he would be late -- a few days ago -- maybe 11?

<BartvanLeeuwen> morning everyone

<BartvanLeeuwen> hi kerry

<kerry> hi bart!

<kerry> do you have webex in the room?

<Linda> not yet

<BartvanLeeuwen> we are working onthat....

<kerry> ok -- ta. It looks a bit like it might need phil to me

<BartvanLeeuwen> kerry: but I'm not able to dail..

<BartvanLeeuwen> as in , I can' start a audio connection

<BartvanLeeuwen> ok, can't you 'take' over the meeting and become host ?

<BartvanLeeuwen> kerry: do you have skype credits ?

<BartvanLeeuwen> okay

<BartvanLeeuwen> can you call long distance on someone elses expense ?

<BartvanLeeuwen> whats your number ?

<BartvanLeeuwen> for now yes

<BartvanLeeuwen> hmmm

<BartvanLeeuwen> that is us

<BartvanLeeuwen> +31 number

<BartvanLeeuwen> kerry: we will do that

<BartvanLeeuwen> you didn't receive a call ?

<kerry> ok, who can chair please?

<BartvanLeeuwen> kerry: linda is chairing

<frans> I will start with the UCR topic

<jtandy> scribenick: jtandy

<scribe> scribe: Jeremy Tandy

<kerry> trackbot, start meeting

UCR document

<trackbot> Meeting: Spatial Data on the Web Working Group Teleconference

<trackbot> Date: 08 February 2016

[we finally get some audio working on the teleconference]

<kerry> scribe: jtandy

<kerry> chair: kerry

<scribe> scribenick: jtandy

<kerry> scribenick jtandy

approve the minutes (of last f2f)

kerry: did we approve the minutes of the last f2f?

[looking]

<kerry> https://www.w3.org/2015/10/25-sdw-minutes

jtandy: query- the minutes from sapporo

kerry: yes- 2-days

propose: approve sapporo f2f minutes

+1

<BartvanLeeuwen> +1

<frans> day two: https://www.w3.org/2015/10/26-sdw-minutes

<Linda> +1

<frans> +1

<kerry> +1

<AndreaPerego> +1 (I was not there, but I trust you)

RESOLUTION: approve sapporo f2f minutes

<ChrisLittle> +0 not there

patent call

<kerry> https://www.w3.org/2015/10/26-sdw-minutes

propose: approve sapporo f2f minutes day 2

+1

<kerry> +1

<BartvanLeeuwen> +1

<Linda> +1

<frans> +1 (I was not there, but I trust you)

RESOLUTION: approve sapporo f2f minutes day 2

<LarsG> +1

kerry: BIG thankyou to Linda and Geonovum for hosting the F2F ...
... does the patent call

<Linda> You're very welcome! Nice having you all here.

<kerry> https://www.w3.org/2015/spatial/wiki/Patent_Call

kerry: reminds people that this follows both W3C and OGC process; particularly for OGC, please declare patent interests ...
... let me know of anything

Linda: we have two new people in the group today; Rob and Stan ...
... from OGC

jtandy: asks Rob and Stan to type their intro into the IRC after they do their verbal intro

new participants

<roblemmens> Intro Rob Lemmens

frans: says great to have you with us ... hopefully you can provide real examples

kerry: [Rob and Stan] are already OGC members ... welcome
... gives some IRC advice to the newcomers

<roblemmens> We're ITC, Univ Twente, working on visual linked geodata explorer and VGI ontology

kerry: to the agenda ... https://www.w3.org/2015/spatial/wiki/Agenda_F2F3

UCR -- steps to ensure that the remaining issues will not stay open indefinitely

kerry: invites frans to lead ...

frans: lists the open actions on the screen https://www.w3.org/2015/spatial/track/actions/open
... there's [quite a large number]
... there has been discussion on the email list, so that probably means these aren't trivial to resolve
... but we need to finish the requirements before we get on with the work
... otherwise things are being done backward
... it's good to take a look at these issues again now
... we actions in the tracker to get things done
... we assigned actions for the deliverables to particular people
... we can't allocate to multiple people

kerry: suggests that we operate in a bit of an agile fashion ...
... leave some requirements open whilst we don't have resources to work on them
... also, the timeline shows we have plenty of time for the UCR
... we don't need to resolve all issues up front

frans: I got the impression that some of these issues are not resolved because they're difficult
... not because we're not working on them yet
... difficult does not mean important

kerry: fair enough - seems reasonable to review these actions [periodically]
... but they don't have to be resolved where we are not actively working on the deliverables

frans: OK - the intention was not to shift the work, just to begin collaboration with these folks
... the other deliverable editors can then help prioritise which issues to focus on
... in some cases, we just need to synthesize the discussions [to form a conclusion]

kerry: what we could do for SSN in particular, is add these issues to tomorrow's agenda
... for the others, it's a bit harder, because they're not started yet

frans: seems sensible!

kerry: better to try to resolve these actions F2F ...
... time deliverable - we might be able to do that
... coverages deliverable - not really started yet
... does that help?

frans: it's good to make sure that the deliverables have a good chain of requirements
... [with assoc use cases]
... I see that BP and SSN have already been through this phase

kerry: tracking deliverables back to requirements and use cases is a good idea
... wouldn't say that SSN has done this yet - BP have done this and this might provide an exemplar to follow
... question to frans - are there any specific actions / issues you want to resolve?

frans: not a good idea to go into the details now - because these are difficult, we'll loose a lot of time
... let's try to keep the UCR doc up to date with the work of the deliverables

kerry: we do have _some_ time right now
... we could try a couple of the harder ones right now
... are there any?

frans: asks for volunteers :-)
... how about CRS?

<Linda> jtandy: re CRS requirement

<kerry> jtandy: best practice addresses crs

<kerry> ...by taking some of the heat away..

<Linda> ... in the BP we are explaining when you need another CRS than WGS84. Do we think that's a good approach and would it resolve the action?

<kerry> fantastic! -- i liked what i saw on this in the BP

<LarsG> jtandy: we need guidance on when WGS84 is not enough and when the publisher needs to be explicit (and how to do it)

<LarsG> ... this also goes into geometry etc.

<LarsG> ... How are we able to meet the requirement we've got

<LarsG> frans: But we've been unable to specify requirments

<LarsG> jtandy: But isn't the req: when is WGS84 not enough and what do I do when not?

<LarsG> frans: Has problem with the way it's done in geosparql (conflating CRS URI with geometry)

<LarsG> ... BP editors can try to formulate requirements as the BP document progresses

ChrisLittle: just to follow up frans comment about chaining back to requirements and use cases
... the spreadsheet we produced
... is it deficient?

frans: we stopped working on that
... it was just input to our UCR FPWD

kerry: did we resolve the CRS issue?

jtandy: responds [...]

<rachel> e.g. in geojson when you specify CRS other than WGS84 most mapping applications that read geojson just ignore that and try to plot it as WGS84. The applications need to show an “unsupported crs” message

jtandy: noting that frans encouraged the BP editors to write down what they understood were the requirements

kerry: can we write these in the TRACKER now????
... please :-)

BartvanLeeuwen: asks that we have a short break to organise webex connection

[... short break ...]

<rachel> ...in other words, requirement that CRS issue needs to be solved on consumer end as well as publisher

[@rachel ... yes I think so]

<BartvanLeeuwen> kerry: do you hear us ?

<ChrisLittle_> bye

jtandy: asks frans to tell us where we got to ...

