[OSM-talk] What Streets are in what Places

Peter Miller peter.miller at itoworld.com
Fri Nov 13 13:44:47 GMT 2009


On 13 Nov 2009, at 12:00, Florian Lohoff wrote:

> On Fri, Nov 13, 2009 at 07:43:20AM +0000, Peter Childs wrote:
>> However in OSM places are points not area's and the areas we do have
>> are either to do with admin (ie Counties, Borough's etc) and hence  
>> are
>> rather less than helpful.
>>
>> The is_in tag is not a lot of use either due to it being  
>> inconsistent.
>>
>> I could do it with some kind of nearest function but I don't think
>> this is going to give me the right answer either.
>>
>> What is in a place, is often to do with other geographic features,
>> like gaps in housing, majro roads etc, and also to do with history,
>>
>> What I would like to do is write a script that takes the planet and
>> gives a list of the places (towns, villages etc) and a polygon/area
>> for each place.
>>
>> While I don't think this list would be worth piping back in to the
>> database, it might be useful for knowing what were missing.....
>>
>> Maybe Places should be area's rather than points anyway.
>
> This is a quite complicated thing to do - Some parts of the world
> fill very quickly with admin boundaries and using them for searching
> in the garmins would definitly give them a strong boost which i'd like
> to see.

We already have data for all 50K places in Britain available for  
import into OSM as part of the NaPTAN dataset. [1] All of the 50K  
places in this dataset have a boundary (which is approximate) which is  
based only on whether a bus stop is in a place or not. Each place also  
has a geocode for its 'centre. Places can have parents and children so  
one can model suburbs,

Here are some examples of the data that is available:
http://www.itoworld.com/static/screenshots?product=NAPTAN&id=3
http://www.itoworld.com/static/screenshots?product=NAPTAN&id=1

Of course there is always 'creep' of desirable and undesirable places  
(where people say they live in the posh bit, not the non-posh bit next  
door - lots of room for negotiation of boundaries, but highly valuable  
and useful data all the same).

I was also talking just this morning about the value of this data and  
if such a boundary-of-places dataset could be produced for Northern  
Ireland within OSM (Northern Ireland is part of the UK but this data  
is not yet available within NaPTAN).

ITO has a new product for directing people by public transport and we  
need to know whether a bus stop is within a place or not - we can  
guess, but it would be much better not to guess and this is what we  
need to do that.

In short, I and my company and indeed Travline/DfT would strongly  
support to option to include boundary polygons for places within OSM  
together with a relation and a point geocode for the place.

Yahoo is also collecting useful data of this sort which I understand  
they are planning to release for reuse.


[1] http://wiki.openstreetmap.org/wiki/NaPTAN


Regards,


Peter


>
> The other point is that people are used to fuzzy or non exact  
> search. For
> example i am living in a Town called Rietberg which has a large area  
> it covers.
> There are multiple suburbs which are very distinct from Rietberg the  
> town
> itself. People are used to be able to search for either Rietberg or
> even the suburbs names e.g. "Mastholte" or "Varensell" - So in google
> ou can search for
>
> 	Alt Hammoor 38, Rietberg, Germany
> 	Alt Hammoor 38, Mastholte, Germany
>
> which is the same place - Postal wise the address is 33397 Rietberg
> but all in car navigations are happy to accept Mastholte aswell.
>
> This gets more complicated taking large citys like Berlin - where the
> individual districts are their own administration and therefor have  
> their
> own admin_level 8 boundary. People on the other hand are used to  
> search
> for Gorßbeerenstraßen, Berlin to work - e.g. spit out all the  
> Großbeerenstraße
> Berlin has (Mentioning the District - TomTom does this).
>
> So - it comes down that people are used to search for admin_level=10  
> names
> and content and for admin_level=6 Content which will then look for all
> content in admin_level=6+  ..
>
> So - i'd vote for strictly using admin boundarys and invent some tags
> for giving some hints - It does not make sense to include  
> admin_level=4
> or most of the time it does not make sense to include admin_level=6  
> for
> searching (Kreis Gütersloh for example makes no sense) - But sometimes
> it should be included that you look for
>
> 	Großebeerenstraße, Berlin
>
> and get
>
> 	Großebeerenstraße, Kreuzberg, Berlin
> 	Großebeerenstraße, Lichterfeld, Berlin
>
> aswell as beeing able to search for
>
> 	Großebeerenstraße, Kreuzberg
>
> Searching for place nodes and guessing is prone to errors and  
> hinting will be
> on a per street level which will be very hard to maintain and  
> include so ...
> Admin boundarys are there for a reason ...
>
> Flo
> -- 
> Florian Lohoff                                         flo at rfc822.org
> "Es ist ein grobes Missverständnis und eine Fehlwahrnehmung, dem Staat
> im Internet Zensur- und Überwachungsabsichten zu unterstellen."
> - - Bundesminister Dr. Wolfgang Schäuble -- 10. Juli in Berlin
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk





More information about the talk mailing list