- From: Phil Archer <phila@w3.org>
- Date: Tue, 20 Sep 2016 08:39:01 +0100
- To: SDW WG Public List <public-sdw-wg@w3.org>
The minutes of yesterday's TPAC F2F are at
https://www.w3.org/2016/09/19-sdw-minutes, linked from the agenda and
pasted below as text.
[1]W3C
[1] http://www.w3.org/
Spatial Data on the Web WG F2F, TPAC 2016 Day 1
19 Sep 2016
[2]Agenda
[2] https://www.w3.org/2015/spatial/wiki/Meetings:F2F4
See also: [3]IRC log
[3] http://www.w3.org/2016/09/19-sdw-irc
Attendees
Present
AZ, AndreaPerego, BartvanLeeuwen, BernadetteLoscio,
ByronCinNZ, Chris_McGlinn, ClemensPortele, DanBri,
DanhLePhuoc, Linda, RaulGarciaCastro, ahaller2,
billroberts, dmckenzie, eparsons, fasr, frans,
hadleybeeman, jtandy, kerry, phila, newton
Regrets
Payam, Chris Little, Josh Lieberman
Chair
Ed
Scribe
Armin, phila, ahaller2, kerry, billroberts,
ClemensPortele, RaulGarciaCastro
Contents
* [4]Topics
1. [5]BP number 7
http://w3c.github.io/sdw/bp/#globally-unique-ids
2. [6]http://w3c.github.io/sdw/bp/#indexable-by-search-en
gines
3. [7]Best Practice 8: Provide geometries on the Web in a
usable way
4. [8]Indirect Identifiers
5. [9]Bart's demonstration of linking to WFS
6. [10]Are we meeting the needs of practitioners- if not,
how can we improve?
7. [11]"How do we kick the RDF habit?"
8. [12]Agenda for tomorrow
* [13]Summary of Action Items
* [14]Summary of Resolutions
__________________________________________________________
<eparsons> Morning all - We are working on getting the webex to
work with the room ... please wait than you
<AZ> hello
<trackbot> Date: 19 September 2016
<kerry> scribe: Armin
<kerry> scribenick: ahaller2
<AZ> AZ = Antoine Zimmermann
Ed: begins with a tour de table
<fasr> fasr = Francisco Regateiro (Lisbon University)
<scribe> scribe: ahaller2
<scribe> scribeNick: ahaller2
<eparsons>
[15]https://www.w3.org/2015/spatial/wiki/Meetings:F2F4
[15] https://www.w3.org/2015/spatial/wiki/Meetings:F2F4
eparsons: going through agenda
<eparsons> [16]https://www.w3.org/2015/spatial/wiki/Patent_Call
[16] https://www.w3.org/2015/spatial/wiki/Patent_Call
eparsons: patent call
<frans> overview of recent changes in the UC&R document:
[17]http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirement
s.html#SPWD-TPWD
[17]
http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirements.html#SPWD-TPWD
frans: UC&R introduction
... some changes to requirements have been made
<frans> [18]https://www.w3.org/2015/spatial/track/products/1
[18] https://www.w3.org/2015/spatial/track/products/1
frans: 6 possible upcoming changes
... other subgroups should be aware of these new requirements
eparsons: are there any requirements we can close now?
frans: more that people are aware of them
kerry: let's work through the list of requirements
frans: issue-75
<phila> issue-75?
<trackbot> issue-75 -- quality metadata out of scope? -- open
<trackbot> [19]http://www.w3.org/2015/spatial/track/issues/75
[19] http://www.w3.org/2015/spatial/track/issues/75
frans: quality for each set of data point may be needed
kerry: quality indicators can be modelled in SSN
<Zakim> jtandy, you wanted to ask if the requirement is a
little more generic ...
kerry: SSN will be linked to Coverage which is one requirement
... in my opinion, it is out of scope here
jtandy: data acquisition may be another use case where you
attach metadata to a data point
<Zakim> phila, you wanted to talk about DQV
jtandy: i have not seen a common practice how to attach
metadata to the data point
... for example, someone says the flood water has come to their
house. it would be useful to see metadata attached to that.
... attach metadata to a dataset record
kerry: it is a pattern, not an ontological requirement
phila: I don't think it is out of scope
... we don't have to do it ourselves, other working groups have
done that
<phila> [20]DQV
[20] https://www.w3.org/TR/vocab-dqv/
phila: it is in scope, it is important to talk about accuracy
and precision. It is important for the editors for those
deliverables to include examples how to use it.
<Zakim> ericP, you wanted to say that clinical sometimes
relegate uncertainty to extensibility mechanisms have lots
include stuff like at-least or at-most in the core
billrobe_: identify individual data points might be a
requirement too.
ericP: clinical domain, where it was very important to talk
about quality of data
eparsons: we could close issue-75 by saying it is in scope
frans: do we have a crowd-sourcing use case?
... use case to attach metadata not for datasets, but for
individual data points
RESOLUTION: Issue-75 is in scope, close issue-75.
<BartvanLeeuwen> +1
<eparsons> +1
frans: make it a requirement for best practices, potentially
also for SSN
+1
frans: Issue-70
<phila> close issue-75
<trackbot> Closed issue-75.
<phila> issue-70?
<trackbot> issue-70 -- Add a requirement for avoiding
coordinate transformations? -- open
<trackbot> [21]http://www.w3.org/2015/spatial/track/issues/70
[21] http://www.w3.org/2015/spatial/track/issues/70
[22]https://www.w3.org/2015/spatial/track/issues/70
[22] https://www.w3.org/2015/spatial/track/issues/70
frans: I don't think anyone is opposed to this requirement
eparsons: can we talk about the wording
... putting a huge effort on the publishers, realistically we
don't project data to multiple coordinate reference systems
ByronCinNZ: default CRS should be the fall back
<frans> proposed requirement: *Requirement: *Data consumers
should be helped in avoiding coordinate
<frans> transformations when spatial data from multiple sources
are combined.
billrobe_: in practice we talk about 2 CRS covering 99% of the
use cases
<Zakim> jtandy, you wanted to suggest that we're missing a best
practice for publishing in multiple formats and
representations?
billrobe_: the publisher taking work of the user is a good
thing
jtandy: if you publish in the CRS of your national agencies
requirement, look at your audience and decide if you can
publish in multiple CRSs, e.g. your national and mercator
... if you can afford it, do it
eparsons: finesse the wording here. publish in multiple CRS to
meet your user requirements
... e.g. if you are in Britain, publish in British national
grid and another one
frans: we need to keep the requirement phrased as a
requirement, not to propose a solution
... one solution is to publish data using multiple CRS, but we
should not include that in the requirement
<phila> phila: Notes BWBP 14
[23]https://www.w3.org/TR/dwbp/#MultipleFormats
[23] https://www.w3.org/TR/dwbp/#MultipleFormats
<phila> phila: As many users as possible will be able to use
the data without first having to transform it into their
preferred format.
kerry: the publishers are our users as well. We should make the
life easy for both.
... how to we identify the CRS and is there a default CRS?
<Zakim> jtandy, you wanted to agree with Frans
jtandy: there is never to be a default CRS
... we can't make a reference to a coordinate without a CRS!
phila: best practice 14 from dwbp
... as many people as possible will be able to use the data
without transforming it
[24]https://www.w3.org/TR/2016/CR-dwbp-20160830/#MultipleFormat
s
[24] https://www.w3.org/TR/2016/CR-dwbp-20160830/#MultipleFormats
ericP: encouraged to publish in standard format is always a
good best practice
... if there is a publisher who does not use the standard
format, a third party can come along and do the transformation
eparsons: can we close that issue?
