Re: [talk-au] Refreshing NSW place names from the GNB database.

[Andrew Davidson]

After a bit of digital archaeology I've found this thread:

https://lists.openstreetmap.org/pipermail/legal-talk/2007-July/000398.html

Which seems to be the point at which people thought they'd got 
permission (but there are doubters in the thread).


The bit I can't figure is why they'd agreed to tagging with the 
copyright notice but then the import seems to have gone ahead without 
this, see:


http://www.openstreetmap.org/changeset/507147

for example.

Then there is a review in 2010:

https://lists.openstreetmap.org/pipermail/legal-talk/2010-August/004136.html

which seems to have resulted in someone still believing that we had 
permission.




On 7/4/16 11:55, Andrew Harvey wrote:

As far as I can tell, this data isn't available under a free and open
license, so unless there is documentation somewhere to suggest
otherwise, it shouldn't have been imported to begin with and certainly
shouldn't be added again.

On 7 April 2016 at 11:30, Andrew Davidson > wrote:

There was an import of NSW places from the GNB database done back in
2008 with a helpful wiki page ;-)

http://wiki.openstreetmap.org/wiki/NSW_Geographic_Names_Import

I'm proposing to review these to see what's changed in the last 8
years but I've run into a number of problems:

1. It would seem that the original import was not complete
2. The nswgnb tags have not survived well
3. The GNB has "helpfully" created entries for the address
localities but these seem to have taken on the reference numbers for
the original town/village/city. They've created new entries for the
original entity but this means that the town/village/city now has a
different reference number.
4. Sometimes the locality entry has the same location but at other
times it can be separated by up to 5km.

Initially I thought that the multiple GNB references could be
entered with multiple values like this:

http://www.openstreetmap.org/node/3118349777

but this doesn't naturally tell you how the place:nswgnb and
ref:nswgnb line up and it doesn't lend itself to adding the
alt_names that are in the database. As an alternative I'd like to
use this scheme*:

http://www.openstreetmap.org/node/113135446

which requires swapping the namespace prefix around to make nswgnb
the namespace. I think this makes it clearer how the GNB categories
line up and can be extended for more names (There is at least one
place in NSW with three different variant names).

I'm also proposing to put the LOCALITY or SUBURB entry at the same
place as the corresponding TOWN/VILLAGE/CITY etc entry (provided
that it still falls inside the admin_level 10 boundary).

Any views on these ideas? I think the most important thing is will
this be useful for the next person who looks at this in 5 to 10
years from now?


*Unusually the admin_level 10 boundary for this area is called Lake
Tabourie and has a separate GNB entry:

http://www.openstreetmap.org/node/4103653600






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




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

Re: [talk-au] Refreshing NSW place names from the GNB database.

[cleary]

I think that "refreshing" the Geographic Names Board (GNB) placenames
would be unhelpful.

If you read the contributors page in the wiki, it is unclear that
appropriate permission was ever obtained to use the GNB data. It seemed
to me that the particular contributor was relying on a generalised
permission for personal and non-commercial use but  without specific
permission for OSM to use the data. I contacted GNB during 2015, seeking
to clarify and confirm permission but did not receive a final response. 
Subsequently, OSM was given specific permission to use to access
specified data from LPI NSW.GNB is now part of LPI NSW and the GNB
database did not form part of the data for which approval was given. So
I remain uncertain that OSM has the necessary permission to use the GNB
database.

A further reservation relates to the inadequacies of the data. Many of
the co-ordinates for locations and features are incorrect. The published
GNB locations of some place names are actually across state borders.
Most errors are small but still significant. The locations on the LPI
NSW Base Map are far more accurate.

Another reservation relates to the appropriateness of including all the
place names in the OSM map. It is a bit like including old railways
lines and stations that are long gone. The GNB database includes places
that were once settlements but are now just memories. For example, GNB
lists a locality called "Bogan" on the Gongolgon Tarcoon Road near the
Bourke Shire/Brewarrina Shire boundary,
(http://www.openstreetmap.org/node/113763363#map=17/-30.17587/146.49569)
 I drove along that road last year, looking for any sign of the place
and there are no buildings, no signs, nothing at all to be seen locally
- only an entry in the GNB database which had been added into OSM. I did
not remove the place from OSM but I think I may have edited the tags and
reduced it from a "hamlet" to a "locality" although I felt sorely
tempted to delete the node as there really was nothing there.

