Re: [Talk-ca] Wikidata in OpenStreetMap

[Frederik Ramm]

Hi,

On 11/11/2016 02:15 AM, James wrote:
> Seeing as ID[1] and Mapbox[2] are pushing to incorporate wikidata into
> OSM, I thought it would be a perfect little project when mappers have
> nothing better to do or in their downtime.

Said the user who then proceeded to add Wikidata tags to 20,000 objects
over the course of 8 days. A "little" project for the "downtime", wow.

> I've started a project on the OSMCanada Tasking Manager here:
> http://tasks.osmcanada.ca/project/40
> 
> Join in for a more data rich Canada :)

These paragraphs made me assume that this was about the community
joining in to add local data or at least something the Canadian
community was doing. However, LogicalViolinist was ambitions and added
Wikidata tags to places on the whole planet.

I had hoped that lessons had been learned from the recent past.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"

___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[Stewart C. Russell]

Hi James,

> This should explain more of what wikidata is capable of doing via 1 tag:
> http://www.openstreetmap.org/user/LogicalViolinist/diary/39872

Yes, I think the wikidata tag is a good idea. It would be nice of there
were an easier query language than Sparql to pull out towns/cities that
Wikidata already knows about … but that might constitute an import.

> is_in:country: Canada

Hey, a little bit Anglocentric there. Why not also
is_in:country:lv=Kanāda, is_in:country:ml=കാനഡ, is_in:country:ne=क्यानाडा, …?

> state_code,country_code, iso_3166_2 are ISO notations that can be used
> and have a standard that is international and can be parsed easily by
> computers.

And also complete duplication of existing boundary information. The
is_in:iso_3166_2 seems to be very rarely used (~200 in all of N.
America) and has no documentation. So it's best not used, or should be
discussed and documented first.

The wikidata tag also contains a geographic membership relationship, so
it's not like we need is_in:* to help them.

> What I don't like in "is_in" tag is as you said it's a bunch of garbage

No, I said that the is_in *family* of tags is deprecated. But we're
fairly heavy users of it in Canada (something like 10% of the world's
is_in:* tags are in Ontario + Quebec:
https://taginfo.openstreetmap.org/keys/?key=is_in#overview), likely due
to the age of our imports.

cheers,
 Stewart


___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[James]

This should explain more of what wikidata is capable of doing via 1 tag:
http://www.openstreetmap.org/user/LogicalViolinist/diary/39872

Stewart I do agree that the tag "is_in" is completely useless, I was more
targeting the sub-tags(I put a star not to list them, but I will)
For example Ottawa:

name: Ottawa
is_in:continent: North America
is_in:iso_3166_2: CA-ON
is_in:country: Canada
is_in:country_code: CA
is_in:state: Ontario
is_in:state_code: ON

state_code,country_code, iso_3166_2 are ISO notations that can be used and
have a standard that is international and can be parsed easily by computers.

What I don't like in "is_in" tag is as you said it's a bunch of garbage
thrown together, which helps no one:
name:Canberra
place:city
is_in:capital_cities; Australian Capital Territory; ACT; Australia
is_in:state=Australian Capital Territory
is_in:state_code=ACT
is_in:country=Australia (use english name of the country)
is_in:country_code=AU



On Fri, Nov 11, 2016 at 9:43 AM, Stewart Russell  wrote:

> On Nov 11, 2016 6:04 AM, "James"  wrote:
> >
> > The other tags are precisions (instead of doing massive spatial joins)
> > you will know that say Ottawa is in Ontario
>
> I disagree that these need to be added. is_in* seems to be deprecated, and
> we have (or should have) robust boundary=administrative data for Canada.
>
> My main complaints about is_in are:
> * it's freeform, so you can't reliably create a relation to the parent
> entity from the text;
> * it's language sensitive, so you could in theory have no end of localized
> versions.
>
> Stewart
>
> ___
> Talk-ca mailing list
> Talk-ca@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ca
>
>


-- 
外に遊びに行こう！
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[Stewart Russell]

On Nov 11, 2016 6:04 AM, "James"  wrote:
>
> The other tags are precisions (instead of doing massive spatial joins)
> you will know that say Ottawa is in Ontario

I disagree that these need to be added. is_in* seems to be deprecated, and
we have (or should have) robust boundary=administrative data for Canada.

My main complaints about is_in are:
* it's freeform, so you can't reliably create a relation to the parent
entity from the text;
* it's language sensitive, so you could in theory have no end of localized
versions.

Stewart
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[James]

Ok to clarify the only thing you are adding is the wikidata tag(which is a
foreign key to wikidata (i.e. Q247372)), which is in the instructions in
the tasking manager on what tags to add.

The other tags are precisions(instead of doing massive spatial joins) you
will know that say Ottawa is in Ontario and is in Canada which is in North
America(also easier to do queries against: Ottawa, Ontario or Ottawa,
Illinois?