... pass it on to the best practices requirements
<phila> PROPOSED: Close issue-70 that we need to pass this on
to BP to handle, encouraging publishers' need to meet broadest
possible community
<AndreaPerego> +1
<eparsons> +1
RESOLUTION: Close issue-70 that we need to pass this on to BP
to handle, encouraging publishers' need to meet broadest
possible community
<ByronCinNZ> +1
<AZ> +1
+1
<phila> close issue-70
<trackbot> Closed issue-70.
<phila> issue-74?
<trackbot> issue-74 -- That uom and precision and accuracy
should be covered in ucr and bp (and respected in other
deliverables too) -- open
<trackbot> [25]http://www.w3.org/2015/spatial/track/issues/74
[25] http://www.w3.org/2015/spatial/track/issues/74
frans: issue-74
... current proposal is to have unit of measurements should
always be included in observations
<Zakim> phila, you wanted to dig into "unit of measurements
should always be included in observations"
phila: dwbp, uom should be included, but not in the string, but
as a separate attribute
danbri: in schema.org we have some workarounds
... space between $ and price, but it is not a best practice as
such
eparsons: wouldn't the uom not be part of the metadata of the
dataset
kerry: it is very particular kind of metadata. If you have
spatial data and you don't have the uom, it is useless. So it
is essential metadata.
<AndreaPerego> There's an example in DQV on the use of UoMs for
data accuracy and precision:
[26]https://www.w3.org/TR/vocab-dqv/#ExpressDatasetAccuracyPrec
ision
[26] https://www.w3.org/TR/vocab-dqv/#ExpressDatasetAccuracyPrecision
<phila> PROPOSED: That it is q requirement to have UoM included
<eparsons> +1
<BartvanLeeuwen> +1
<jtandy> +1
<Linda> +1
<billrobe_> +1
+1
<AndreaPerego> +1
<ByronCinNZ> +1
RESOLUTION: That it is a requirement to have UoM included,
close issue-74
<kerry> +1
<phila> close issue-74
<trackbot> Closed issue-74.
<frans> issue 74 proposal 1: We expand CRS definition
requirement a bit to make it clear that a CRS definition should
include an indication of UoM. For instance:
<frans> "There should be a recommended way of referencing a
Coordinate Reference System (CRS) with a HTTP URI, and to get
useful data about the CRS when that URI is dereferenced. The
CRS data should include the unit of measurement of the CRS."
<frans> issue 74 proposal 2: We add a new requirement:
<frans> "The use of precision that matches uncertainty in
coordinate data should be facilitated and encouraged"
issue-76?
<trackbot> issue-76 -- New requirement for multiple CRSs? --
open
<trackbot> [27]http://www.w3.org/2015/spatial/track/issues/76
[27] http://www.w3.org/2015/spatial/track/issues/76
<phila> issue-76?
<trackbot> issue-76 -- New requirement for multiple CRSs? --
open
<trackbot> [28]http://www.w3.org/2015/spatial/track/issues/76
[28] http://www.w3.org/2015/spatial/track/issues/76
jtandy: there is no requirement to have multiple CRSs, but data
should be accessible, which could be multiple CRSs
kerry: it is same as issue-70
<ericP> issue-70
<trackbot> issue-70 -- Add a requirement for avoiding
coordinate transformations? -- closed
<trackbot> [29]http://www.w3.org/2015/spatial/track/issues/70
[29] http://www.w3.org/2015/spatial/track/issues/70
jtandy: geojson does not support multiple representations of
the same data
... so you split data out in different representations
ericP: 70 says to minimise transformations, 76 is a solution to
that
... rephrase 76 to make it a requirement
frans: to publish multiple CRS can have different reasons, i.e.
to follow standards, not just to make it easier for users
<kerry> +1 to eric's comment -- make sure issue 70 is just
phrased right
eparsons: close 76 and it is 70, but rephrased
frans: publishing in multiple CRS is already a practice, but
they don't know what the best practice is
<phila> PROPOSED: Issue-76 is close enough to issue-70 that we
can cover it in the way we closed 70, although there is a
difference in perspective
eparsons: close enough, they will both be covered in best
practices document
<phila> PROPOSED: There are cases where there is a requirement
for more than one CRS, so we can close issue-76
<eparsons> +1
<RaulGarciaCastro> +1
<BartvanLeeuwen> +1
<jtandy> +1
+1
<ClemensPortele> +1
<ByronCinNZ> _1
<AndreaPerego> +1
<ByronCinNZ> +1
RESOLUTION: There are cases where there is a requirement for
more than one CRS, so we can close issue-76
<phila> close issue-76
<trackbot> Closed issue-76.
<eparsons> Coffee break back at 10:30 local - 15mins form now
<eparsons> Slowly returning...
<kerry> scribe: kerry
<scribe> scribenick: kerry
<RaulGarciaCastro> I’ll come back after lunch
<ByronCinNZ> Yes
<jtandy> BTW: the summary of SDW bps is here -
[30]http://w3c.github.io/sdw/bp/#bp-summary
[30] http://w3c.github.io/sdw/bp/#bp-summary
<jtandy> agenda for the BP session is here
[31]https://www.w3.org/2015/spatial/wiki/Meetings:F2F4#Monday
[31] https://www.w3.org/2015/spatial/wiki/Meetings:F2F4#Monday
jtandy: topic: best practice
... we don't have time to cover all 17 issues. please help to
priorities
... thankyou for all the expert opinion so far -- but tricky to
distil into workable bps
... linda nad I have bben working hard on this but it is harder
than we expected
... abd we have less time to devote going forward
... we can still do edit and style and steer and rank, but we
need WG members to own sections of the doc (a BP)
<Zakim> phila, you wanted to talk about engagement
jtandy: think of this as you can volunteer for a BP topic
phila: people read the doc getting close to publication , as i
have on the plane here
... see kadvice from bernadette
bernadette: one person read the whole doc at a very late point
and made a *lot* of comments which were helpful but took a long
time to process
... it is really important if people read the details before
the last minute ..
... we had text that had been there right through that then got
questioned right at the end.
... we had to have meetings on specific points that were
raised, the doc got much better
... but it is important to read t he details asap is better
jtandy: and specific targeted meetins with the person raising
the issue
bernadette: also good was f2f to foucs
jtandy: and you got the WG to comment on particular BPs, but it
all comes in a rush at the end
phila: in theory we need to every thing finalised right now --
time is running out
jtandy: until 12:30 need to shake down best practices -- which
are the least clear and/or most important?
... linda is preparing a tally -- which has top priority to
nominate of the 17?
clemens: bp 1, bp 7, a missing one on using complex iso models
and how to translate to rdf json and others -- how do we advise
data publishers on this?
ClemensPortele: is there a simple solution? Needed for inspire.
billrobe_: 4, 7, 8
eparsons: 4, 7 ,14
BartvanLeeuwen: 4,7,14
AndreaPerego: terminology: spatial thing, features, and its
effect on the draft ontology
<eparsons> Note hadleybeeman
s/spaatial/spatial/
AndreaPerego: 7
frans: 8 and 9
Linda: 4, 7, 8
jtandy: 9 (fuzzy boundaries),, 7, something on crowdsourcing
<Linda> any voters on webex?
hadleybeeman: process question: are you planning
implementations?
jtandy: we plan to point to implementations in the wild -- but
we may have trouble finding some
hadleybeeman: are you separating existing form not yet
existing?
jtandy: we expect to indicate that -- things will not be
"proper best practices" if not implemented in the wild
ByronCinNZ: 3, 4, 9 but if we are doing 3 anyway i vote for
formats (8)
<ByronCinNZ> Yes 8
AZ: 7
jtandy: so we will do top 4.
