[Talk-us] Addresses and Tiger

Alan Millar am12 at bolis.com
Tue May 26 15:18:35 BST 2009


> I have been playing around with the TIGER 2008 data, which, for some of
> the counties around Atlanta seems to be much better than the old data.
> If I import it for some of the counties, though, I would like to import
> the addresses in a format that would be usable for routing software.
> The most popular schema seems to be the Karlsruhe Schema, but making use
> of it would require generating three OSM ways for each TIGER way - one
> on either side to represent the houses on that side of the road.  That
> seems unnecessarily complex, but there does not seem to be any widely
> accepted schema that places the data on the way itself.  Does anyone
> have any thoughts on that question?

I had the same thoughts on the Karlsruhe schema.  Having 3 ways for street
and address interpolation does seem complex.  But when I look at OSM areas
mapped using it, and also recall problems I've encountered with geocoding,
I think the Karlsruhe schema works pretty well.  (At least if we keep full
street information on the addressing ways, and not just the house
numbers.)

I've also been thinking about how to retrofit the addressing data to
existing areas which have already been edited and/or adjusted.  Having
multiple TLIDs per way, and then having ways split and/or recombined, will
make it difficult.  Just having multiple TLIDs per way could be hard, when
the way has been moved from its original Tiger location.  We'll have to
interpolate the interpolation points, I guess.  Hmmm....

- Alan






More information about the Talk-us mailing list