[OSM-talk] administrative boundaries and is_in

Michael Collinson mike at ayeltd.biz
Thu Jan 10 10:14:13 GMT 2008


At 10:11 AM 1/10/2008, Robin Paulson wrote:
>On 10/01/2008, Michael Collinson <mike at ayeltd.biz> wrote:
> > >which seems counter-intuitive, not to mention requiring huge amounts
> > >of work. do we set this for every item - roads, churches,
> > >supermarkets,....thousands of other items?
> > >is there anything underway to enable OSM to calculate where an object
> > >is, based upon knowledge of administrative boundaries - after all,
> > >they are only a polygon-shaped bounding box?
> >
> > Yes, sort of.  But the other way around, I am working on deriving
> > administrative boundaries from "is_in" and "place" tags. *If* it
> > works, the answer to your main question would be to randomly use
> > is_in tags on low level items such as roads and churches and let the
> > computer work out a boundary around them.  I should be able to report
> > back in February.
>
>sorry mike, i'm sure you've put a lot of work in, but that sounds even
>more backwards, and very difficult to control well - i foresee a lot
>of fudging to make it work in a lot of areas.
>
>the boundary data should be relatively easy to come by, and require a
>*lot* less points to be drawn/edited, than your method. plus, your
>method is never going to be one hundred percent perfect. for example,
>what happens in areas of open country side, with no POIs to mark, when
>the boundary changes direction - something that will happen a lot?
>and what when a place is in two administrative boundaries? there are
>fairly major cities that straddle for instance state lines in the US.
>i think arkansas may be one?
>
>but i guess itches should be scratched....something will doubtless
>come from this, whether unintended or not. ok, it'll be interesting to
>see your results

Robin, yes all your points are valid except I think getting accurate 
boundary data globally without violating copyrights will prove a long 
drawn out process.  Hope I'm wrong!  Meanwhile, mining the is_in mass 
observation resource indeed gives me an itch I've been longing to scratch.

I should have made one important qualification and that is my main 
interest is really "is near" rather than "is in", so I'm not so 
worried about precise boundaries - only place in the world, the rough 
areal extent and a reasonable low resolution approximation of shape, 
in that order of priority.  I think this is very important for 
searching.  On a local scale, it probably more useful to search for a 
street on the outskirts of Balham rather than determining for sure 
whether it is in Balham or the next suburb. At a wide scale, I'm 
hoping it will provide an easy method to zero in on San Francisco 
California when searching for "San Francisco, USA" without returning 
small villages in Spain and the Philippines.

Mike





More information about the talk mailing list