Linda: 7, 4, 8, 9
jtandy: 20 minutes on each
<jtandy> [32]http://w3c.github.io/sdw/bp/#globally-unique-ids
[32] http://w3c.github.io/sdw/bp/#globally-unique-ids
<phila> [33]BP7
[33] http://w3c.github.io/sdw/bp/#globally-unique-ids
jtandy: pls read the bp for number 7 now
... there is also a meeting thread
<jtandy> see email summary for BP 7:
[34]https://lists.w3.org/Archives/Public/public-sdw-wg/2016Aug/
0139.html
[34]
https://lists.w3.org/Archives/Public/public-sdw-wg/2016Aug/0139.html
<jtandy> ahh - that was the top of the thread ... I'm finding
the summary now
<jtandy> summary =
[35]https://lists.w3.org/Archives/Public/public-sdw-wg/2016Sep/
0096.html
[35]
https://lists.w3.org/Archives/Public/public-sdw-wg/2016Sep/0096.html
BP number 7 [36]http://w3c.github.io/sdw/bp/#globally-unique-ids
[36] http://w3c.github.io/sdw/bp/#globally-unique-ids
eparsons: need some context, e.g. explaining that now you need
to publish stuff at a finer granularity that you are used to
... when we publish we need that every thing/feature/atomoc
piece needs an identifier
<jtandy> [37]http://w3c.github.io/sdw/bp/#how-to-use
[37] http://w3c.github.io/sdw/bp/#how-to-use
jtandy: I was trying to to unpack that conecept here
[38]http://w3c.github.io/sdw/bp/#how-to-use
... so here's your starting point
[38] http://w3c.github.io/sdw/bp/#how-to-use
<ByronCinNZ> Lost audio
[phone dropped out ... messing around trying to fix]
<ByronCinNZ> Thanks
<jtandy> and the identifiers context is provided here:
[39]http://w3c.github.io/sdw/bp/#what-r-u-talking-about
[39] http://w3c.github.io/sdw/bp/#what-r-u-talking-about
<AZ> audio back
<ByronCinNZ> Cheers
jtandy: ... in irs is link to what are u talking about --
context for identifying -- is that the context we need?
eparsons: yes, but we need bits of it restated again within the
relevant bp presenation: this will be very new for SDI audience
although old hat for web people
... will laso apply to other BPs where a fundamental change is
recommended
<Zakim> ClemensPortele, you wanted to talk about "Reuse
identifiers when you can"
ClemensPortele: reuse identifiers when you can -- how is this
going to work?
e.g using dppedia or geonames for the uir, but i want to
publish something else -- if you retirve the uri you will not
come to my information that i am publishing
s/iur/uri/
ClemensPortele: so my publishing will still be dark becuase
noone will come to it
Linda: responding to ed -- is the "why" section of bp not good
enough?
... e.g. BP 7 starts with heading "Why". What is missing?
eparsons: the granularity in particular, not just and enpoint
for a wfs, but every object there
Linda: so if you are used to using wfs -- this is how it is
different
eparsons: addressed to particular users -- highlght how this is
going to be different for you.
jtandy: examples will help with this
billrobe_: on ClemensPortele point about identifiers,
... it depends on your data model and say you are doing
datacube you do want someeone else's identifier so that you can
interoperate
... one very useful case is the n-ary relations wheret ehe
place is the object of the statement
<danbri> hmm
[40]https://en.wikipedia.org/wiki/Smeaton%27s_Tower "Smeaton's
Tower is the third and most notable Eddystone Lighthouse." ->
[41]https://www.wikidata.org/wiki/Q3995634 50°21'51.8"N,
4°8'30.5"W vs
[42]https://en.wikipedia.org/wiki/Eddystone_Lighthouse ->
[43]https://www.wikidata.org/wiki/Q546122 50°10'48"N, 4°16'12"W
[40] https://en.wikipedia.org/wiki/Smeaton's_Tower
[41] https://www.wikidata.org/wiki/Q3995634
[42] https://en.wikipedia.org/wiki/Eddystone_Lighthouse
[43] https://www.wikidata.org/wiki/Q546122
ByronCinNZ: in realtion to data on the we bp, in our case we
need the spatial things themselves
jtandy: dwbp has 2 cases, , thes second is item-level
phila: adwbp a lot of the examples use it as well
... noone is going to store full uris as a waste of space -- it
may not be stored as a uri but must be translatable to one
easily
ericP: could be a relative URI so properly managed
AndreaPerego: responding to phila , in geospatial catalogues we
have unique uids, these can be added to a base uri, is this ok?
<Zakim> danbri, you wanted to ask about syntax (e.g. json-ld
would allow foo:bar qualified names, other contexts allow gzip,
...)
danbri: what syntax ? can have different abbreviations
depending on sytax
... can rely on storage tool
<danbri> another e.g. if you're in an XHTML regime you could
use entities; in JSON-LD qnames or however JSON-LD calls
qualified names.
eparsons: so the best approach is syntax-specific, but the
general idea is the same and we need to force the issue
jtandy: content we are heading the right direction
... billrobe_ what are you specific issues?
billrobe_: BP mostly covers it ... item 3 "stable identifiers"
is a complex modelling problem to decide if it does change
(e.g. boundaries change and sometimes that is critical) but
this is covered in bp as it is
BartvanLeeuwen: agree with clemes becuase need a uri that
resolves to what I want to say about it, and the way I want it
resolved
<Zakim> phila, you wanted to panic a little
AZ: bp 7 is very important but i do not have a specific concern
Linda: we discussed indirect identifiers in mailing list which
was almost resolved and should go here
phila: want to hear from dan, erric, hadley on this
... you do not want to use a goenames uri as the subject in
your rdf?
ClemensPortele: yes, but not necessarily rdf -- i want my
representation to be the target of resolultion
phila: ok, so you don't want geonames to be the response -- so
should we recomment owl:sameas?
... can we use owl:sameAs?
... so maybe 3 people all point to anne frank's house
AndreaPerego: the point is also about provenance to know who is
saying what, maybe this solves that
<Zakim> danbri, you wanted to discuss owl:sameAs being a very
strong claim
ericP: don't want sameas as as it is not the same thing just in
the same place
... want to addreess use case of a location and you need to
make a decision -- you invent more predicates
danbri: what eric just said
... sameas is almost like swearing
... e.g. whiioch poland? and smeaton's tower is it hte same as
edystone lighthouse?
phila: but there is aconnection and we want that reoirded
<AndreaPerego> Wonder whether rdfs:seeAlso could to the job
instead of owl:sameAs?
eparsons: so anne franks house in my database (BP 14 and 15)
linking my datababse of houses in amsterdam with bart's
database of fires in amsterdam
<danbri> seeAlso is pretty weak, owl:sameAs is super strong,
there are things also like skos mappings that sit along the
spectrum between those
eparsons: I need to say that I have a consistent uri within my
database (reuse is elsewhere)
hadleybeeman: but "reuse identifiers" is in here!
ClemensPortele: we have a different understanding of reuse
<eparsons> kerry : duty to create predicate "sameplace"
<ericP> kerry: i believe it is our duty to construct a
predicate that captures samePlaceAs
ClemensPortele: bp 14 is establishing the links,, and that is
different from...
<frans> sameplace could be one of the spatial relationships we
need to have defined
ClemensPortele: we shouln't just look at OWL e.g. geosjson,
e.g. scheam.org sameAs
eparsons: call for wap of scribes
phila: please read DWBP best practice 10 as it contradicts us
hadleybeeman: yes!
<phila> scribe: phila
<scribe> scribeNick: phila
hadleybeeman: If I understand you correctly, people are asying
that they don't want to use other people's identifiers because
you want to make your own statements about things
BartvanLeeuwen: One of the issues we have is, are we too
RDF-centric
... If you put a spatial thing on a map, and people want to
click it, then people can find out more.
