[talk-au] [OpenStreetMap] intersections

Stephen Hope slhope at gmail.com
Thu Dec 1 02:28:14 GMT 2011


On 30 November 2011 21:51, Anthony <osm at inbox.org> wrote:

> And in the case of something like sparseness I think you can come up
> with at least equivalently "correct" results, by developing the
> algorithm hand in hand with input from a "good human mapper".


Oh, definitely.  That's why I think we can make one good enough.

 It seems to me that what you are saying here is not that one of the

> suburbs is more important than the others, but that the group of them
> is commonly referred to by the name of the one.
>

Pretty much. Not quite that, necessarily, but more if there is only going
to be room for one name, that's the one everybody would expect. It's not
more important, so much, as better known.

But based on what you say about the weather reports and traffic radio,
> it doesn't seem to me that "importance" is what we're talking about.
>
> My question would be - is it common to refer to a location within a
> particular suburb as part of [the name of some other suburb]?  If so,
> then I don't think it's "importance".  At least, it's not the
> importance of the suburb, but rather the importance of the name.
>

That's pretty much it. I usually refer to my work place as being in the
suburb next door to the one it really is in - many more people know where
that is (even if just "oh, somewhere on the north edge of the city") than
have heard of my real location. Only if I have to give an address do I
actually list the real suburb.

For each small grouping of suburbs/localities, if you asked a bunch of
people from within say 100km which of their names should be shown if there
was only room for one, you'd generally get a pretty good consensus. How we
get our renderers to know that is the problem.  I'm not sure relations are
the best solution - I'm not saying that A should always replace B & C, just
if I have to pick between them I'd pick A.  If I can fit both A and C, then
that's fine.

I haven't come up with a good solution. There's no objective, visible on
the ground, easily verifiable data I can think of that we can map to handle
it.  But the effect is real, and if we can utilise it, it will make our
maps feel better.


Stephen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-au/attachments/20111201/fd9225b1/attachment.html>


More information about the Talk-au mailing list