I have edited quite a number of GNB place names when undertaking other
edits. I have relocated the nodes to the actual sites as shown on
satellite imagery or LPI NSW Base Map and have altered the place tags
where it seemed appropriate.(e.g replacing the "hamlet" tag when there
are no residents anywhere in the area. Overall I prefer to use survey
data and the approved LPI NSW data sources and would be disappointed if
GNB data were to be "refreshed'.







On Thu, Apr 7, 2016, at 11:43 AM, Ian Sergeant wrote:
> Hi,
> 
> What are you actually trying to achieve here?
> 
> As I understand, the purpose of the original GNB update was slot in
> GNB names where OSM didn't already have coverage.
> 
> If there is already a town/village/suburb/locality in OSM, and it is
> already well located, then there is no issue that I can see.
> 
> What type of changes in the GNB have happened that you want to reflect in
> OSM?
> 
> I'm really not sure how putting a locality node bang on top of a town
> node is going to be anything put confusing.  I don't think it
> communicates any information at all.
> 
> Ian.
> 
> 
> 
> On 7 April 2016 at 11:30, Andrew Davidson  wrote:
> > There was an import of NSW places from the GNB database done back in 2008
> > with a helpful wiki page ;-)
> >
> > http://wiki.openstreetmap.org/wiki/NSW_Geographic_Names_Import
> >
> > I'm proposing to review these to see what's changed in the last 8 years but
> > I've run into a number of problems:
> >
> > 1. It would seem that the original import was not complete
> > 2. The nswgnb tags have not survived well
> > 3. The GNB has "helpfully" created entries for the address localities but
> > these seem to have taken on the reference numbers for the original
> > town/village/city. They've created new entries for the original entity but
> > this means that the town/village/city now has a different reference number.
> > 4. Sometimes the locality entry has the same location but at other times it
> > can be separated by up to 5km.
> >
> > Initially I thought that the multiple GNB references could be entered with
> > multiple values like this:
> >
> > http://www.openstreetmap.org/node/3118349777
> >
> > but this doesn't naturally tell you how the place:nswgnb and ref:nswgnb line
> > up and it doesn't lend itself to adding the alt_names that are in the
> > database. As an alternative I'd like to use this scheme*:
> >
> > http://www.openstreetmap.org/node/113135446
> >
> > which requires swapping the namespace prefix around to make nswgnb the
> > namespace. I think this makes it clearer how the GNB categories line up and
> > can be extended for more names (There is at least one place in NSW with
> > three different variant names).
> >
> > I'm also proposing to put the LOCALITY or SUBURB entry at the same place as
> > the corresponding TOWN/VILLAGE/CITY etc entry (provided that it still falls
> > inside the admin_level 10 boundary).
> >
> > Any views on these ideas? I think 

Re: [talk-au] Refreshing NSW place names from the GNB database.

[Andrew Davidson]

Hmm, yes that is a problem. It one of the more annoying things that I've 
found with OSM; that the level of documentation for "approval" is not 
consistent. I had been assuming that this had been approved on the basis 
of it being in the list of imports.


In that case is it a matter of removing the tags?

On 7/4/16 11:55, Andrew Harvey wrote:

As far as I can tell, this data isn't available under a free and open
license, so unless there is documentation somewhere to suggest
otherwise, it shouldn't have been imported to begin with and certainly
shouldn't be added again.

On 7 April 2016 at 11:30, Andrew Davidson > wrote:

There was an import of NSW places from the GNB database done back in
2008 with a helpful wiki page ;-)

http://wiki.openstreetmap.org/wiki/NSW_Geographic_Names_Import

I'm proposing to review these to see what's changed in the last 8
years but I've run into a number of problems:

1. It would seem that the original import was not complete
2. The nswgnb tags have not survived well
3. The GNB has "helpfully" created entries for the address
localities but these seem to have taken on the reference numbers for
the original town/village/city. They've created new entries for the
original entity but this means that the town/village/city now has a
different reference number.
4. Sometimes the locality entry has the same location but at other
times it can be separated by up to 5km.