... I want to create a page that says Anne Frank's house is on
fire that it has interesting shutters etc.
hadleybeeman: But you're saying you want an ID for the fire at
Anne Frank's House
jtandy: So Bart describes an incident that happens and Ed is
describing the shutters..
hadleybeeman: Draws diagram of the house, its fire and its
shutters
... But they're both about the same place
<ByronCinNZ> Yes I am here
hadleybeeman: You need diff IDs for the incident and the
shutters but they're both about the same thing and for that we
say use the same URI
billrobe_: Agrees with hadleybeeman and gives example. Also,
we're not only in the RDF world, so the majrity of use cases
work well with using some definitive ID for Anne Frank's House
... I can deliver my info about Amsterdam as HTML and RDF...
<hadleybeeman> Or CSV
<Zakim> ericP, you wanted to say that reuse identifiers is
always great. also complicated, also worth reminding people,
but creating the abstract relationships ducks the controversy.
jtandy: And then maybe use schema:sameAs
ericP: Some perspective... every BP in every environment says
use URIs as IDs but the zeroth law is don't lie
... A little bit of ontology that allows you to say 'same
location as' - people get to reuse IDs there. They're not
encouraged to reuse IDs where they shouldn't
... Also get to avoid complex conversations...
... Coming up with hard and fast rules about saying two
incidents are the same is very hard
... You don't want to get into worrying about why two things
are the same
ahaller2: Eric said what I wanted to say. Same location as is
the only predicate we need.
ClemensPortele: I'm worried about this strong 'same'. Many
things in a map will have IDs, and finding 'the canonical
identifier' is nigh on impossible and no one will do it.
... A more relaxed linkage is good
jtandy: I think we're concluding that we need a more relaxed
relationship for this. Maybe this is a missing BP
... We're saying that if you want to relate two things as being
in the same place, we should show how to do it without
necessarily using the same ID.
eparsons: I think that's BP 14 or 15, not 7
<danbri> proposal:
[44]https://gist.github.com/danbri/12cbbdb26cfa25a5bc6ac2060788
766f (too long for IRC :)
[44] https://gist.github.com/danbri/12cbbdb26cfa25a5bc6ac2060788766f
frans: Spatial same as - in spatial data, we talk about spatial
things and geometries. You have a well established system for
saying geometries are the same
<kerry> +1 to frans remark
frans: Maybe we need a set of relationships for spatial things
jtandy: So the predicate that Kerry suggested... some sort of
geometric calculus?
kerry: No, some sort of (non-computable) spatial relations.
<Zakim> hadleybeeman, you wanted to ask if "best practice" (vs
normative spec) allows for this kind of fuzziness already
kerry: You know we talked about backlinks? This is where I
think it's relevant
hadleybeeman: I hear you discussing this BP and it sounds as if
you agree but you're looking for edge cases where it doesn't
work. Does this being a Note help to allow some fuzziness?
jtandy: In GeoSPARQL - OGC is creating an ontology for this
hadleybeeman: Butr you're writing BPs which doesn't need to be
as precise as a Rec
kerry: I don't think they're edge cases, they're normal
BernadetteLoscio: When you're talking about datasets and the
items within them, the item is Anne Frank's house? Not stuff
within it?
jtandy: Anne Frank's house could have a point, 2 or 3 D polygon
[More discussion of IDs for Anne Frank's House]
<hadleybeeman> [and assertions about Anne Frank's House]
-> [45]http://sws.geonames.org/6618987/ Anne Frank's House
[45] http://sws.geonames.org/6618987/
[46]http://w3c.github.io/sdw/bp/#indexable-by-search-engines
[46] http://w3c.github.io/sdw/bp/#indexable-by-search-engines
<danbri> "Search engines should receive a metadata response to
a HTTP GET when dereferencing the link target URI." … what is a
link target URI?
billrobe_: We spent a lot of time talking about creating
precise machine readable data about things that, AFAIK, search
engines are ignoring.
... schema.org might be a route to providing something useful
for search?
... Is there sometehing already in place that we can use in our
BPs?
jtandy: We've suggested that schema.org provides a vocab that
helps search engines index things like schema:Place
... Does that help search engines answer questions like find
coffee shops near here?
danbri: Potentially
... schema.org works because it sits on top of what was in
place already.
... RDF tried to build a parallel Web that ignored the existing
billions of pages.
... If you express your data in schema.org, there's no
guarantee it'll be used.
... I don't understand what a link target URI is?
jtandy: For a URL, if you defref a URL, you should get back an
HTML page that might have embedded data
... but the granularity might change. You want a URL for
everything in a WFS
ClemensPortele: That exists already
jtandy: But it's not crawlable
danbri: A lot of sites used to hide things behind HTTP POST
... I think the same thing happened around SDIs - things are
hidden. You need to make a Web page, make sure robots.txt
doesn't exclude it etc.
... Don't treat it as a special universe.
billrobe_: Makes sense, but it's not often practiced.
danbri: Content negotiation is a tricky one. Se Web loves it
ericP: JSON likes it
... It's a problem if the data is too large
ClemensPortele: In the Geonovum test, what we did was what we
said here. We ctreated an HTML paghe for every resource that we
had and included schema.org in that.
... the tricky part
... theoretically it's a BP, in the real world it's not
exploited.
... We could't really argue that it made a big impact
... If we look at reviews, it's definitely a BP. But not really
for everything.
ByronCinNZ: I feel like it's tryiung to say too much
... There's a lot in there that I find contentious-ish. A BP on
fail metadata, well that's about keeping metadata up to date.
... Maybe it could be more succicnt. What actually is the
point?
... Some BPs have really good examples
... could be more direct and more usable.
<danbri> for Google's use, see also
[47]https://developers.google.com/search/docs/data-types/local-
businesses (opening hours oriented),
[48]https://developers.google.com/search/docs/data-types/events
(events…)
[47]
https://developers.google.com/search/docs/data-types/local-businesses
[48] https://developers.google.com/search/docs/data-types/events
frans: Should this BP not simply say, make HTML models of the
data you provide?
... It you at least make HTML pages you've made progress on
making your data search engine searchable.
... The other thing is linkage between the thing and the
metadata
... I imagine a SE requests a page, looks for links and then
follows those links
... What's required is a link from the data to the metadata and
then links within that. Maybe to subsets and other subsets
<Zakim> danbri, you wanted to discuss w3c instruments
<ericP> [49][media-types] Review request for
application/geo+json-seq media type registration
[49] https://mailarchive.ietf.org/arch/search/?email_list=media-types
Linda: Like a sitemap
ClemensPortele: But sitemaps are limited to 25K links
... We had 8 million addresses
Linda: It can be paged
danbri: Is the word 'Best' correct. W3C likes to attract people
to try out new stuff. It seems a lot of what we're talking
about is new.
... I spend a lot of time trying to get people to make use of
schema.org data.
... If all we can write is BPs then we're limited. If we can
say emerging practice we perhaps can go further
eparsons: There's a heirarchy of BPs. There are simple things
you can do, like exposing what's behind your WFS. Next step is
to create HTML pages, next step is to add in structured data
<ericP> i believe hierarchies like this are expressed on coffee
mugs
eparsons: i.e. take a stepwise approach. BPs can be
incremental.
<Zakim> jtandy, you wanted to ask "is something _on the web_ if
search engines can't see it?"
<danbri> I was just reviewing [50]https://www.w3.org/TR/mwabp/
Mobile Web Application Best Practices (2010). A lot of it is
both precise and has survived the test of time. It updates even
earlier work,
[51]https://www.w3.org/blog/BPWG/2010/12/14/mobile_web_applicat
ion_best_practices_is_2 from
[52]https://www.w3.org/TR/mobile-bp/ (2008). Earliest I can
find is [53]https://www.w3.org/TR/2005/WD-mobile-bp-20051017/
[50] https://www.w3.org/TR/mwabp/
[51]
https://www.w3.org/blog/BPWG/2010/12/14/mobile_web_application_best_practices_is_2
[52] https://www.w3.org/TR/mobile-bp/
[53] https://www.w3.org/TR/2005/WD-mobile-bp-20051017/
jtandy: Is it on the web only if search engines can't see it?
