[OSM-talk-be] Nominatim administrative boundaries

[Glenn Plas]

On 06/16/2013 07:30 PM, Kurt Roeckx wrote:
> On Sun, Jun 16, 2013 at 07:02:01PM +0200, Glenn Plas wrote:
>> Kurt (a.o),
>>
>> I checked the Rotselaar/Werchter setup and I made a single change to
>> the Rotselaar relation:
>>
>> http://www.openstreetmap.org/browse/relation/214462
>>
>> The only thing I think was missing is adding the Werchter boundary
>> relation as a 'subarea' to the Rotselaar one.
> Thanks for fixing that.

No problem, it's a joint effort isn't it ?
>
>> Did the same setup for Rijmenam/Bonheiden.  There aren't many
>> 'fusiegemeentes' being mapped -unfortunately- although it would be
>> highly interesting to have them, not only from a nominatim (search)
>> point of view, but also for addressing in general.
> The problem is ussually finding the right data, and making sure
> that there is no license problem with adding that data.
>
True, concerning the license on this sort of data, I'm feel this should 
somehow be open data by nature.   Only problem is finding an absolute 
reliable source indeed

>> The change I made will probably trigger some changes in the
>> nominatim search result in a few days , I now expect that Leuven
>> will be replaced by Rotselaar in the search result set when looking
>> up Werchter in a few days.
> Nominatim shows all the levels, including those I think it
> shouldn't.  It currently shows:
> Werchter, Rotselaar, Leuven, Flemish Brabant, Flanders, 3118,
> Belgium, European Union
I see what you mean but that's the administrative hit.  And it's pretty 
realistic.   If you try 'werchter' search on google maps you only get 
back 1 hit (the city).  Nominatim shows 2 hits, and the 
non-administrative hit is comparable to the Google result with a big 
exception, the postal code is missing from nominatim.  (looking at 
http://nominatim.openstreetmap.org/details.php?place_id=367851 ).
>
> I just noticed that if you look up "3118 Rotselaar" in google
> maps, that it properly shows what Werchter looks like, because
> it has it's own postal code.  It's more correct than what I
> added.
Strange, It looks pretty good indeed now, I was looking at something 
else before where I noticed a gap above Werchter (still in Rotselaar).  
The upper borders didn't match... But it sure looks ok in google maps.

Now I wonder where that data comes from now.  (and who paid for it)

Glenn

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-be/attachments/20130617/fe579fef/attachment.htm>