On Nov 11, 2016 1:51 AM, "Frederik Ramm"  wrote:

> Hi,
>
> On 11/11/2016 05:08 AM, Devon Fyson wrote:
> > With respect to linking wikidata and OSM, all those properties should be
> > added to wikidata if they don't already exist. Then link to it by simply
> > adding the wikidata ID in OSM. Then if OSM ever wants to merge wikidata
> > in, it can all be done automatically via the wikidata ID. Most of what
> > is in that Ottawa example is already in wikidata and so it doesn't make
> > sense to duplicate it.
>
> I don't think James was advocating for adding stuff to OSM, but if he
> was, yes then I'm totally with you - the Wikidata link should be
> sufficient, and we don't want to copy stuff from Wikidata.
>
> (Nor do we want to copy stuff from OSM *into* Wikidata because this
> would then make parts of Wikidata ODbL licensed, something that Wikidata
> certainly won't approve of.)
>
> Bye
> Frederik
>
> --
> Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"
>
> ___
> Talk-ca mailing list
> Talk-ca@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ca
>
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[Frederik Ramm]

Hi,

On 11/11/2016 05:08 AM, Devon Fyson wrote:
> With respect to linking wikidata and OSM, all those properties should be
> added to wikidata if they don't already exist. Then link to it by simply
> adding the wikidata ID in OSM. Then if OSM ever wants to merge wikidata
> in, it can all be done automatically via the wikidata ID. Most of what
> is in that Ottawa example is already in wikidata and so it doesn't make
> sense to duplicate it.

I don't think James was advocating for adding stuff to OSM, but if he
was, yes then I'm totally with you - the Wikidata link should be
sufficient, and we don't want to copy stuff from Wikidata.

(Nor do we want to copy stuff from OSM *into* Wikidata because this
would then make parts of Wikidata ODbL licensed, something that Wikidata
certainly won't approve of.)

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frede...@remote.org  ##  N49°00'09" E008°23'33"

___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[Devon Fyson]

With respect to linking wikidata and OSM, all those properties should be
added to wikidata if they don't already exist. Then link to it by simply
adding the wikidata ID in OSM. Then if OSM ever wants to merge wikidata in,
it can all be done automatically via the wikidata ID. Most of what is in
that Ottawa example is already in wikidata and so it doesn't make sense to
duplicate it. See this discussion

on wikidata.
As for cross referencing existing OSM nodes with existing wikidata IDs, the
fastest way would be to use a script to search between the two databases
with a best fit algorithm then have user intervention to ensure quality.
(eg node 18886011 at 45, -75 is a capital with name 'Ottawa' and looks to
match Q1930 with a 90% fit, next match of ___ is a 50% fit). Even lower
hanging fruit would be to start with cross-referencing via the wikipedia
tags which could probably be fully automated. eg OSM → wikipedia → wikidata
to generate OSM → wikidata.
Devon

On Thu, Nov 10, 2016 at 8:19 PM, John Marshall  wrote:

> Great work James. I'm in.
>
> John
>
> John Marshall
>
> On Nov 10, 2016 20:17, "James"  wrote:
>
>> Seeing as ID[1] and Mapbox[2] are pushing to incorporate wikidata into
>> OSM, I thought it would be a perfect little project when mappers have
>> nothing better to do or in their downtime.
>>
>> The wikidata link can help developers get vital information like the
>> current population, current mayor, flags, coat of arms,etc. The list goes
>> on. This is a lot of data that can't necessarily be encoded into OSM. This
>> would help simplify the use of OSM data for small developers.
>>
>> I've started a project on the OSMCanada Tasking Manager here:
>> http://tasks.osmcanada.ca/project/40
>>
>> Join in for a more data rich Canada :)
>>
>>
>> [1] https://www.mapbox.com/blog/id-wikidata/
>> [2] https://www.mapbox.com/blog/geocoding-wikidata/
>>
>> ___
>> Talk-ca mailing list
>> Talk-ca@openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/talk-ca
>>
>>
> ___
> Talk-ca mailing list
> Talk-ca@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ca
>
>
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Wikidata in OpenStreetMap

[John Marshall]

Great work James. I'm in.

John

John Marshall

On Nov 10, 2016 20:17, "James"  wrote:

> Seeing as ID[1] and Mapbox[2] are pushing to incorporate wikidata into
> OSM, I thought it would be a perfect little project when mappers have
> nothing better to do or in their downtime.
>
> The wikidata link can help developers get vital information like the
> current population, current mayor, flags, coat of arms,etc. The list goes
> on. This is a lot of data that can't necessarily be encoded into OSM. This
> would help simplify the use of OSM data for small developers.
>
> I've started a project on the OSMCanada Tasking Manager here:
> http://tasks.osmcanada.ca/project/40
>
> Join in for a more data rich Canada :)
>
>
> [1] https://www.mapbox.com/blog/id-wikidata/
> [2] https://www.mapbox.com/blog/geocoding-wikidata/
>
> ___
> Talk-ca mailing list
> Talk-ca@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ca
>
>
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

[Talk-ca] Wikidata in OpenStreetMap

[James]

Seeing as ID[1] and Mapbox[2] are pushing to incorporate wikidata into OSM,
I thought it would be a perfect little project when mappers have nothing
better to do or in their downtime.

The wikidata link can help developers get vital information like the
current population, current mayor, flags, coat of arms,etc. The list goes
on. This is a lot of data that can't necessarily be encoded into OSM. This
would help simplify the use of OSM data for small developers.

I've started a project on the OSMCanada Tasking Manager here:
http://tasks.osmcanada.ca/project/40

Join in for a more data rich Canada :)


[1] https://www.mapbox.com/blog/id-wikidata/
[2] https://www.mapbox.com/blog/geocoding-wikidata/
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca