Fixing wiki* -> brand:wiki*

classic Classic list List threaded Threaded
76 messages Options
1234
Reply | Threaded
Open this post in threaded view
|

Fixing wiki* -> brand:wiki*

Yuri Astrakhan-2
Here is a query that finds all wikidata IDs frequently used in "brand:wikidata", and shows OSM objects whose "wikidata" points to the same. I would like to replace all such wikidata/wikipedia tags with the corresponding brand:wikidata/brand:wikipedia.  Most of them are in India, but there are some in Europe and other places.  This query can be used directly from JOSM as well.


BTW, this type of queries might be good for maproulette challenges once they can work more like osmose.


_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

PanierAvide-2
Hello,

Thank you for this query, which will be very useful for detecting these
issues. I'm not sure if this is possible in the current state of
Wikidata, but can't we retrieve all shop chains brands, and then query
OSM to find object having wikidata tag pointing to one of the brands ?
If data is structured enough on Wikidata side, we would have a better
way to retrieve more of the OSM issues.

Regards,

Adrien.


Le 27/09/2017 à 00:47, Yuri Astrakhan a écrit :

> Here is a query that finds all wikidata IDs frequently used in
> "brand:wikidata", and shows OSM objects whose "wikidata" points to the
> same. I would like to replace all such wikidata/wikipedia tags with
> the corresponding brand:wikidata/brand:wikipedia.  Most of them are in
> India, but there are some in Europe and other places.  This query can
> be used directly from JOSM as well.
>
> http://tinyurl.com/y72afjpy
>
> BTW, this type of queries might be good for maproulette challenges
> once they can work more like osmose.

--
PanierAvide
Géomaticien & développeur


_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

SimonPoole
In reply to this post by Yuri Astrakhan-2

While I can understand adding WP and WD tags to objects of note, why on earth would we want to add all this redundancy to OSM objects at all? Particularly given that object type + brand(s) should essentially always be unique, anybody that wants to look up WD keys could do so via a simple external table.

Simon

Note I'm talking about the usefulness for OSM here, that it would be useful for WD is clear.


Am 27.09.2017 um 00:47 schrieb Yuri Astrakhan:
Here is a query that finds all wikidata IDs frequently used in "brand:wikidata", and shows OSM objects whose "wikidata" points to the same. I would like to replace all such wikidata/wikipedia tags with the corresponding brand:wikidata/brand:wikipedia.  Most of them are in India, but there are some in Europe and other places.  This query can be used directly from JOSM as well.


BTW, this type of queries might be good for maproulette challenges once they can work more like osmose.



_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk


_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

signature.asc (499 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
In reply to this post by Yuri Astrakhan-2
On 26 September 2017 at 23:47, Yuri Astrakhan <[hidden email]> wrote:

> Here is a query that finds all wikidata IDs frequently used in
> "brand:wikidata", and shows OSM objects whose "wikidata" points to the same.
> I would like to replace all such wikidata/wikipedia tags with the
> corresponding brand:wikidata/brand:wikipedia.

Thank you. I support this proposal.

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
In reply to this post by SimonPoole
On 27 September 2017 at 10:07, Simon Poole <[hidden email]> wrote:

> While I can understand adding WP and WD tags to objects of note, why on
> earth would we want to add all this redundancy to OSM objects at all?

The question incudes the false premise that this is redundancy: It is
not, it adds disambiguation.

> Particularly given that object type + brand(s) should essentially always be
> unique

You're prepared to guarantee that there are not two different chains,
in the same business, in different parts of the world, with the same
name? I'm not.

For example, until the UK version went titsup a few years back, there
were chains of stores in the UK and in Australia, each called
"Woolworths". Though they had common roots, they were not the same.

> anybody that wants to look up WD keys could do so via a simple external table.

Please can you tell us where to find these tables?

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Richard Fairhurst
Andy Mabbett wrote:
> in different parts of the world

IIRC OSM stores spatial information. I might be wrong.

Richard



--
Sent from: http://gis.19327.n8.nabble.com/General-Discussion-f5171242.html

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

SimonPoole
In reply to this post by Andy Mabbett


Am 27.09.2017 um 13:30 schrieb Andy Mabbett:
>
> For example, until the UK version went titsup a few years back, there
> were chains of stores in the UK and in Australia, each called
> "Woolworths". Though they had common roots, they were not the same.
>
>
Why would that matter to OSM?

Given that they are already disambiguated by geography?

And that if you are looking for a "Woolworths" in Australia you don't
really care if there is/was a chain of the same name in the UK?

PS: in OSM  such brands tend end up in the name tag in any case


_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

signature.asc (499 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
On 27 September 2017 at 12:57, Simon Poole <[hidden email]> wrote:

>> For example, until the UK version went titsup a few years back, there
>> were chains of stores in the UK and in Australia, each called
>> "Woolworths". Though they had common roots, they were not the same.

> Why would that matter to OSM?

It may not, It certainly matters to OSM's users.

Tim Berners Lee coined the "Five Stars of Open Data"

    http://5stardata.info/en/

defining best practice in publishing open data. OSM already meets the
first four, well. The fifth is:

 *  link your data to other data to provide context

and that's what including Wikidata iDs in cases such as the above does.

In fact, since that makes OSM more useful and thus more attractive to
re-users, it *does* matter to OSM.

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
In reply to this post by Richard Fairhurst
On 27 September 2017 at 12:49, Richard Fairhurst <[hidden email]> wrote:

> Andy Mabbett wrote:

>> in different parts of the world
>
> IIRC OSM stores spatial information. I might be wrong.

For some reason I can't determine, you quote me out-of-context; the
context was that we were discussing the assertion that "object type +
brand(s) should essentially always be unique".

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

lsces
In reply to this post by Yuri Astrakhan-2
( Thought I hit 'reply list' :) )

On 26/09/17 23:47, Yuri Astrakhan wrote:
> Here is a query that finds all wikidata IDs frequently used in
> "brand:wikidata", and shows OSM objects whose "wikidata" points to the
> same. I would like to replace all such wikidata/wikipedia tags with the
> corresponding brand:wikidata/brand:wikipedia.  Most of them are in
> India, but there are some in Europe and other places.  This query can be
> used directly from JOSM as well.

wikidata provides a section which documents a range of LINKS which
identify the same object on other databases. It would be nice if there
was a stable identity on OSM that would populate an entry in THAT list.
This would then allow cross checking of any link.

At some point it might be nice to eliminate all of the unnecessary links
like this in OSM if wikidata or somenewopesourcedata become a more
reliable source of such data. Just a single link to a substantial set of
additional data.

So ... where does 'brand:' come from? Should this not simply be 'link:'?

--
Lester Caine - G8HFL
-----------------------------
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk
Rainbow Digital Media - http://rainbowdigitalmedia.co.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

SimonPoole
In reply to this post by Andy Mabbett


Am 27.09.2017 um 15:00 schrieb Andy Mabbett:

> ...
>> Why would that matter to OSM?
> It may not, It certainly matters to OSM's users.
>
> Tim Berners Lee coined the "Five Stars of Open Data"
>
>     http://5stardata.info/en/
>
> defining best practice in publishing open data. OSM already meets the
> first four, well. The fifth is:
>
>  *  link your data to other data to provide context
You are assuming
a) that an arbitrary best practice definition is relevant for OSM
b) that we get any real life brownie points for this outside of academia

>
> and that's what including Wikidata iDs in cases such as the above does.
>
> In fact, since that makes OSM more useful and thus more attractive to
> re-users, it *does* matter to OSM.
You still haven't given any specific real world use case in which
brand:wikidata would actually be helpful and somebody would -really-
want to consume it. Outside of than that it would obviously be good for
wikidata and for people going to linked data conferences.

My take is that it adds a nearly impossible to maintain (consider your
own Woolworth's example), non-speaking, foreign key for additional
information that is already  that is already quite adequately provided
by the brand key.
 
Simon








_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

signature.asc (499 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
In reply to this post by lsces
On 27 September 2017 at 14:28, Lester Caine <[hidden email]> wrote:

> wikidata provides a section which documents a range of LINKS which
> identify the same object on other databases. It would be nice if there
> was a stable identity on OSM that would populate an entry in THAT list.

Indeed so, but OSM does not, nor is it likely to in the foreseeable future.

> At some point it might be nice to eliminate all of the unnecessary links
> like this in OSM if wikidata or somenewopesourcedata become a more
> reliable source of such data.

While it is not yet complete, in what way is Wikdiata failing to be
sufficiently reliable?

> Just a single link to a substantial set of additional data.

That is exactly what Wikidata is for.

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Frederik Ramm
In reply to this post by SimonPoole
Hi,

On 27.09.2017 15:37, Simon Poole wrote:
> My take is that it adds a nearly impossible to maintain (consider your
> own Woolworth's example), non-speaking, foreign key

We generally discourage foreign keys (that are only usable together with
a different data set and that are not signposted locally).

When Wikidata keys were first added to OSM, I thought this was something
limited to place names of a certain importance and I didn't object.

Seeing that this now leads to the automatic assumption that Wikidata IDs
are practically admissible *everywhere* where Wikidata has defined an
ID, I am having second thoughts about the whole issue.

In theory, almost everything we map could be expressed by a Wikidata ID.
If welcoming a Wikidata link on a city place node means that by
extension I also have to welcome "amenity:wikidata=Q123456" on something
that is, say, an ice cream parlour because Q123456 is the generic
Wikidata category for ice cream parlours, then I think I'd rather not
have any Wikidata links in OSM at all.