<eparsons> Morning !!

[resuming ...]

frans: so we've done topic for point 1 "https://www.w3.org/2015/spatial/track/actions/open"
... before we finish that, are there any other volunteers to talk about open issues

<kerry> chair: eparsons

frans: what about ChrisLittle - he sent an email about time requirements

ChrisLittle: no - I think we're fine there

frans: also - what about SSN folks?
... there's still an open action to turn the reqs from the SSN incubator group into use cases [?]
... that's a question to kerry: are there still missing SSN requirements in the doc

kerry: I put a summary on the mailing list to cover this
... still an open action on me
... leave at that for now and remind me again
... I've done the analysis

frans: can we do that at the same time as we go through the wiki page for SSN requirements?
... is that possible?

kerry: ... hmmm ... hmmm ... yes

eparsons: what's the doubt?

kerry: there isn't one
... the wish list is 98% constructed by me; it's not the original incubator group list, its what people have asked me for
... these might not actually constitute _requirements_

eparsons: [agrees] we need to trim the wishlist; having more concrete requirements might help this
... is this a good mechanism [for managing the scope] of the SSN deliverable

kerry: yes ...
... we can keep the bigger list for context

eparsons: thanks kerry - that helps

kerry: notes that in the spirit of W3C chair training, have a wishlist of things that you _want_ to do, but might not have the resources to get to
... the SSN wishlist is done in that spirit

frans: no other volunteers?
... we've talked about trying to sync the UCR with other deliverables ...
... let's move to the next topic

<kerry> https://www.w3.org/2015/spatial/wiki/SSN_Wish_List is the "wish list" of things for ssn I have picked up

Use cases missing arising from recent work?

<phila> Thansk Kerry

frans: I have a new use case ... the web is a distributed system
... it's not always easy to find what you need
... does the BP doc provide a solution to this?

billroberts: we can't provide a _complete_ resource

eparsons: [agrees] we provide a starting point

frans: proposes that we add this use case, and explicitly say that the BP doc resolves this

eparsons: this could work - the link has always been implicit, it would hurt to make this explicit
... what do the BP editors think?

Linda: hmmm ...

frans: so the BP doc provides a starting point to get people going.
... similar to the "linked data book" that I started with for that topic

Linda: yes- this could work

ChrisLittle: the only thing that's not explicit in the charter that was provide the _single point of reference_
... we provide some best practice

eparsons: we're not trying to be canonical
... it's only as good as it is when the doc is published
... it's ok to have this "meta requirement"
... but don't oversell this as the "final word"
... maybe what we need is a _short_ summary that helps people get started; the BP doc will [likely] be quite long

frans: yes - the BP doc needs to be read in many ways

<phila> jtandy: I noted that you cited the Linked data books as a place to learn about that. Are you hoping that the BP doc will be as informative?

<phila> frans: It's called a book but it's a single Web page

<phila> billroberts: It's the one Tom Heath did

<BartvanLeeuwen> http://linkeddatabook.com/editions/1.0/

<phila> frans: It was a complete narrative with pointers to relevant information so if you needed to pick a single starting point, it was a good one

eparsons: @frans ... you've given yourself a task to write the requirement there

<phila> frans: I think that's something to keep in mind when writing the Bp doc

<phila> BartvanLeeuwen: I think there's so much info around about spatial data, it's hard to create a single starting point

<phila> BartvanLeeuwen: But I like the objective

<scribe> scribe: phila

eparsons: We recognise that we're not starting with a blank sheet of paper
... We're opening up that info

<scribe> scribe: phila

<scribe> scribeNick: phila

jtandy: I'm concerned that we're setting ourself too big a target
... I don't think we can aim for publishing a complete ref for spatial data
... We should limit our focus to getting spatial data 'on the Web' - although I don't think we all agree what that means
... not about spatial data in general

eparsons: We should recognise 2 broad communiites. Someone who is publishing spatial data for the first time
... the other one is slightly different. I already have data that I probably already publish using OGC services
... but I need to do x y z
... So they start from different points and therefore need to give them both a starting point.
... You're a good citizen of the Web, it's about URIs, etc.
... It doesn't need to be an encyclopaedia

frans: If you say on one side that you have a requirement but we can also say that we can't necessarily meet them all.

jtandy: When you word this requirememnt, pls make sure you're talking about publishing spatial data on the Web.

ChrisLittle: The contents of the LD book - there's some good structure there you can reuse (nick)

<eparsons> action frans "to write meta requirement that the BP will provide guidance to spatial data publishing"

<trackbot> Created ACTION-135 - "to write meta requirement that the bp will provide guidance to spatial data publishing" [on Frans Knibbe - due 2016-02-15].

jtandy: We have an action to think about how to structure the BPs to give people an 'in'
... especially for people with a short attention span

frans: Are thereany other use cases arising from recent work.

roblemmens: We're new of course. I've been looking at the UCR and BP doc
... I've been looking at the VGI paradigm. Are there use cases for that?

<eparsons> VGI

<AndreaPerego> Another source to look into is "Linked Data Patterns" (by Ian Davis and Leigh Dodds): http://patterns.dataincubator.org/book/

<eparsons> https://en.wikipedia.org/wiki/Volunteered_geographic_information

roblemmens: I saw some frontier issues in the UCR, but maybe they're spread here and there. It's related to provenance of course.
... For those datasets, you want to know how they came into being

<ChrisLittle> +1 to roblemmens comment

frans: We have this general discoverability requirement that everyone thinks is important. Not going far enough?

roblemmens: You have data in general, but volunteered data...

s/froniter/volunteered/

<jtandy> +1 to @roblemmens thoughts about VGI; especially regarding provenance of that info

eparsons: Playing devil's advocate - don't issues of provenance solve that?
... All that's different is the people who create it
... If you know that this temperature record was made by the Met office vs. someone in their garden, that tells you what you need, no?

roblemmens: But they sometimes form communities that have a common purpose, and that is also discoverable.

jtandy: So we tried to pick up the VGI stuff in the BP doc. In many situations where people are volunteering that info, they're using existing channels, like Twitter
... So you end up with unstructured info that the service will probably have to process.
... But yes, VGI is often as structured as official info. OSM is an obvious example.

ChrisLittle: Ancenstry info is an example., In the UK they've all agreed to use pre-1972 boundary change data

roblemmens: OK, but there are different levels of formality, and depending on that you may need to do more or less pre-processing

<kerry> https://www.w3.org/2015/spatial/wiki/Working_Use_Cases#Crowd_sourced_earthquake_observation_information_.28Best_Practice.2CSSN.29

kerry: I wanted to point to that use case ^^
... It's a VGI use case, not sure if it made it into the UCR
... This talks about official info that is volunteered.
... So maybe we've already covered it?

eparsons: Asks roblemmens to look through the requirements to see if VGI is covered sufficienetly.

roblemmens: Yes, OK. I'm not sying it's missing, I'm just checking that it has been discussed.

Linda: Do we have requirements on provenance.

frans: Prov is a general data issue
... So this req is about alignment with existing practices.

<eparsons> action roblemmens Review current Reqs and add one if needed for VGI issues

<trackbot> Error finding 'roblemmens'. You can review and register nicknames at <http://www.w3.org/2015/spatial/track/users>.

frans: Many issues that people deal with, like prov, is not specially spatial

Linda: responding to Frans' question about additional UCRs. I sent an e-mail to the list a while back with an addition to the construction use case
... Do you intend to add that Frans?

frans: I think more discussion is needed. I had questions about it - I created an issue so it won't be forgotten.