They only look at webpages
danbri: Nope, images, etc.
jtandy: Is the BP more along the lines of creating a human
readable page and then maybe a structured version
eparsons: WxS isn't on the Web, it's on the dark Web
jtandy: If we want our data to be on the Web, people should be
able to find it with a normal browser. Better still using some
structured data as well (schema.org)
<Zakim> hadleybeeman, you wanted to talk about how this works
for browser standards
jtandy: if we can encourage people to do that then the SEs
might start to use it.
<eparsons> zakim close queue
jtandy: I think we've made progress with BP4, yes
phila: It's consistent with DWBP's advice on making (meta)data
human and machine readable
Best Practice 8: Provide geometries on the Web in a usable way
-> [54]http://w3c.github.io/sdw/bp/#describe-geometry BP8
[54] http://w3c.github.io/sdw/bp/#describe-geometry
jtandy: Proposes to take 60 mins for lunch
... Then we can pick up BP8
... Rather than try and rush it in 8 mins
hadleybeeman: On Best/Emerging practices...
<eparsons> Lunch at 12:30 - will be back at 13:30
hadleybeeman: We've been discussing this a lot in the TAG and
whether W3C is where standards are created or ratified
... What the HTML Web Browser world is that for any new idea,
they want it hammered out in a Community Group.
... They'll form a group within the Web Incubator Community
Group
... So that by the time it's in a WG it's already in the wild
... Then WGs aren't working from scratch
[Discussion around future work, life of the WG etc.]
Discussion of difference between OGC and W3C in terms of end of
work for WGs. OGC's carry on indefinitely, even in dormant, W3C
has to start again
danbri: Can it transition to a CG
phila: Yes of course
<eparsons> Lunchtime everyone
[Adjourned for lunch]
<eparsons> Slowly returning from lunch...
<jtandy> so ... we're just restarting ...
<jtandy> in the room we've decided to try to complete the
discussion on BP7 about "indirect identification" ... see
summary of email thread at
[55]https://lists.w3.org/Archives/Public/public-sdw-wg/2016Sep/
0096.html
[55]
https://lists.w3.org/Archives/Public/public-sdw-wg/2016Sep/0096.html
<eparsons> Webex back I hope - can you hear us ByronCinNZ ?
<ByronCinNZ> Yes
<eparsons> Perfect thx
<ByronCinNZ> Will be jumping over to Orlando shortly for the
DCAT metadata OGC. Will return after
<jtandy> for ref, see
[56]http://www.w3.org/TR/2004/REC-webarch-20041215/#indirect-id
entification
[56]
http://www.w3.org/TR/2004/REC-webarch-20041215/#indirect-identification
<billroberts> scribe:billroberts
<scribe> scribenick:billroberts
Indirect Identifiers
<DanhLePhuoc> pressent+ DanhLePhuoc
jtandy: summarises the ideas of 'Indirect Identification' based
on the link above
... do people recognise this practice as useful and something
that happens a lot?
<eparsons> billroberts Has not experienced any issues with
indirect identification
<eparsons> billroberts metadata solves Last updated problem
danbri: schema.org has a vocabulary that is quite agnostic. You
could use that rigorously in terms of differentiating
identifiers and documents about them
<Zakim> danbri, you wanted to say two things: forcing thing vs
page-about-it distinction everywhere comes with costs (for
publishers and consumers); muddling up things with their
danbri: but you can also use it 'scruffily', using a web page
about say Brad Pitt as a way of referring to Brad Pitt
... it was found to be hard to get the idea of distinguising
web address and identifier to developers. It would be possible
if there was a distinct benefit, but that probably isn't hte
case. There isn't the incentive at the moment
... there is a cost to enforcing the distinction, and there is
a cost to mixing them up. So you have to weigh up the pros and
cons
<roba> hi - joined via webex but its behaving differently - not
hearing anything and it offering a video session, no mic mute
option.
jtandy: can we say that it's common practice to conflate the
identifier and the document, and that's ok?
eparsons: we can say it's common practice for sure
kerry: but should we endorse it?
phila: it becomes a problem to use an indirect reference, if
you use it in the wrong way
... eg to say the mountain is 374kb, but if you are saying
something sensible in that context, then it causes no problems
... this is a widely used approach and it generally doesnt'
cause problems
AndreaPerego: if your use case doesn't need the distinction,
then don't do it
<danbri> All I was going to say: at this relatively early
stage, when a given thing doesn't have a lot of machine
readable Web descriptions, strict separation of
thing-vs-description identifier is overkill. But once we have
more adoption it may prove increasingly valuable.
<eparsons> billroberts depends upon context - relies on human -
but thats ok
<phila> billroberts: We're agreeing that it depends on context
and we're relying on the human operator to apply that context
<roba> q
<DanhLePhuoc> +q
frans: we can probably assume that people won't generally be
confused about spatial things - i.e. they won't think it's a
document
ClemensPortele: but we then may need to distinguish Spatial
Thing and Feature?
kerry: how do we convey to a user of the data what we mean by a
URI?
roba: one use case is citing an object. In that case you want
an identifier for the thing, not a representation of it. If the
data provider wants his representation to be cited, then you
might have to make the distinction
... one possible way to do that is the URI redirection
approach.
jtandy: do you mean something like a 303 redirect to a WFS
endpoint?
roba: doesn't matter too much what you redirect to, you can
just use the test on whether it is redirected or not
... it's like referencing a geometry not a feature
jtandy: tries to summarise and play back Rob's point. If you do
redirect, you've separated out the thing and the representation
roba: not quite as strong as that.
... if it doesn't redirect, you could tell that the URL is not
safe to be an identifier
jtandy: how much context do we need? how to express it?
roba: one approach is to get the context by dereferencing it,
but I don't think we can say that is a best practice
jtandy: so, as Dan says, we're all blundering around
DanhLePhuoc: a data snippet can be valid and useful without
having an http URI
... so could use non-http identifiers (URNs eg)
jtandy: I think we want to recommend HTTP identifiers, even if
they don't resolve on the web
DanhLePhuoc: using identifiers that are not HTTP means that you
don't have the cost of setting up a web server to allow
dereferencing
<Zakim> phila, you wanted to talk about dumb strings
phila: URIs should be treated as dumb strings, you can't infer
any meaning from them
<eparsons> billroberts redirects not practical for mass market
users
kerry: also uncomfortable about using the redirect behaviour as
a way of inferring context
roba: I think that's ok, but a best practice should be to use
redirects. If you are using a representation URL you need to be
clear it's not an identifier
jtandy: trying to summarise - you shouldn't be obliged to
separate identifier and document, unless you see a good reason
to do so
roba: a good reason to do so is to make it clear to users
jtandy: will park the discussion on Spatial Thing and Feature
Bart's demonstration of linking to WFS
BartvanLeeuwen: we're working for emergency services in
cross-border or cross-discipline contexts
... there is more and more spatial data being shared between
these partners in ad hoc ways
... For example, water boards collaborating with fire
departments on flood evacuation plans
... The water board is opening up its SDI for the fire
department
... For example the fire department comes across information
about a critical section of a dike (i.e. high risk of a flood)
... misinterpretation between the two organisations of what was
meant by 'critical'. Water board had a different idea to the
fire department
... [Bart shows a GIS WFS system with attributes of objects on
the map]
... so we suggest having a rdf:seeAlso attribute on all WFS
systems, to link to a place to store more information
... Fire department is happy with an extra column in their
system, but don't want to worry about supporting all kinds of
formats
... following the seeAlso link goes to a Linked Data page about
the thing, which in turn can link to definitions of concepts
and other terminology
... so this is a simple and generic way of linking a WFS to a
Linked Data representation of the objects
... We use standard linked data principles for dereferencing
the URIs to get data
... Example showing a map of an incident, with icons to
represent different situations, (eg a flaming icon to indicate
a fire)
... this approach allows icons to be connected to definitions
of what they mean
... and makes it possible to swap in different sets of icons
for the same meaning, to make it familiar to someone from a
different organisation
... From the opposite perspective, there is currently no
standard way to link from the Linked Data representation to the
feature on the map
ClemensPortele: there could be just a WFS request that returns
the feature
BartvanLeeuwen: Jeremy and I have discussed this as a possible
best practice
jtandy: so in essence, you are supplementing an existing SDI by
putting one column in the database that links to a Linked Data
representation, where all the semantic integration can take
place - but you can still display it on a map
... beautiful in its simplicity
... (1) (maybe controversial) web mapping is explicitly out of
scope - is this web mapping?
consensus: no this isn't web mapping, it's about linkability
jtandy: (2) I'm minded of discussions back in Amersfoort, where
billroberts mentioned some hybrid approaches using triple
stores alongside other things
... Bart's work seems a similar kind of thing
<roba> re embedding a link, of course the issue is whether the
uri should be for a specific information resource or for an id
which should dereference :-) There's your use case to consider
<Zakim> danbri, you wanted to say yay
jtandy: this seems to fit under the heading of data access -
different ways of getting to the info
danbri: many examples that we were talking about earlier were
very sparse, so not much benefit for the effort. This seems
much more obviously beneficial as you can link together lots of
things all at once
frans: can understand the perspective of the water boards on
not wanting to do lots of work on detailed definition. But
there is software that makes setting up a WFS pretty easy. If
people act on our BPs then maybe publishing the supporting data
will be easy in future too
Linda: thinking about how to fit this into the document. Is
this a new BP, or a possible approach to implementing an
existing BP?
... could maybe fit in BP11 about convenience APIs?
eparsons: this is more about the linkability best practice
roba: although I couldn't see the demo, I think I got the idea:
this is a general use case of embedding a link to information
about an object in another context
... so what kind of link do you embed in your data? a link to
an identifier or another resource?
... probably need a standard practice here so that people know
what to expect. It should probably be the identifier
... in that case the identifier can then link to various
representations. Otherwise the implementer has to make a choice
of which kind of representation to link to
phila: points out that in Bart's example, there is both
information about the thing and information about the document
about the thing. So this example makes the distinction between
thing and representation
ClemensPortele: this example is mainly about easy data access.
In QGIS you just have a string about the attribute. In the LD
version, you can link off to definitions of the terminology
<roba> do you want t force all WFS to use exactly the same set
of choices as to how to link to different resources and
additional information - or make that the URI dereferencing
practice?
ClemensPortele: it doesn't naturally fit in one of the existing
best practices
... not sure if it's 'best' or 'common' or 'emerging'
<roba> best does not imply common, but if common works its
probably "best". Where "common" is missing or doesnt meet
identified needs best is closer to "good"
billroberts: I think it probably fits into our existing best
practices on linkability and on making links
ClemensPortele: it's important to make it self-descriptive
Linda: maybe there is a DWBP we could link to
phila: point of process. Is there anything proprietary in
Bart's work?
BartvanLeeuwen: no
<roba> its obviously about linkability - and if the practice
link is to something via a URI that dereferences to a document
that provides metadata, then it meets several BP cases
BartvanLeeuwen: would like documentation of which attribute to
use for this, to try to make it more standardised
jtandy: different applications might require different data
models
... so not sure we should specify always rdf:seeAlso or
whatever
<BernadetteLoscio> I think is this one:
[57]https://www.w3.org/TR/dwbp/#ChooseRightFormalizationLevel
[57] https://www.w3.org/TR/dwbp/#ChooseRightFormalizationLevel
eparsons: this is good because you don't need to do anything
difficult but has many benefits
<ClemensPortele> scribe: ClemensPortele
<scribe> scribenick: ClemensPortele
jtandy: next topics: General BP issues, Narrative, Plan for
next draft
Are we meeting the needs of practitioners- if not, how can we
improve?
jtandy: not writing BPs for managers, but for those doing the
work
... are we meeting the needs?
frans: No. Related to BP on geometry, etc. Many options to
choose from, but no real guidance how to do things
jtandy: agrees, we don't say how to make the choice
... Unlikely there is a single choice that fits everywhere.
Example: GeoJSON.
... Struggles how to introduce the questions to ask yourself in
the BP text.
frans: Can we also identify the characteristics a good format
has?
jtandy: Probably difficult, depends on the perspective and use
case. Support for one or multiple CRS is an example.
eparsons: We may be meeting the needs of the wrong
partitioners, ie. the GIS community, less the "Web community"
... distinguish implementation recommendations / options based
on the specific needs
... what meets the 99% of the cases, let's make that the
default
Linda: are Web developers still in the audience, not so clear
from the current text
(yes they are)
ByronCinNZ: A couple of things that may be missing:
... wider meaning of spatial beyond geospatial (but not really
addressed in the text)
... clarify gaps that are relevant for bridging between SDI and
Web developer community, e.g. spatial accuracy depending on the
CRS/projection
jtandy: I.e., provide more help on how to pick the right
datum/projection?
ByronCinNZ: Yes
jtandy: Any similar topics?
eparsons: Publish a raster or vector data?
billroberts: responds to frans "what is a good format"
question: what will likely be used. So probably providing
multiple options, e.g. GeoJSON for Leaflet and Shapefile for
their GIS
... publish once, use many times
frans: happy to read in the current BP to focus on the use of
the data and keep the users in mind
... points to limited choice of data types, which is currently
missing for geometry
... we should work towards a single way of expressing geometry
... Also, current text is too much about geospatial data, less
useful for use cases like architecture / BIM etc.
<Zakim> phila, you wanted to pick up on non-geo spatial
phila: content depends on contributors. Asks Chris from the BIM
domain ...
Chis(?): Welcomes guidance. CRS guidance relevant and different
CRSs will be used (e.g. inside the building). There are open
questions how to do this best.
phila: Probably GeoFencing group did not consider the case
where the CRS changes between two different polygons.
BartvanLeeuwen: Have we outreached to the "Web developer"
community and asked them, if it is useful what we are doing?
phila: Trying to reach as many people and communities as
possible
eparsons: Yes, what we do depends on the people who turn up
<Zakim> phila, you wanted to talk about this evening
[58]http://www.meetup.com/GeeksIn-Lisbon/events/233972259/
[58] http://www.meetup.com/GeeksIn-Lisbon/events/233972259/
dmckenzie: We need to identify and contact the communities
where we want feedback. Can use OGC communications channels.
<phila> [59]Dev Meetup this evening
[59] http://www.meetup.com/GeeksIn-Lisbon/events/233972259/
phila: that might have been an opportunity to pitch our work
and get feedback
dmckenzie: many of these will be very regional, so hard to
cover this broadly
billroberts: the extra day at Amersfoort may be a good example
to follow
dmckenzie: OGC University DWG may be a channel for outreach
... or other DWGs that link to larger communities
<eparsons> coffee break until 14:50
<eparsons> coffee break until 15:50 sorry
"How do we kick the RDF habit?"
jtandy: you can use Linked Data in many representations, not
just using RDF
frans: since SDWBP is an extension of DWBP, how is DWBP?