Bye
Frederik

--
Frederik Ramm  ##  eMail [hidden email]  ##  N49°00'09" E008°23'33"


_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

signature.asc (484 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
In reply to this post by SimonPoole
On 27 September 2017 at 14:37, Simon Poole <[hidden email]> wrote:

> Am 27.09.2017 um 15:00 schrieb Andy Mabbett:

>> Tim Berners Lee coined the "Five Stars of Open Data"
>>
>>     http://5stardata.info/en/

> You are assuming
> a) that an arbitrary best practice definition is relevant for OSM

It's not "arbitrary". You can find TBL's credentials here:

   https://en.wikipedia.org/wiki/Tim_Berners-Lee

> b) that we get any real life brownie points for this outside of academia

I don't give a fig about (and most certainly did not refer to) "brownie points".

>> and that's what including Wikidata iDs in cases such as the above does.
>>
>> In fact, since that makes OSM more useful and thus more attractive to
>> re-users, it *does* matter to OSM.

> You still haven't given any specific real world use case in which
> brand:wikidata would actually be helpful and somebody would -really-
> want to consume it.

You didn't ask for one. But since you now do:

* A data consumer wants to render a map with logos of shops and restaurants

* A data consumer wants to link to Wikipedia articles about same, in
the local language

* A data consumer wants to calculate and compare the distances between
(or density of) the outlets of a particular brand, in different
countries, regardless of the script in which the plaintext name of the
brand has been entered.

> Outside of than that it would obviously be good for
> wikidata and for people going to linked data conferences.

That's the second time you've mentioned a supposed benefit to
Wikidata; any such benefit is minimal, and certainly not my - nor, I'm
sure, Yuri's - reason for wanting to fix the issues outlined in the
OP.

The conference comment is pure snark.

> My take is that it adds a nearly impossible to maintain (consider your
> own Woolworth's example), non-speaking, foreign key for additional
> information that is already  that is already quite adequately provided
> by the brand key.

Your take is mistaken, as shown previously.

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

sabas88
In reply to this post by Frederik Ramm


2017-09-27 15:59 GMT+02:00 Frederik Ramm <[hidden email]>:
Hi,

On <a href="tel:27.09.2017%2015" value="+12709201715">27.09.2017 15:37, Simon Poole wrote:
> My take is that it adds a nearly impossible to maintain (consider your
> own Woolworth's example), non-speaking, foreign key

We generally discourage foreign keys (that are only usable together with
a different data set and that are not signposted locally).

When Wikidata keys were first added to OSM, I thought this was something
limited to place names of a certain importance and I didn't object.

Seeing that this now leads to the automatic assumption that Wikidata IDs
are practically admissible *everywhere* where Wikidata has defined an
ID, I am having second thoughts about the whole issue.

In theory, almost everything we map could be expressed by a Wikidata ID.
If welcoming a Wikidata link on a city place node means that by
extension I also have to welcome "amenity:wikidata=Q123456" on something
that is, say, an ice cream parlour because Q123456 is the generic
Wikidata category for ice cream parlours,

This is wrong also for me. The "translation" between tagging systems should be kept externally.
 
then I think I'd rather not have any Wikidata links in OSM at all.


Bye
Frederik

--
Frederik Ramm  ##  eMail [hidden email]  ##  N49°00'09" E008°23'33"


_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk



_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Marc Gemis
In reply to this post by Frederik Ramm
I fail to understand how an external database can link to an OSM
location in case we do not allow foreign keys.
I know there is some vague "find something with a name similar to X in
some area Y" kind of strategy, but did somebody ever implemented such
a thing ?
I doubt that "area Y" is always known.

Suppose a tourist agency has some additional information an an hotel.
How can they link that information to the OSM object ? Do they have to
store the coordinates, then do a query for hotel with name X around
that location ? Is that the best method ?

Just curious.

m.

On Wed, Sep 27, 2017 at 3:59 PM, Frederik Ramm <[hidden email]> wrote:

> Hi,
>
> On 27.09.2017 15:37, Simon Poole wrote:
>> My take is that it adds a nearly impossible to maintain (consider your
>> own Woolworth's example), non-speaking, foreign key
>
> We generally discourage foreign keys (that are only usable together with
> a different data set and that are not signposted locally).
>
> When Wikidata keys were first added to OSM, I thought this was something
> limited to place names of a certain importance and I didn't object.
>
> Seeing that this now leads to the automatic assumption that Wikidata IDs
> are practically admissible *everywhere* where Wikidata has defined an
> ID, I am having second thoughts about the whole issue.
>
> In theory, almost everything we map could be expressed by a Wikidata ID.
> If welcoming a Wikidata link on a city place node means that by
> extension I also have to welcome "amenity:wikidata=Q123456" on something
> that is, say, an ice cream parlour because Q123456 is the generic
> Wikidata category for ice cream parlours, then I think I'd rather not
> have any Wikidata links in OSM at all.
>
> Bye
> Frederik
>
> --
> Frederik Ramm  ##  eMail [hidden email]  ##  N49°00'09" E008°23'33"
>
>
> _______________________________________________
> talk mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/talk
>

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Andy Mabbett
In reply to this post by Frederik Ramm
On 27 September 2017 at 14:59, Frederik Ramm <[hidden email]> wrote:

> We generally discourage foreign keys

We do? Citation please.