Linda: So let's discuss..

issue-38

<trackbot> issue-38 -- decide if and how to add the proposed use case -- raised

<trackbot> http://www.w3.org/2015/spatial/track/issues/38

Linda: This is from the construction industry. They find they have to integrate data as construction is in progress.
... Some data is about real world things, some data is about virtual things/records in a database. Those differences matter of course. So we need some way to model these things.
... They wrote their whole use case down and I'm hoping it can be added to the UCR.

-> https://lists.w3.org/Archives/Public/public-sdw-wg/2016Feb/0018.html E-mail thread

Linda: The key thing is that difference in perspective
... They talk about the chart perspective/map perspective. Those are the spatial things and they need to integrate that with the real world.

phila: Waffles on about HR14
... Emphasising that a URI is a dumb string and you don't know what it identifies until you dereference it

eparsons: Yes, we do need to think about it. In the spatial world, we only talk about data about real world things.

frans: We don't have the same confusion

<jtandy> [note: we had this discussion on the mailing list, see thread: https://lists.w3.org/Archives/Public/public-sdw-wg/2015Oct/0125.html]

eparsons: I think we do, in that don't think in terms of the physical thing that you can walk into

billroberts: You can't think in terms of who owns the spatial data, you think about who owns the building.

ChrisLittle: The building would be a point on a map

<jtandy> [also see URLs in data primer, W3C WD https://www.w3.org/TR/urls-in-data/]

ChrisLittle: then you get closer and it has extent and so s

frans: That's an issue, you have multiple representations of the thing. The idea of having to do something with the real world thing never arises.

LarsG: You said, it's not your problem, Frans, but someone else will have the problem.
... If we don't separate the thing from the metadata, that might not be a problem for you but it might create a problem downstream.

jtandy: I just wanted to highlight that we had a lot if this discussion on the email thread. We also discussed it f2f in Saporro

<kerry> s/saporro/Sapporo/

jtandy: A guy called Tim said that, yes, we've talked about this for a long time and in the end people generally think it really doesn't matter. using a VCard as a proxy for the person is commonly done and we need to get over it.
... We can use 'punning' so that we can treat it as a real world thing or a record.

frans: Can we ignore it unless it becomes a problem.

jtandy: SDOs have this problem. But schema.org looks at what people do and people do this all the time...

billroberts: Is anyone sufficiently familiar with Jeni's work on trying to solve this?

-> https://www.w3.org/TR/urls-in-data/ URLS in Data

jtandy: What she said is - the best thing to do is to give differnet identifiers for the real world thing and the record. Most of you don't, but if you don't, it would be helpful if you ...
... distinguished between a direct property about the thing itself (creation date etc.) and an indirect one about the page about the thing.

frans: So that's an example where people use web pages as the real thing.

Linda: That's what this use case from the construction industry is about. They need to make that distinction.
... I think one of them is coming to the workshop on Wednesday.

BartvanLeeuwen: I was amused that HR14 comes up in every F2F meeting I've ever been to.
... I am sympathetic with Frans's view of ignoring it until and if it bites us.
... But we might get comments...
... In the end, we are probably going to have to formulate an answer.

<Zakim> LarsG, you wanted to talk about licensing content vs licensing metadata

frans: Points to some proposed text... you mint a URI for the building ...

(lost some detail)

jtandy: INSPIRE gives identifiers to information records. And then you have thematic identifiers and these can be confusing.
... I think we're saying that you should create a URI for the real thing.

LarsG: We've said in the library world that we need differnet URIs for the document and metadata about the doc.
... There's a huge difference in the licensing of the two

AndreaPerego: We had the same in DCAT which has the datasets and catalogue records.
... It's usually the record that gets the reliable URI.
... It's probably related to how we use the IDs.
... Any feedback on the UK work on URIs for spatial objects?
... Did it meet the requirements?

phila: That work is withering on the vine, not because it isn't good, but because the individuals are otherwise engaged.

jtandy: Since that work, there seems to have been a groundswell of using the context to determine whether we're talking about the real world thing or the data.

eparsons: I have an app on my phone that opens the door to my hotel.
... So my phone is interacting with a real world thing

<jtandy> [wow - he can use bluetooth LE to open the lock on the door after online check-in!]

eparsons: We can skirt around it but I think we do have to draw that line in the sand and say that it is going to be necessary to have different identifiers for the two. And that would be BP

<jtandy> [key point is that we need to be _clear_ which is the real world lock, which is the information record about that lock]

Linda: They're not trying to describe the building, they're trying to integrate different datasets related to the building.

eparsons: Talks about links between thing and n records about the thing.

AndreaPerego: The catalogue record is defined in the context of the catalogue, it's not an absolute resource.
... In this case, the prov of the metadata is important. It's important to tell the record from the description of the resource and the resource itself.
... It makes sense to me to apply that as a general requirement for data on the Web.

frans: So the req from this use case is being able to integrate data with different perspectives.
... I'm looking for a case where the difference really matters.

<Zakim> jtandy, you wanted to note that we can make general statements about data on the web- but through a spatial lens

eparsons: My lock on the door is my example

billroberts: We need to be careful not to be drowned in the swamp of HTTP Range 14

eparsons: I think it's a bigger issue for the spatial community than it is for the general data community.
... We're both communities that have grown up digitising representations of the real world. But now we're seeing more need to refer to the real world thing.

<scribe> ACTION: frans to clarify use case requring differentiation between real world and representation identifers, during Wendesday's workshop [recorded in http://www.w3.org/2016/02/08-sdw-minutes.html#action01]

<trackbot> Created ACTION-136 - Clarify use case requring differentiation between real world and representation identifers, during wendesday's workshop [on Frans Knibbe - due 2016-02-15].

jtandy: I think we're Ok to cover this as long as we show the issue through the spatial lens.

eparsons: Any more UCR issues to tackle?

kerry: I don't have another use case, but I was going to suggest we might move on to the topic of @@ ?

eparsons: Can you clarify who is coming on Wednesday?

Linda: Our regular Platform Linked data Nederland will be here. Some are very technical, RDF heads, others are more business
... Come from academia, business etc.
... This Wednesday we have a lot coming from outside NL
... There's a EURO SDR workshop as well as us so that's increasing interest

eparsons: Thayt's more cadstral, mapping agencies, IGN etc.

Linda: The theme of the meeting is Spatial Data on the Web. More from Geo world than normal. 160 registered

eparsons: Have we thought about how we want to do this? Picking some UCs that we think might be useful cf. asking for ideas for UCs?

jtandy: Linda asked me to give a 10 minute talk on what we're doing with a focus on the BP doc. The people working on the Geonovum tender will be there too.
... All I'd mention in terms of UCs, the 30 BPs we have so far are underpinned by multiple UCs.
... We have a parallel threada. If people want to engage with this group, then the // thread is the way to do that.

Linda: So not all 160 people will be in the workshop - which is good as the room isn't that big.
... So we'll have about 1.5 hours to engage with those folks.

jtandy: I'm explicitly asking them... we need you if we're to provide something useful to you.
... So really asking for help with what they need help with.

(scribe paraphrase)

roblemmens: Maybe a short walk through the document?

jtandy: We could do that in the // track?

eparsons: This is about doing a santity check on the work we've done, talking to customers.
... I would hope that there won't be loads of new use cases popping up. WE perhaps shouldn't fish for new UCs.

frans: If there are any, it's better to have them ASAP.

eparsons: Yes, but I don't want to put UCs front and centre. The view into the use cases is probably going to be the BP doc. If there's something misisng, let us know.

jtandy: if there's something grossly missing, tell us the use case and cite the real world case where this is important.
... We may be able to amend an existing UC to include that.
... But we can't take pet projects.

-> http://www.pilod.nl/wiki/Geodata_on_The_Web_Event_10_February_2016 Workshop website

eparsons: We've done a lot of homework gathering these UCs. This isn't our pet project - we have evidence that these things are important.

jtandy: I'll give a short plenary talk at the start, and then in our // track we can have the full conversation.

frans: I agree with not fishing for new use cases, but we can seek new requirements

jtandy: And they need to be hooked on to an existing UC.

Time

ChrisLittle: We've talked about this before. Time is not specifically spatial - it's more fundamental than that.
... There's a lot of confusion and mix up of time and dates.

eparsons: So what is the real problem.

ChrisLittle: The real problem is the confusion between precison and accuracy
... ISO 8601 gives you a false sense...
... It's the same as the confusion with WGS84.
... If you don't care about precision to the nearest minute, then OK.

jtandy: many communities ignore what other people want.
... It's when we look in from the outside, it doesn't necessarily line up.

eparsons: We have a bunch of expert communities who know more than we do.

jtandy: Our biggest problem is that there is no formal way of writing down time as an ontology.
... If Simon were here, he'd say there is no way to talk about non-Gregorian calendars, or indeed ones that include leap seconds.
... cretaceous period etc. Lots of things it doesn't do.
... Our charter asks us to finish the work off. It matters when something is where.

<AndreaPerego> s/peroiod/period/

frans: To respond to Chris's remark on precision and accuracy - it's a general data thing, not spatial.

<rachel> s/whewn/when/

frans: Maybe we can ask the DWBP about this.
... Maybe we should make a list of things to talk to DWBP about on the call next week (17/2)

eparsons: anyone else?

aharth: I think it's a problem that we don't have an onology for time - but our use cases don't say that. An ontology is a long way from the UCs.
... ISO 8601 gets you a long way. How big is the community that need non-Gregorian time?

<eparsons> action ChrisLittle "bring up time issues with DWG on next weeks telecon"

<trackbot> Created ACTION-137 - "bring up time issues with dwg on next weeks telecon" [on Chris Little - due 2016-02-15].

ChrisLittle: The problem is that ISO8601 allows accuracy but it's not precise. It doesn't include leap seconds so GPS time is currentely 30 adrift from the real time.

aharth: So who cares?

ChrisLittle: The defence community thinks it's a big problem. Just as people think of bombing the wrong place, you can be equally inaccurate about when.
... We're committed to leap seconds until at least 2022. So we'll move a few more seconds adrift from real time.

<Zakim> jtandy, you wanted to talk about archaeology

ChrisLittle: There are use cases of people trying to extract meteorological info from old ships logs. If you write down gthat info, which calendar were they using?

<rachel> s/sheeps/ships/

jtandy: We have use cases from archaeological and cultural history
... And geologists who want to talk agbout the anthropocene

aharth: Is unconvinced. I don't think the community is a big one. If we look at things like SSN...

ChrisLittle: becomes animated
... Let's talk about the IoT people - a growing community. Each of those things will have a time on them. But where does that time come from.
... Different computers may have different times. They need to be synchronised.

eparsons: It works now, doesn't it?

ChrisLittle: Not really, it might.

aharth: It's about raising a flag about the difference between accuracy and precision.

<Zakim> phila, you wanted to ask about internet time

eparsons: Internet time generally runs of GPS time
... We've signed up to finishing the ontology about time. We just need to make sure it fits within the community we're serving.

ChrisLittle: My concern is that as we fix the OWl ontology, we need to go beyond spatial context.

roblemmens: A more prinicpalled problem - often there is no time label at all, when sometimes you need it.
... You find the spatial data but you don't know when it was true.
... But that has nothing to do with the ontology as such.

frans: It's common practice to use XML schema datatypes
... Most standards seem to recommend xsd:dateTime which goes to a time
... So false info is created

roblemmens: That's why time labels are often missing

frans: The other thing I wanted to say... some communities don't have a big Web presence at the moment. We shouldn't only look at current Web presence - some may be waiting for us to help them.
... So we should look at some who are not really on the Web yet. people dealing with professional data, gov data, historical data.

kerry: I wanted to say ... this is something I've been troubled with. What does SSN look like? Yes, it's an ontology. But we need to talk about how to use it.
... I don't think I'd include that in the BP doc.
... It's within scope if we have the resources.

<ChrisLittle> +1 to Kerry's comments

<eparsons> +1 to kerry

kerry: So I think we need to talk about time and how to handle it on the Web.

<kerry> https://www.w3.org/2015/spatial/wiki/Time_Wish_List

kerry: Modulo a check over the UCs, I think this pretty much covers things on the e-mail list.
... Maybe that's enough?
... I'd like us to talk about how to use this.

eparsons: What's you sense, ChrisLittle, if the work is completed - how much does that solve where the issues are and how much does the BP doc need to give a narrative.

ChrisLittle: I think fixes to the ontology will help encourage better practice.
... People use 8601 time when they don't really mean it because that's what the javascript libraray offers.

<Zakim> jtandy, you wanted to say that BP editors intend to use examples to illustrate how time should be used properly

jtandy: BP editors... section 6.2.3 which talks about temporal aspects of data that currently has no BPs in it.
... What we thought we'd do, rather than talk about that, was to use examples of how we write spatial data that had temporal aspects.
... I'm sure there's more work to do there to cover archaeology etc.
... That's where I'd like to show people how to do it properly.

eparsons: But there's still that wider requirement around spatiotemporal. The time ontology has a wider scope than ours.

jtandy: I think the work has more or less been done. Simon has written his proposal for extending the time ontology to cover non-Gregorian stuff.
... I'd take what Simon has done and see if it meets our needs

ChrisLittle: I think we need some BP exemplars, saying don't do this etc.
... Just as we do on precision and accuraxcy on CRSs

phila: Can Simon publish what he's done as a ReSpec document?

BartvanLeeuwen: Since we said a long time ago that we'd talk about best practice, not best theory.
... How do you say that this building was built in 1980 in LD? There's no BP for that yet.

eparsons: To be brutal, our charter says we'll do the time ontology. We haven't said we'll give BPs on how to use it.
... We're all going to be time pushed this year.

kerry: I was going to say, in terms of Time Ontology influencing BPs, there's plenty out there already.

<kerry> https://www.w3.org/2015/spatial/wiki/Time_Wish_List

kerry: There's a lot on that list already

<Zakim> phila, you wanted to say BPs for time not needed

phila: Talks about diff between implementation experience - needed for the Recs - and BPs, which will come later.
... Explains kind of evidence needed to get to rec for an ontology. Each term used in data >= 2 times

(scribe missed aharth's point, because it's lunch time)

<Zakim> AndreaPerego, you wanted to ask about time URIs and the use of Time ontology by reference.data.gov.uk

AndreaPerego: We haven't talked about the UK time URIs. You get back time ontology data
... Do we think they're good?
... Also I'd be interested in experience of using them.

billroberts: We have used those time interval URIs in data quite a lot. There is the question of what is the point of doing that rather than using a literal.
... Not all the would be literals are convenient

<eparsons> @kerry might finish a few mins early

billroberts: and it allows you to have a time as the subject.
... The downside is that you can't always take advantage of some of the SPARQL features for ordering
... although the URIs can be ordered.

AndreaPerego: I think Ian Davis did some work on this and geometries for time and space
... I think maybe you, Bill, made comments about having URIs for geometries in different formats
... I think these topics are very internlinked.

<aharth> phila: my point was that OWL Time depends heavily on OWL 2 DL, but I don't know whether reasoners exist that support the translation from xsd:date to OWL Time and back

ChrisLittle: When you have geometries in space, you can recognise difference based on CRS. Most time formats look similar and people assume that there is no transformation but that's not right.

billroberts: I need to understand these differences.

eparsons: Straw man - we need to get the ontology to a point where it becomes a Rec later this year. To cover these other aspects about best practice, not BP, maybe we need to have key references in the BP doc.
... We don't have other mechanisms to get things out.

akc me

<Zakim> phila, you wanted to talk about cross links

<jtandy> +1 ... agree with eparsons

phila: Talks about linking from the BP doc to all 3 of our other docs

<Linda> +1 to having key references to time issues in the BP doc

eparsons: We all need food

== Lunch ==

<eparsons> Returning from Lunch...

<BartvanLeeuwen> scribe: BartvanLeeuwen

<scribe> scribenick: BartvanLeeuwen

Best Practices

jtandy: pointers to real examples in the wild
... we need to provide real examples, examples need to be informative, short
... it needs massaging to get them compact enough to get into our document

<ChrisLittle> * no idea why I keep timiing out (sic)

jtandy: the working group asked us for examples from the BP's
... we need you ( The WG ) to give us these examples

eparsons: I hunted long and hard to use schema.org to have a best practice around that

<eparsons> http://queue.acm.org/detail.cfm?id=2857276

eparsons: I came up with this article
... it was the best place I could find to show what schema.org can do, and described in the best way

jtandy: which best practices does this apply to

eparsons: that is the issue, there are multiple, discoverabilty / findabilty
... will we extract multiple best practices, or one document which reference multiple BP's

jtandy: this is one of the challenges of the document, we might end up with 120 examples
... in the BP's themselves we try to pull out snippets of the examples, which show you how to do small things
... have the full examples reference the BP's which contain the snippets.
... we also might want to reference external resources where things are working together
... do we want something like jsfiddle to let people play with examples.

eparsons: how much of examples do you want to show it is real, and how much do you want the examples to be usefull

jtandy: the first is about 'this is practice, not theory', the examples in the document are illustrative
... we might say this illustrative example is composed of a certain example in the wild

frans: there is case to be made to create your own examples, examples in the wild my disappear. We need small concise examples where examples in the wild my be containing multiple BP's

jtandy: this is a good point, its similar to what I was suggesting.
... we need ilustrative examples that we 'own' but they should point to examples in the wild

<rachel> +1 to illustrative easy to read examples (patterns) and real uses in the wild

eparsons: we use small snippets to explain how to implement a BP, but also point to real world examples

jtandy: we might want to have fixed set of examples which hang together, which we could publish as a appendix

eparsons: we need to think about a sample. e.g. acme.org which wants to publish data

BartvanLeeuwen: it needs the BP's to be reordered

<billroberts> BartvanLeeuwen makes a presentation on work with Dutch fire department

<billroberts> ...supplement a WFS with additional information about each feature

<billroberts> ...help the user understand the meaning/background of the geo features they are looking at

<billroberts> ...Bart added a new feature name, to connect the feature to the thing it is about

<billroberts> ...(rdf:about in Bart's example, but that choice is up for discussion)

<billroberts> frans: there are mechanisms already in WFS to use XML schema to do this

<billroberts> eparsons: we should agree a standard way to do this. Which way is for us to discuss

<billroberts> BartvanLeeuwen: also wants a backlink. If you know the thing of interest, how do you find a WFS URL for that thing

<billroberts> eparsons: most valuable thing may be simply to put the extra field into the standard WFS response

<billroberts> BartvanLeeuwen: it's hard for the fire department users to figure this out in the middle of the night. Need to make it standardised and hence easier

<billroberts> jtandy: for the BP, a tangible example, such as emergency response, could make a good consistent thread that we could use in several examples, to help make the BP doc more coherent and easier to understand

<billroberts> BartvanLeeuwen: the Dutch fire department would definitely be interested and an emergency response example makes it particularly relevant for them

<billroberts> eparsons: which of the best practices would Bart's example be appropriate for?

<billroberts> agreement that there are several relevant BPs for this example

<billroberts> stanislav: there is a lot of relevant data around, eg from cadastral services

<billroberts> jtandy: challenge for users of WFS is that in many cases the only info a user can get is from the layer names

<billroberts> eparsons: one could argue that metadata catalogues would help with that problem

<billroberts> eparsons: nice idea, but need more than just Bart to work on it, because of volume of work

<billroberts> frans: other gruops are working on similar things: how to modify WFS to be more 'webby'

<billroberts> [BartvanLeeuwen highlights a diagram in the presentation]

<eparsons> @kerry may have lets check...

<billroberts> BartvanLeeuwen: system uses Geoserver to do WFS plus PostGIS with R2RML to create RDF, with the RDF connected in a 1:1 mapping with the WFS

<billroberts> stanislav: there is software LOD4WFS that can help in creating new RDF for publishing via WFS

<billroberts> BartvanLeeuwen: how can I create a triple which has a WFS URL as object? to relate other data to the relevant WFS

<billroberts> Frans: why do this?

<billroberts> BartvanLeeuwen: to support various user communities as well as we can

<billroberts> jtandy: work in Australia (didn't catch who?) on publishing identifiers for places and links to info about them

<eparsons> Aussie chap Rob Atkinson

<billroberts> jtandy: value for this group is to bring together the RDF and WFS communities

<ClausStadler> (hi, what's the webex meeting password?)

<ClausStadler> ah nm

<billroberts> eparsons: if we can't point to existing best practices, can present our preferred option (even if not yet well established)

<billroberts> eparsons: not so helpful if we say, well you could do one of 4 things. Better to be specifci

<billroberts> BartvanLeeuwen: need enough people to start implementing the extra column in WFS

<billroberts> eparsons: could propose this to OGC for the WFS standard (or a profile for WFS)

<billroberts> BartvanLeeuwen: could probably avoid changing the WFS spec by using a profile

<billroberts> jtandy: if phila were here, he'd remind us that BPs should be durable

<billroberts> jtandy: could be more than one way to achieve something (to reflect changing fashions and practices) while still following the underlying, persistent BP

<billroberts> eparsons: but make document usable by practitioners so not loads of different choices

<billroberts> frans: WFS is not hte only kind of spatial data service we might have to consider

<billroberts> jtandy: so, emergency response looks good for a common narrative through many examples in BP doc

<billroberts> jtandy: how best to structure doc to make it easy to add value. How can we help spatial data publishers to add value without having to do very complicated things. What are the first/easiest things?