<ByronCinNZ> audio please. Sounds like an interesting
conversation
jtandy: not heavy, but many of the examples use RDF
kerry: use of link type registry is one example that can help
as it is general
jtandy: yes, publish semantics in the registry of IANA
... temporal relationships there is a proposal discussed with
Simon Cox
... spatial relationships - there has been no feedback on which
of the options to use
... on topology, direction, distance
<kerry> +1 in principle
(general agreement)
<AndreaPerego> IANA Link Relations:
[60]http://www.iana.org/assignments/link-relations/
[60] http://www.iana.org/assignments/link-relations/
Linda: but we need to agree on the list
AndreaPerego: have checked, if anything is there?
jtandy: Yes, nothing there
AndreaPerego: May introduce overhead on the publisher side. But
has advantage that it can also be used directly in HTML
jtandy: Whatever we do, there will be some burden on the
publisher, currently it simply is just no option that a
publisher could use
AndreaPerego: maybe the profile link relation could be an
option
kerry: it is important to capture the more informal spatial
relationships that are used on the social level
<AndreaPerego> Definition of the "profile" link relation (from
IANA registry): "Identifying that a resource representation
conforms to a certain profile, without affecting the
non-profile semantics of the resource representation."
kerry: ie the topological ones are not the most important ones
... focus on those that are used in common language
frans: why do we want to "kick the RDF habit"?
<ByronCinNZ> audio please
jtandy: if we focus on RDF people would quickly conclude the
document does not apply to them
Linda: we have a link to the Linked Data BP this document
becomes very RDF centric
<roba> using RDF is a practice - perhaps best for some things
but not the only option.
jtandy: two schools of people, a) Linked Data must use RDF and
b) takes a more relaxed position. Like the BP document...
... Need to use examples that are based on other approaches,
GML, GeoJSON, maybe OData, etc
<Zakim> AndreaPerego, you wanted to comment on Fransis's point
AndreaPerego: Just to say that link relations are already used
in HTML documents, e.g., to link to stylesheets. So, this makes
it easier to use them to express also other relationships.
BartvanLeeuwen: Mention of RDF is a religious thing. People
reject something just based on the reference to RDF
BernadetteLoscio: DWBP had similar discussion. Introduction has
discussion of the relationship and avoid specific focus. But
many examples are in ttl.
<Zakim> danbri, you wanted to discuss RDF
danbri: Have used RDF in other groups, but without making a big
fuss about it
<jtandy> BP doc tries to present no bias to RDF here:
[61]http://w3c.github.io/sdw/bp/#linked-data
[61] http://w3c.github.io/sdw/bp/#linked-data
frans: having other examples is a good idea and the Linked Data
text should state that it is not abut RDF
jtandy: Chapter 7 already does this, see link above
billroberts: you want web identifiers and you want linking,
then you are nearly at RDF
<eparsons> ClemensPortele : RDF point was mine - current draft
better
eparsons: Both the GIS and Web developer communities consider
RDF a "nasty beast". So not highlighting RDF will help
communicating the BP
jtandy: general consensus and going in the right direction
... how can we address kerrys point about the spatial
relations. Can someone propose a list of spatial relationships?
phila: can we just register the ones from GeoSPARQL?
jtandy: Which of the three sets?
... and these are only the topological ones
eparsons: The directional and distance related ones are more of
a challenge
kerry: Some of them are context dependent (near/far)
<Zakim> ericP, you wanted to ask about wikipedia infoboxes
ericP: could the wikipedia boxes provide any insight?
(possibly)
jtandy: is there a link to more information?
<kerry>
[62]https://en.wikipedia.org/wiki/Category:Geography_infobox_te
mplates
[62]
https://en.wikipedia.org/wiki/Category:Geography_infobox_templates
<Zakim> phila, you wanted to talk about IANA Links don't need
to be in a W3C standard, and to ask DanBri whether he can help
phila: topological ones could easily be added by contacting
IANA / Mark Nottingham to add them
... for the others, is there something that we can point to?
danbri: we could also add them to schema.org
phila: how widely implemented are the relationships?
eparsons: typically widely implemented by GIS, so we would need
to analyse what has been implemented in tools
<Zakim> billroberts, you wanted to ask about progress on
relating things to geometry
<ericP> +1 to dim[B(a)∩I(b)]=1&<arg(x∨x_) ˚͜˚
jtandy: and how do we identify the directional /
distance-related ones? any sources to use as a basis?
billroberts: related to the work on the spatial ontology
... can schema.org help?
danbri: there are existing properties that relate objects to
geometries
PROPOSED: submit topological GeoSPARQL Simple Feature relations
to the IANA link relation registry.
frans: there could a difference between the spatial relation of
spatial things or their geometries
eparsons: the topological ones depend on the existence of
geometries
... ... the topological ones are always computable from the
geometries
ericP: and how about the more fuzzy relationships
<Zakim> phila, you wanted to talk about Young's Calculus
DanhLePhuoc: (sorry, I missed that)
phila: if we have the list of well-defined relationships for
topo ones, should we do this for the temporal ones (Allan's
calculus), too?
<DanhLePhuoc> there is some relationship can be calculate
without geometric information, for instance, located in, part
of, can be computed via transitive reasoning
jtandy: I can make topologial assertions without geometry.
<danbri> phila, see "We don't nitpic about whether they're
alive, dead, real, or imaginary. " in
[63]http://xmlns.com/foaf/spec/#term_Person
[63] http://xmlns.com/foaf/spec/#term_Person
eparsons: but you cannot prove them
kerry: the informal ones are more valuable to the formal ones
<eparsons> +1 tp both
<Zakim> danbri, you wanted to mention 'equals'
danbri: only one seems only useful in a mathematical sense
(equal), the others also make sense in a colloquial sense
<Zakim> kerry, you wanted to mention
[64]https://jena.apache.org/documentation/query/spatial-query.h
tml
[64] https://jena.apache.org/documentation/query/spatial-query.html
jtandy: It would be a well-defined topic to make a proposal for
the link relations. Any volunteers?
phila: where to put, in schema.org, W3C space? Should it also
go to the BP, too?
... IANA wants to reference something stable
<Zakim> ericP, you wanted to propose email
ericP: simplest could be an email to the mailing list
kerry: this might be a starting point:
[65]https://jena.apache.org/documentation/query/spatial-query.h
tml
[65] https://jena.apache.org/documentation/query/spatial-query.html
<Linda> And schema.org has
[66]https://schema.org/containedInPlace I saw
[66] https://schema.org/containedInPlace
<RaulGarciaCastro> scribe:RaulGarciaCastro
jtandy: we want to include: computable relationships and
assertive relations (not necessarily based on computations)
... … who can take the lead for doing that?
eparsons: what has to be done?
jtandy: name + description
<phila> [67]An excellent example of a namespace document
[67] https://www.w3.org/ns/csvw
eparsons: I take the lead
<ericP> [68]An adequate example of a namespace document
[68] https://www.w3.org/ns/ldp
jtandy: there is no best practice for these relationships;
there is a gap there
<ericP> (though it does introduce some convention for
properties of a class)
jtandy: … could this be part of the namespaces work?
(yes)
jtandy: how to use spatial relationships for uncertain
boundaries? There is a common practice to do it
<eparsons> action eparsons to work with chairs to define
spatial relations namespace document
<trackbot> Created ACTION-198 - Work with chairs to define
spatial relations namespace document [on Ed Parsons - due
2016-09-26].
kerry: we can handle it talking about fuzzy relationships
... … and I don’t even need to knwo the geometry
eparsons: sometimes there are things with no geometry
associated
roba: relationships depend on the use case; we need a mechanism
to specify what you need in your context
frans: if we relax the relationships to things without
geometry, anyone can make a statement about anything
<Zakim> ericP, you wanted to ask if time management mechanisms
can be used here
ericP: you should expect people acting in good faith (a trust
issue)
jtandy: some expert should help me with the examples of spatial
relationships
... bill, what are your thoughts about producing statistical
data?