> If welcoming a Wikidata link on a city place node means that by
> extension I also have to welcome "amenity:wikidata=Q123456" on something
> that is, say, an ice cream parlour because Q123456 is the generic
> Wikidata category for ice cream parlours, then I think I'd rather not
> have any Wikidata links in OSM at all.

This is a slippery-slope fallacy. Such equivalence should be entered
one, on the applicable key= or tag= page of the OSM wiki, like this:

   https://wiki.openstreetmap.org/w/index.php?title=Tag:amenity%3Dice_cream&diff=1510821&oldid=1424407

just as it it should be recorded once in Wikidata, on the
corresponding item, using the "OpenStreetMap tag or key" property,
P1282, like this:

   https://www.wikidata.org/wiki/Q1311064#P1282

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Jean-Marc Liotier
In reply to this post by Frederik Ramm
On Wed, 27 Sep 2017 15:59:34 +0200
Frederik Ramm <[hidden email]> wrote:
>
> "amenity:wikidata=Q123456" on something that is, say, an ice cream
> parlour because Q123456 is the generic Wikidata category for ice
> cream parlours

I thought wikidata tags were for unique objets, which usage I believe
is welcome... If they are applied to categories then it is something
else entirely.

I would agree that the matching of a generic Openstreetmap tag to a
Wikidata identifier is Wikidata's business... But then maybe a
Wikidata-centric view of the model will consider it is Openstreetmap's
business...

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

John F. Eldredge
In reply to this post by Richard Fairhurst
The spatial information will tell you where each business location is; it
is not sufficient to tell you whether these are multiple locations of the
same brand, or two unrelated brands that share the same name and category
of business.


On September 27, 2017 6:51:32 AM Richard Fairhurst <[hidden email]>
wrote:

> Andy Mabbett wrote:
>> in different parts of the world
>
> IIRC OSM stores spatial information. I might be wrong.
>
> Richard
>
>
>
> --
> Sent from: http://gis.19327.n8.nabble.com/General-Discussion-f5171242.html
>
> _______________________________________________
> talk mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/talk



_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: Fixing wiki* -> brand:wiki*

Christoph Hormann-2
In reply to this post by Frederik Ramm
On Wednesday 27 September 2017, Frederik Ramm wrote:
>
> In theory, almost everything we map could be expressed by a Wikidata
> ID. If welcoming a Wikidata link on a city place node means that by
> extension I also have to welcome "amenity:wikidata=Q123456" on
> something that is, say, an ice cream parlour because Q123456 is the
> generic Wikidata category for ice cream parlours, then I think I'd
> rather not have any Wikidata links in OSM at all.

I am inclined to concur.

The basis of accepting wikidata IDs in OSM originally was that it is
difficult to reference a specific feature from our database from the
outside and that wikidata provides stable IDs for real world objects
and that it can be of significant use for OSM data users to be able to
directly reference features in the OSM database describing the same
real world objects through this ID.  There are a number of
prerequisites for this however all of which have been put into question
in recent discussion:

* There would need to be a 1:1 relationship between OSM features and the
tagged wikidata ID (which apparently is often not the case for example
for populated places, island countries etc.).
* The wikidata ID would need to be stable (recent statements that the
wikidata IDs in the OSM database require constant maintenance to not
become stale indicate otherwise).
* The wikidata ID would need to be verifiable - a local mapper with
knowledge of the real world object represented by a certain OSM feature
would need to be able to falsify the wikidata ID based on information
readily available from wikidata (which does not always seem to be the
case either).
* The wikidata IDs are at least for the most part manually added and
verified by mappers with lokal knowledge - just like any other data in
OSM (which clearly does not seem to be the case any more - i have no
solid numbers here but certainly the majority of wikidata tags have
been added without individual verification by mappers with local
knowledge).

Everyone remember there is a big cultural difference between OSM and
wikipedia (and i assume wikidata can be included there).  OSM is
founded on local knowledge and original research while wikipedia
rejects original research and values secondary sources of information.  
This fundamental difference also translates into differences in data
models and different approaches to solving problems.  It is a good idea
not to try pretending these differences do not exist and that you can
intermix the two worlds without problems.

--
Christoph Hormann
http://www.imagico.de/

_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
1234