Initially I thought that the multiple GNB references could be
entered with multiple values like this:

http://www.openstreetmap.org/node/3118349777

but this doesn't naturally tell you how the place:nswgnb and
ref:nswgnb line up and it doesn't lend itself to adding the
alt_names that are in the database. As an alternative I'd like to
use this scheme*:

http://www.openstreetmap.org/node/113135446

which requires swapping the namespace prefix around to make nswgnb
the namespace. I think this makes it clearer how the GNB categories
line up and can be extended for more names (There is at least one
place in NSW with three different variant names).

I'm also proposing to put the LOCALITY or SUBURB entry at the same
place as the corresponding TOWN/VILLAGE/CITY etc entry (provided
that it still falls inside the admin_level 10 boundary).

Any views on these ideas? I think the most important thing is will
this be useful for the next person who looks at this in 5 to 10
years from now?


*Unusually the admin_level 10 boundary for this area is called Lake
Tabourie and has a separate GNB entry:

http://www.openstreetmap.org/node/4103653600






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




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

Re: [talk-au] Refreshing NSW place names from the GNB database.

[Andrew Davidson]

On 7/4/16 11:43, Ian Sergeant wrote:

Hi,

What are you actually trying to achieve here?

As I understand, the purpose of the original GNB update was slot in
GNB names where OSM didn't already have coverage.

If there is already a town/village/suburb/locality in OSM, and it is
already well located, then there is no issue that I can see.

What type of changes in the GNB have happened that you want to reflect in OSM?


New place names get created, old names get discontinued, new variants of 
names get registered, spelling of place names are changed, dual names 
are registered, place names that for some reason have never made it to 
OSM are missing, alternative names that weren't included in the initial 
import should be included.





I'm really not sure how putting a locality node bang on top of a town
node is going to be anything put confusing.  I don't think it
communicates any information at all.


Did you look at the examples? I meant the opposite of that; there should 
only be one OSM place node even if that means having the LOCALITY or 
SUBURB entry significantly offset from the GNB lat/lon.


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

Re: [talk-au] Refreshing NSW place names from the GNB database.

[Andrew Harvey]

As far as I can tell, this data isn't available under a free and open
license, so unless there is documentation somewhere to suggest otherwise,
it shouldn't have been imported to begin with and certainly shouldn't be
added again.

On 7 April 2016 at 11:30, Andrew Davidson  wrote:

> There was an import of NSW places from the GNB database done back in 2008
> with a helpful wiki page ;-)
>
> http://wiki.openstreetmap.org/wiki/NSW_Geographic_Names_Import
>
> I'm proposing to review these to see what's changed in the last 8 years
> but I've run into a number of problems:
>
> 1. It would seem that the original import was not complete
> 2. The nswgnb tags have not survived well
> 3. The GNB has "helpfully" created entries for the address localities but
> these seem to have taken on the reference numbers for the original
> town/village/city. They've created new entries for the original entity but
> this means that the town/village/city now has a different reference number.
> 4. Sometimes the locality entry has the same location but at other times
> it can be separated by up to 5km.
>
> Initially I thought that the multiple GNB references could be entered with
> multiple values like this:
>
> http://www.openstreetmap.org/node/3118349777
>
> but this doesn't naturally tell you how the place:nswgnb and ref:nswgnb
> line up and it doesn't lend itself to adding the alt_names that are in the
> database. As an alternative I'd like to use this scheme*:
>
> http://www.openstreetmap.org/node/113135446
>
> which requires swapping the namespace prefix around to make nswgnb the
> namespace. I think this makes it clearer how the GNB categories line up and
> can be extended for more names (There is at least one place in NSW with
> three different variant names).
>
> I'm also proposing to put the LOCALITY or SUBURB entry at the same place
> as the corresponding TOWN/VILLAGE/CITY etc entry (provided that it still
> falls inside the admin_level 10 boundary).
>
> Any views on these ideas? I think the most important thing is will this be
> useful for the next person who looks at this in 5 to 10 years from now?
>
>
> *Unusually the admin_level 10 boundary for this area is called Lake
> Tabourie and has a separate GNB entry:
>
> http://www.openstreetmap.org/node/4103653600
>
>
>
>
>
>
> ___
> Talk-au mailing list
> Talk-au@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-au
>
___
Talk-au mailing list
Talk-au@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au

Re: [talk-au] Refreshing NSW place names from the GNB database.