<billroberts> jtandy: step 1 might be: get people to use HTTP URIs in their data

<billroberts> jtandy: look for things that are profound but don't take much effort

<billroberts> eparsons: different approaches might be appropriate for (a) 'green field' data publisher starting from nothing or (b) someone with a well-established set up that needs modification

<billroberts> frans: examples could start simple (tweet about a fire) but bring in lots of detail later

<billroberts> BartvanLeeuwen: flooding might be better example than fire for some BPs

<billroberts> eparsons: if you didn't have GIS adn WFS 'baggage' where would you start

<billroberts> BartvanLeeuwen: good question, but people always want to see stuff on a map so GIS may be a good place to start

<billroberts> eparsons: could just have an HTML page per record - simple, discoverable. May not need fancy GIS functionality

<billroberts> eparsons: if I haven't got a GIS, do I need one? (for simple examples)

<billroberts> BartvanLeeuwen: in a flood warning, someone wants to know where to go to a safe place - doesn't want to start a GIS to get that

<billroberts> roblemmens: time also relevant

<billroberts> stanislav: in a complex emergency situation, improtant to have access to all relevant data in one place

<billroberts> eparsons: emergency services 99.9% likely to have a GIS infrastructure

<billroberts> ChrisLittle: these organisations may be in their own data silo but probably shouldn't be. Better to also be looking at eg social media and other external data sources to get whole picture

<billroberts> eparsons: eg need to find all Starbucks stores (say) in response to a potential threat. Starbucks probably have a GIS but it's not accessible quickly to emergency services, whereas their store locator web page probably is accessible

<billroberts> roblemmens: emergency services not the only users, also perhaps reporters/journalists

<billroberts> eparsons: so we could recommend to companies to publish machine readable store locator info on their websites (whether or not they also have a GIS)

<billroberts> jtandy: BBC uses linked data for underlying data infrastructure eg for sporting events. Could potentially be combined with other emergency response data

<billroberts> BartvanLeeuwen: even if you can see a GIS you may not have enough info to interpret correctly. (eg opaque short codes for layers)

<billroberts> BartvanLeeuwen: also multilingual issues sometimes - eg between Germany and the Netherlands in river flooding situations

<billroberts> eparsons: so let's work with our current list of BPs and see if we need to restructure to allow us to use flooding as a narrative

<billroberts> jtandy: also useful as a way to bring in Coverage data

<billroberts> jtandy: as remote sensing very useful in flood situations

<billroberts> eparsons: should also include fire based examples - Australian bush fires also brings some broader interest

<billroberts> BartvanLeeuwen: flooding tends to be more interdisciplinary than fires,hence more interesting data exchange problems

<billroberts> jtandy: what about examples of existing 'in the wild' best practices

<billroberts> eparsons: should go through BPs and ask: who has a good example?

<billroberts> jtandy: if you don't volunteer something, we'll have to get our dentist's pliers out

<billroberts> jtandy: anyone got prepared examples relevant to specific BPs?

<billroberts> jtandy: who has done their homework?

<billroberts> BartvanLeeuwen: BP22. DBpedia uses Geonames

<LarsG> BP 22: http://w3c.github.io/sdw/bp/#link-to-auth-identifiers

<billroberts> ClementsPortele: (audio hard to follow but...) he is using Geonovum testbed. Experiment to supplement existing WFS infrastrucure to be more webby

<billroberts> ClemensPortele: has x-ref'd that with the BPs.

<billroberts> ClemensPortele: offers to share more detailed information about htat, or could present during the meeting if required

<billroberts> eparsons: please send a link to online info

<billroberts> jtandy: (1) sounds great, thanks (2) could you show it on webex, or is that practically difficult

<billroberts> ClemensPortele: yes, could show on webex

<billroberts> webex being screen transferred to ClemensPortele...

<billroberts> [ClemensPortele presents from his screen]

<billroberts> ClemensPortele: data from the Netherlands, including kadaster

<billroberts> ClemensPortele: based on existing SDI in NL

<billroberts> ClemensPortele: make a new 'layer' to access the data in a more webby way

<billroberts> ClemensPortele: using schema.org, using HTML and JSON-LD, supporting content negotiation, allow all data to be indexed

<billroberts> ClemensPortele: [shows list of BPs - which have been implemented?]

<billroberts> ClemensPortele is currently writing this up, report available around end of month and will share it when ready, so we can read more about which aspects of BPs have been implemented

<billroberts> ClemensPortele: thinking about providing a Swagger description for the API, but not sure when taht will be done

<billroberts> ClemensPortele: not implementable BPs: reusing authoritative identifiers...

<billroberts> and other not implementable (for this work) BPs listed

<billroberts> ClemensPortele: experience of trying to x-ref to BPs has revealed some cases where revising the BPs might make them easy to work with or test

<billroberts> ClemensPortele: eg need metadata in both DCAT and schema.org - content negotiation not sufficient to distinguish these two

<billroberts> ClemensPortele: struggling to implement some best practices.

<billroberts> ClemensPortele: Agricultural data example. Shows links from RDF data to the WFS resources

<billroberts> eparsons: thanks Clemens for v useful presentation

<billroberts> jtandy: great piece of work, thanks. Thanks also to Geonovum for the testbed. As well as stuff in the BP doc, can we also have interactive resources, such as Clemens' demonstrated systems?

<billroberts> frans: can W3C host persistent example services. Question for phila when he's back

<billroberts> eparsons: to be pragmatic, treat persistent testbed as a nice to have. Often practical difficulties about ongoing funding for hosting

<billroberts> ClemensPortele will post a link in IRC

<billroberts> ChrisLittle: Clemens, how confident are you of categorisation of BPs as implementable/not-implementable. Was it difficult to decide?

<billroberts> ClemensPortele: some of those categorisations of BPs need some detail/qualification

<billroberts> jtandy: useful to know if the BP is clear as a standalone text. Asks Clemens for further insights as we go on

<billroberts> eparsons: next item, scoping issues

<Zakim> rachel, you wanted to mention examples of named epochs and events for BP1

<billroberts> rachel: has done some work on BP1, names of epochs etc

<billroberts> rachel: Linked Open Data Finland has useful examples of referring to events and places

<rachel> example

<scribe> scribe: BartvanLeeuwen

<scribe> scribenick: BartvanLeeuwen

<rachel> http://www.ldf.fi/ Linked Open Data Finland

<rachel> e.g. Event: Atrocities in Namur http://ldf.fi/ww1lod/d75dce08

<rachel> Geological epoch http://resource.geosciml.org/classifier/ics/ischart/Jurassic

<rachel> Fosse Way Roman road http://pleiades.stoa.org/places/656280677

<ClemensPortele> Link to my slides: https://github.com/geo4web-testbed/topic4-general/blob/master/geo4web-topic4-sdwbp-20160208.pdf

<rachel> Namur http://vocab.getty.edu/tgn/7007960

<rachel> Getty Iconography Authority http://vocab.getty.edu

<rachel> Getty Iconography Authority (cultural objects including historical events) a module of Cultural Objects Name Authority (CONA)

<LarsG> scribe: Lars

<LarsG> scribenick: LarsG

Stanislav: presents a spatio-temporal content explorer
... can navigate from a class of objects and show instances on the map
... (all in one screen)
... result filtering features
... particularly limiting the number of results (for SPARQL endpoint reasons)
... then navigate through relations (properties) and then filter again (e. g addressable objects) and connect to house numbers
... backend is GeoSparql
... can also filter on time

phila: Do you filter the subset or all the data on the same time?

