[Tagging] Topographic place names

Andrew Guertin andrew.guertin at uvm.edu
Thu Dec 12 11:09:47 UTC 2013


On 12/12/2013 05:53 AM, Martin Koppenhoefer wrote:
> 2013/12/12 Steve Bennett <stevagewp at gmail.com>
>>> IMHO it would be nice to have an alternative dataset in lower zoomlevels
>>> for geographic regions and extended/blurry features, something like a set
>>> of shapefiles with translations into all languages we can provide,
>>> something similar to what natural earth data provides, but distributed and
>>> modified/translated by us, not just English and for higher zoom levels
>>> (i.e. more detailed) than what NE has. Still we could start with their
>>> geographic regions dataset and refine it, as "All versions of *Natural
>>> Earth* raster + vector map data found on this website are in the public
>>> domain."
>>>
>> Are you saying that this kind of data is a poor fit for OSM itself?
>
> yes, for the reasons described above: no clear boundaries / fuzzy borders.
> A solution could also be a new datatype in OSM for fuzzy objects, (e.g. a
> collection of objects and the consumer would create a hull area around
> them, possibly also roles for objects that are to exclude), but at least
> currently this kind of stuff does not fit into how osm works.

Many villages or other small human settlements have no clearly defined 
boundaries, and we just represent them as a node. Similarly, many 
objects (say, shops) DO have clearly defined boundaries, but only have a 
node in OSM. In both cases, it's understood that the thing is an area, 
and the node means "it's somewhere around here".

Those are common examples of nodes representing fuzzy objects, and I see 
no reason that a way couldn't also be fuzzy. Just as with nodes, it 
would be up to the consumer to either understand the level of fuzziness, 
ignore the feature entirely, or pass it through and let a human 
interpret it.

--Andrew



More information about the Tagging mailing list