AndreaPerego: When are we talking about SpatialThing vs
Feature?
eparsons: let’s plan now with everyone here
<Zakim> phila, you wanted to ask BernadetteLoscio and newton
their thoughts on the usefulness of the running example
phila: Is narrative important?
BernadetteLoscio: The running example was useful, even if in
some cases it was difficult to come up with it
jtandy: We were thinking on a flooding example, but it is
complex
... … I don’t expect the best practices to be used alone, but
with other documents
eparsons: Maybe we could reduce the scope in the narrative?
jtandy: Some answers that the narrative is supposed to answer
are already answered in the document
frans: How about using the narrative in the examples to give
coherence? Right now it is separated
eparsons: Maybe we can prioritize the best practices
... … trying to include everything makes things complex
billroberts: have been trying to find population statistics and
examples, and this raised a number of questions that can be
performed (e.g., is the population data in machine-readable
form?)
... … this can give hints to data publishers
ClemensPortele: If you remove the narrative and put it into the
examples it may not be so convincing
jtandy: how do we plan net steps? (will think on the narrative
for tomorrow)
Linda: how far is the current document for the next working
draft?
<Zakim> phila, you wanted to be annoying
<kerry> +1 it has moved a lot from previous version
<eparsons> +1
phila: The document is already very good; please publish it as
soon as possible
... … feel free to take things out
... … right now it is more than expected
eparsons: publishing it is the way of getting more people
involved
phila: get what you got in a published document (even with open
issues) in a week or two
jtandy: Pending things: update the glossary with missing
definitions (anyone?), bibliography, open issues, changelog…
phila: I will help with the document (formatting, language,
etc.)
<danbri> [69]http://pending.webschemas.org/GeospatialGeometry
(based on
[70]https://github.com/schemaorg/schemaorg/blob/sdo-callisto/da
ta/ext/pending/issue-1375.rdfa (based on
[71]https://en.wikipedia.org/wiki/DE-9IM )). Various
'deliberate mistakes' included to check if anyone reads it.
[69] http://pending.webschemas.org/GeospatialGeometry
[70]
https://github.com/schemaorg/schemaorg/blob/sdo-callisto/data/ext/pending/issue-1375.rdfa
[71] https://en.wikipedia.org/wiki/DE-9IM
AndreaPerego: Do we need the notion of a spatial thing? Not in
every case we need to differentiate between real thing,
geometry, etc.
<roba> josh lieberman was working on an abstract spatial
ontology - i think we need this to be a lightweight core
<roba> ...updating geosparql may end up with something too
complex ?
<Zakim> danbri, you wanted to re-iterate SpatialThing was from
a random chat
jtandy: In the document we already state that a spatial thing
can be different things
danbri: the origin came due to trying to adopt CYC
AndreaPerego: And people have used it since then
danbri: We can still change it
jtandy: The concept of spatial thing for representing things
with extent is good for us
... the GeoSPARQL ontology is being refactored
... … anyway, review the document to see if the current use of
the term makes you happy
Linda: we have reviewed the document
jtandy: Ensure that the glossary is consistent with the wiki
(anyone?)
... … it is just a compilation thing, no need to write new
content
<BernadetteLoscio> [72]https://www.w3.org/TR/dwbp/#requirements
[72] https://www.w3.org/TR/dwbp/#requirements
<danbri> re basic geo ns, it came from a
[73]https://www.w3.org/wiki/ScheduledTopicChat meeting.
[74]https://www.w3.org/wiki/GeoInfo which has 404 cyc reference
-> [75]http://www.cyc.com/cycdoc/vocab/geography-vocab.html -
earlier version,
[76]https://web.archive.org/web/20070203153714/http://cyc.com/c
ycdoc/vocab/geography-vocab.html
[73] https://www.w3.org/wiki/ScheduledTopicChat
[74] https://www.w3.org/wiki/GeoInfo
[75] http://www.cyc.com/cycdoc/vocab/geography-vocab.html
[76]
https://web.archive.org/web/20070203153714/http://cyc.com/cycdoc/vocab/geography-vocab.html
<phila> [77]BP cross ref
[77] http://w3c.github.io/sdw/bp/#requirements
<danbri> so yes, SpatialThing came via Cyc, e.g.
#$SpatialThing-Localized
<danbri>
[78]http://sw.opencyc.org/concept/Mx4rvVjpUZwpEbGdrcN5Y29ycA
[78] http://sw.opencyc.org/concept/Mx4rvVjpUZwpEbGdrcN5Y29ycA
newton: I made a script to build the cross-reference table for
our best practices document; I can help with this document
<eparsons> Action billrobe_ to check Glossary for completeness
<trackbot> Error finding 'billrobe_'. You can review and
register nicknames at
<[79]http://www.w3.org/2015/spatial/track/users>.
[79] http://www.w3.org/2015/spatial/track/users
<Zakim> danbri, you wanted to confirm SpatialThing was indeed
Cyc-inspired, see
[80]http://sw.opencyc.org/concept/Mx4rvVjpUZwpEbGdrcN5Y29ycA
(also to report
[80] http://sw.opencyc.org/concept/Mx4rvVjpUZwpEbGdrcN5Y29ycA
danbri: went through the wikipedia infoboxes for relationships
properties and there are proposals in schema.org; but nothing
has been assessed by experts
jtandy: Regarding bibliography, proper references in ReSpec
must be found (in yellow)
phila: I can manage that
<phila> ACTION: phila to help improve the bibliography for the
BP doc [recorded in
[81]http://www.w3.org/2016/09/19-sdw-minutes.html#action01]
[81] http://www.w3.org/2016/09/19-sdw-minutes.html#action01]
<trackbot> Created ACTION-199 - Help improve the bibliography
for the bp doc [on Phil Archer - due 2016-09-26].
<Zakim> kerry, you wanted to talk about tomorrow agenda before
we leave
phila: I can help in placing the icons for the benefits
jtandy: we will try to have a stable release in two weeks from
Wednesday (15th October) so it can be published the following
week
... Thanks for all the comments
<danbri> @phila, to answer your [82]http://schema.org/ process
question - my actions fall under project webmaster role
documented in
[83]http://schema.org/docs/howwework.html#webmaster
[82] http://schema.org/
[83] http://schema.org/docs/howwework.html#webmaster
<Zakim> AndreaPerego, you wanted to ask about the agenda for
the SDW workshop @ INSPIRE 2016 (Sep, 30th)
[84]http://inspire.ec.europa.eu/events/conferences/inspire_2016
/page/wsl
[84]
http://inspire.ec.europa.eu/events/conferences/inspire_2016/page/wsl
AndreaPerego: do we want feedback from INSPIRE in the best
practices?
jtandy: tell them about the new future draft so they can give
feedback
AndreaPerego: We must highlight what we want feedback on
... … the workshop is next week on Friday
jtandy: We can talk about it
eparsons: I can present if you give me the content
Agenda for tomorrow
kerry: (reviews agenda)
phila: There is the AC meeting tomorrow at 15:00
... … it may affect the meeting
meeting closed
Summary of Action Items
[NEW] ACTION: phila to help improve the bibliography for the BP
doc [recorded in
[85]http://www.w3.org/2016/09/19-sdw-minutes.html#action01]
[85] http://www.w3.org/2016/09/19-sdw-minutes.html#action01
Summary of Resolutions
1. [86]Issue-75 is in scope, close issue-75.
2. [87]Close issue-70 that we need to pass this on to BP to
handle, encouraging publishers' need to meet broadest
possible community
3. [88]That it is a requirement to have UoM included, close
issue-74
4. [89]There are cases where there is a requirement for more
than one CRS, so we can close issue-76
[End of minutes]
__________________________________________________________
Received on Tuesday, 20 September 2016 07:39:17 UTC