Stanislav: it's a subset already
... Navigating data (e. g. buildings) filtering by temporal extent
... showing a timeline when buildiings were erected

ChrisLittle: highlights how important it is to tell what calendar was used

roblemmens: This example shows the value of linking data

eparsons: This way linked data can add back to geospatial

phila: billroberts does this kind of visualisation for a living hiding triples from the end users

Stanislav: in geospatial the human interfaces are extra helpful (easier to show it on a map than to explain)
... uses Parliament (only triple store that supports full geosparql)

roblemmens: usability aspects are still to be solved
... UI hard to grasp for inexperienced users

jtandy: With geosparql you are constrained to geosparql vocabulary

Stanislav: Yes, we use WKT for geometries

ChrisLittle: particular case of presenting spatio-temporal data

aharth: has played with web data and one important point is how to visualise people on a map
... datasets differ on how they attach lat/long to objects. DBPedia attaches directly to the location, others do it differently
... sometimes you have to go through the attached geometry

phila: foaf:Person is subclass of wgs84:SpatialObject

kerry: how would jtandy want people to contribute to the BP? One wiki page per BP?

jtandy: hasn't thought about that yet
... but acknowledges the need for a place to share content. Wiki page sounds good

eparsons: it needs to be directed. Start with empty page and then fill iin during the calls.

Linda: One BP per time?

eparsons: yes. Perhaps a separate telecon if necessary

kerry: Prepare a wiki page _before_ the meeting would be ideal

frans: one page per BP or one per example?

<rachel> +q

frans: and link that to the BPs

jtandy: better to start with one big page

rachel: +1 to jtandy

eparsons: should suit the needs of the editors
... that would naturally work its way through

Scoping

jtandy: much in BP is about exposing data through web services
... all questions were asked through the lense of exposing feature services
... but it's all appliable to other kinds of data, too
... so it might be out of scope but otoh it needs to be in the document for the geospatial community to understand that it's important

eparsons: So are we OK with publishing some BPs that aren't particular to _geospatial_ data? eparsons is OK with that
... if we end up with recommendation à la "publish geospatial data the same way you'd publish other data" that is OK

<AndreaPerego> +1

frans: anxious we don't have time to deal with hard geospatial problems (CRS)
... is the point that we would want to point to BPs yet to write by other groups?

jtandy: DWBP works on a different level, so they might not solve our problems

phila: If it's not in DWBP now, it won't get there, so if SDW has needs, we must put it into our own document

AndreaPerego: We must provide examples of how we want our BPs to be done, even if it overlaps with other BPs
... sometimes we just extend things (e. g. GeoDCAT as an extension to DCAT without disturbing existing infrastructure)
... it's about showing people how to get linkable data within their own infrastructure even if that means giving general advice

Linda: ClemensPortele has commented on this

ClemensPortele: comments were specifically on out-of-scope for _our_ implementation, not general comments

<Zakim> phila, you wanted to talk about https://www.w3.org/TR/dwbp/#StructuralMetadata

phila: points to DWBP example

<phila> Structured

phila: example talks about location (use of dcterms:description is incorrect)
... we need to give guidance

frans: haven't seen anything about precision, significant digits etc. Worried we might have too much work to do and not have time to focus on geospatial
... we should focus on those.
... it's a matter of priorities

eparsons: but we need to build on general foundations on how to publish data. What do we percieve to be missing from DWBP?

frans: there are BPs out there not just described in DWBP
... e. g. web documents on how to publish linked data, mint URIs etc. Probably also use of significant digits. We need to find those.

jtandy: many haven't read common practice because it's not particularly about spatial data. We need to corral those and put all the BPs in one place
... even if this means stating well-known info about e. g APIs

eparsons: +1. How much do we need to do before talking about spatial data?
... Is it a matter of work to do or just work we need to collect?
... Let's do what we have to do and then the missing pieces will emerge.
... be it minting URIs or whatever. Let's not panic yet.

jtandy: other scoping question: sensor data
... many say this doesn't fit with spatial data

<kerry> +q

jtandy: How much do we want to pick up sensor data as a thing vs spatio-temporal aspects of sensor data?

kerry: doesn't have an opinion. SSN ontology is important but that's not a BP document and perhaps we don't need that (nor as a part of another BP doc)

<jtandy> http://w3c.github.io/sdw/bp/#describe-process

jtandy: BP 15 http://w3c.github.io/sdw/bp/#describe-process is very specific, describing a data stream, but not specific to sensor data

eparsons: what is a sensor, someone with a twitter account (from telecon last week)
... essentially everything is a sensor

kerry: BP 15 is also about provenance etc so it's not sensor data specific
... crowd sourced data is similar

jtandy: is nervous about scope re. consuming sensor data

kerry: perhaps summarise it all in a single BP ("use ssn ontology for sensor data")

eparsons: perhaps we shouldn't make BPs for sensors and time until we have the ontologies to describe them

kerry: those ontologies should make it into the BP doc

phila: The BPs came out of use cases and if we strip them out of BP we're ignoring needs coming from UCs.
... Those are needs from geospatial and web community

jtandy: many UCs came as input from SSN deliverable, not BP

ChrisLittle: The ontologies sit on microformats and feed into the ontologies. We cannot do one without the other

eparsons: We have problems both in temporal aspects and the fact that much will come from sensors
... We need to extract the problems from possible solutions (such as SSN and time ontologies)
... We can say that there are problems with particular technologies without having to provide a solution to them
... perhaps we need to rephrase some BPs not to cite other deliverables

jtandy: In emergency response there is much sensor data
... so instead of saying exactly how it works we just use them as examples (e. g. the location of a smoke detector) and from there convey the connection to the SSN ontology

frans: it's good to link between the deliverables

BartvanLeeuwen: No problem to sneak sensors into the emergency rescue narrative is not a problem

billroberts: and sensors can move around, tooo

eparsons: time and ssn are both valuable but they might not be BP (yet)
... if it's not mature we cannot really say it's a best practice

phila: BP is about "providing sensor data in a structured way; have a look at ..."

jtandy: (recapping)
... use web services to publish spatial data. This was an inhibitor.
... Much future data will be sensor data, but we'll not say how to publish this data unless it's particular to spatial (e. g. an air quality monitor on a bus)
... or how to deal with streaming data. This is appropriate to sensor but not particular

<rachel> [river gauges as sensor input to flooding emergency response http://www.gaugemap.co.uk/ using http://www.shoothill.com/environment-agency-liveapi/]

jtandy: Wild fire observations can come from humans. This is more about semi-structured information than about sensors

eparsons: "If you want to deep dive, have a look at ..."
... and point people to upcoming work (including CRS, non-Gregorian calendars etc.)
... but we have to provide some pointers on how not to make certain mistakes

frans: Important question is how to put geometry on the web. This is not only sensors but everywhere
... BP right now is to do it in a fashion that fits your need, so we might need to develop our own ontology. This we need to address

eparsons: do you talk about a simple features ontology or something else?

frans: not only for buildings or geographical data, but also microscopes and geometry in general (i. e. on the web)

jtandy: What's wrong with the current five ways?

frans: the problem is that there are five and those are much more about geographical than about geometry. There are too many ways to express time and space
... We need one way to rule them all

AndreaPerego_: there are reasons that there are more than one
... applications are built to consume different kinds of data
... services should ideally serve data in different ways (GML, WKT, ...)

<phila> The Core location Vocabulary

AndreaPerego_: we have recommendation on how to represent geometries (GML, WKT, GeoJSON). It depends on the use case which one you pick.