[Ian Sergeant]

Hi,

What are you actually trying to achieve here?

As I understand, the purpose of the original GNB update was slot in
GNB names where OSM didn't already have coverage.

If there is already a town/village/suburb/locality in OSM, and it is
already well located, then there is no issue that I can see.

What type of changes in the GNB have happened that you want to reflect in OSM?

I'm really not sure how putting a locality node bang on top of a town
node is going to be anything put confusing.  I don't think it
communicates any information at all.

Ian.



On 7 April 2016 at 11:30, Andrew Davidson  wrote:
> There was an import of NSW places from the GNB database done back in 2008
> with a helpful wiki page ;-)
>
> http://wiki.openstreetmap.org/wiki/NSW_Geographic_Names_Import
>
> I'm proposing to review these to see what's changed in the last 8 years but
> I've run into a number of problems:
>
> 1. It would seem that the original import was not complete
> 2. The nswgnb tags have not survived well
> 3. The GNB has "helpfully" created entries for the address localities but
> these seem to have taken on the reference numbers for the original
> town/village/city. They've created new entries for the original entity but
> this means that the town/village/city now has a different reference number.
> 4. Sometimes the locality entry has the same location but at other times it
> can be separated by up to 5km.
>
> Initially I thought that the multiple GNB references could be entered with
> multiple values like this:
>
> http://www.openstreetmap.org/node/3118349777
>
> but this doesn't naturally tell you how the place:nswgnb and ref:nswgnb line
> up and it doesn't lend itself to adding the alt_names that are in the
> database. As an alternative I'd like to use this scheme*:
>
> http://www.openstreetmap.org/node/113135446
>
> which requires swapping the namespace prefix around to make nswgnb the
> namespace. I think this makes it clearer how the GNB categories line up and
> can be extended for more names (There is at least one place in NSW with
> three different variant names).
>
> I'm also proposing to put the LOCALITY or SUBURB entry at the same place as
> the corresponding TOWN/VILLAGE/CITY etc entry (provided that it still falls
> inside the admin_level 10 boundary).
>
> Any views on these ideas? I think the most important thing is will this be
> useful for the next person who looks at this in 5 to 10 years from now?
>
>
> *Unusually the admin_level 10 boundary for this area is called Lake Tabourie
> and has a separate GNB entry:
>
> http://www.openstreetmap.org/node/4103653600
>
>
>
>
>
>
> ___
> Talk-au mailing list
> Talk-au@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-au

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

[talk-au] Refreshing NSW place names from the GNB database.

[Andrew Davidson]

There was an import of NSW places from the GNB database done back in 
2008 with a helpful wiki page ;-)


http://wiki.openstreetmap.org/wiki/NSW_Geographic_Names_Import

I'm proposing to review these to see what's changed in the last 8 years 
but I've run into a number of problems:


1. It would seem that the original import was not complete
2. The nswgnb tags have not survived well
3. The GNB has "helpfully" created entries for the address localities 
but these seem to have taken on the reference numbers for the original 
town/village/city. They've created new entries for the original entity 
but this means that the town/village/city now has a different reference 
number.
4. Sometimes the locality entry has the same location but at other times 
it can be separated by up to 5km.


Initially I thought that the multiple GNB references could be entered 
with multiple values like this:


http://www.openstreetmap.org/node/3118349777

but this doesn't naturally tell you how the place:nswgnb and ref:nswgnb 
line up and it doesn't lend itself to adding the alt_names that are in 
the database. As an alternative I'd like to use this scheme*:


http://www.openstreetmap.org/node/113135446

which requires swapping the namespace prefix around to make nswgnb the 
namespace. I think this makes it clearer how the GNB categories line up 
and can be extended for more names (There is at least one place in NSW 
with three different variant names).


I'm also proposing to put the LOCALITY or SUBURB entry at the same place 
as the corresponding TOWN/VILLAGE/CITY etc entry (provided that it still 
falls inside the admin_level 10 boundary).


Any views on these ideas? I think the most important thing is will this 
be useful for the next person who looks at this in 5 to 10 years from now?



*Unusually the admin_level 10 boundary for this area is called Lake 
Tabourie and has a separate GNB entry:


http://www.openstreetmap.org/node/4103653600






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