<phila> XKCD 927, the one about having too many standards

jtandy: sometimes it's too complicated to do simple things in the "best" vocabulary

frans: we shouldn't say it's impossible to come up with one single model. We should attempt to find an interoperable way.

jtandy: Do you want a conceptual model (we've got one) or do you talk about implementations.
... Should we say "this is the only one we recommend"? That way we can alienate parts of the community

aharth: There are differences in conceptual model. WGS84 doesn't differentiate between thing and geometry. It would be nice to describe the differences between them and align that with the web data model

Linda: it would help people if we can order the existing models according to ease of implementation ("but if you want polygons, you must use X")

AndreaPerego_: Like having a decision matrix

phila: OGC is happy to do GeoSparql 1.1 if necessary
... it's not helpful to define a sixth vocabulary

ClausStadler: there is a difference between how data providers work and how tool implementers work.

billroberts: If we say there is only one way, we'll be ignored. Instead we should give advice on how to document our decisions (vocabulary, use of CRS) and how to convert between them.

<Zakim> phila, you wanted to say the thing I forgot to say about addresses

billroberts: If there are gaps we should close them

phila: Addresses are important. vCard is well implemented and is not INSPIRE compliant. We should give advice on how to publish an address

<billroberts> +1 to importance of addresses

ChrisLittle: All this fits into precision and accuracy. Not only "where is the house" but also "which door". And how to do that transition

<eparsons> thanks rachel !

BartvanLeeuwen: We also need a remark on who we expect to consume our data. People often want to publish data because they have it but their customers don't have the tools to consume it
... e. g. no support of arcs in OpenLayers

<BartvanLeeuwen> +1 to addresses

frans: no consumer group would be happy with five or six ways to do the same thing since they will need to implement support for all of those
... that's too much if-then. They will want one spatial ontology
... we should try to make progress here

eparsons: we can look at how to close gaps but shouldn't spend too much time

frans: our extensions should be compatible with all existing (a sort of super-model)

billroberts: much of data consumption is driven by existing tools. So often we need to publish using several formats/vocabularies

jtandy: points to BP 7 where this topic is addressed

frans: we should not start from geography but from mathematics when creating this ontology

<aharth> scribenick: aharth

<phila> scribe: aharth

LarsG: our discussion relates to Clemens' point earlier: we need to be able to support different representations/schemas of the same thing
... there's an Internet draft in the works regarding Accept-Schema
... client and server can negotiate the schema used

eparsons: how does it help if we talk about a very abstract geometry model first, leading to the final encoding

frans: we need an abstract model that is powerful

phila: how much effort would it be to come up with a table to help people make a decision regarding the vocabulary?

frans: geosparql could be the model

phila: most people might think of terms of addresses rather than geometries...

<jtandy> [@lars: see RFC 6906 from Eric Wilde - The 'profile' Link Relation Type]

frans: it would be helpful to converge to one vocabulary for spatial data

phila: we could get spatial relations into the link registry

billroberts: what are the most common use cases? addresses, geocoding, postcodes, points and polygons...
... but mostly polygons are not available
... then, especially in statistics, the next level of complexity is how polygons relate to each other
... there are many pragmatic problems around that are a big barrier

frans: how do you put a geometry into a triple store?

billroberts: some of the triple store technology is not advanced enough, use postgis, elasticsearch
... we would use more geosparql if there are implementations available

frans: geometry is just a datatype like numbers, string

aharth: i'd say a polygon is an object, rather than an image

<ClausStadler> base64 encoding - webpack optimizes html pages by inlining all images like that

ChrisLittle: numbers are not enough to encode geometries, need numbers + other data

<ClausStadler> what i wanted to say is that binary data could be encoded using e.g. base64, so in a way it would be possible to support it in rdf

billroberts: is it an object, is it a database-y type of thing? for most things we need both
... we are in a tricky situation where we need both

jtandy: i'm reminded of discussions of the JSON-LD space... people were not really fans of RDF triples
... maybe there's a way to use a triple data model without requiring much of the semantic web machinery

ChrisLittle: maybe link to an object, dismantling it does not necessarily have to be done using rdf tools

frans: there are functions in geosparql to do that

eparsons: we carry on discussing this lateron, let's come back to that
... next topic is the engagement with the DWBP

phila: presents the current version of the DWBP document
... the document has many navigational aids: challenges, benefits, used to classify the best practices
... only one best practice is open (related to APIs)
... now the thing left to do is to create examples and tests
... the DWBP document will be a Recommendation
... which means for each of the best practices there need to be two organisations that have made the same recommendation
... in other words, find two independent implementations of each best practice
... charter of DWBP runs out in july
... the goal is to provide a running example with actually testable data
... there's currently much effort to finalise the DWBP bp document

jtandy: in preparation for next week's call, what version of the document should we read?

phila: i would read the editor's draft
... for next week, we're flagging places where things related to spatial can be improved

frans: do they have public comments?

phila: we've got some comments, but would like to have more

<phila> data on the Web Best Practices Editors' Draft

jtandy: please, read through the document and find gaps relative to our work

AndreaPerego: there are other deliverables (a couple of vocabularies) that might be worth looking at

<phila> Data Quality Vocabulary

<phila> Dataset usage

phila: permissions and obligations is another topic that's coming up

<Zakim> AndreaPerego, you wanted to ask if DAQ / DUV will address also data granularity

phila: data shapes (SHACL) is a "schema"-language, there's recent progress there

AndreaPerego: what about granularity in the vocabs? might be relevant to us, for example the way to model spatial resolutions

phila: dcterms:conformsTo is a start
... candidate rec for the two vocabularies is targeted for march

jtandy: I wonder if the two vocabs provide examplars for how we should publish vocabs

phila: yes. a class diagram is usually the first thing people look at
... it's the same pattern that the other ontologies (DCAT...) use

jtandy: look to see if they handle location correctly, identify whether they cover our issues (otherwise we need to get the issues into our document)
... best practice: we have a room of implementation community with us on wednesday
... what are the questions we want to ask them?

<eparsons> action jtandy to email public list with homework re DWBP call next week

<trackbot> Created ACTION-138 - Email public list with homework re dwbp call next week [on Jeremy Tandy - due 2016-02-15].

jtandy: suggestions: tool chains/software architecture
... e.g., triplestores with sparql, elasticsearch, or something in between, HTML in vi...
... is that a useful question to ask?
... for the data publishing folks: what kind of spatial data are you trying to publish? geometries vs. points?
... for the data consumers: what is preventing you using spatial data in applications and decision making?

frans: i wonder what we do with the answers

jtandy: that might help us prioritise in which order to put different approaches/formats into the bp document
... which formats do you typically publish data in?

frans: ask for what people like to do, but find impossible or difficult to do now

jtandy: format one is interesting, because it gives a hint about the currently used toolchain
... provide examples from the real world
... how should we structure the bp doc for readers with short attention span

eparsons: we should use the event for sense-checking, not gathering new requirements

BartvanLeeuwen: maybe explain how we end up at the WG

jtandy: yes we will do that

BartvanLeeuwen: careful to not raise expectations we cannot fulfil

jtandy: "what do you want from this" would be another question

Linda: if people asked the public draft, are we going to ask for specific questions?

jtandy: will do
... our sensor data discussion might be worth mentioning

Linda: we have 1 hour 45 minutes, keep time in mind

eparsons: we're about done for the day

Linda: tomorrow we are at a different location, house number 2, in the meeting center, lunch and afternoon session at the same place as today

eparsons: thanks and see you