[Talk-ca] Merging OSM + Geobase

[William Lachance]

Hey all,

I've been following the discussions about the geobase import with
interest. There seems to be some concern about how to correlate the
existing OSM data with the Geobase dataset. In particular, how do we get
the superior positioning and connectivity of geobase without blowing
away all the great work people have done tagging their manually inputted
ways?

This topic is of some interest to me, as I've been using both GeoBase
and OSM data in a project of mine called hbus.ca (linky:
http://hbus.ca), which is basically a transit mapper for the Halifax
Regional Municipality. Right now, I'm using OpenStreetMap data for the
map view, and geobase information (the NRN road network) for the site's
geocoding and pathfinding capabilities. From my perspective, it would be
really great to get the GeoBase information in for Halifax, as OSM
coverage for the outer areas of the HRM is rather poor.

One alternative which I haven't heard mentioned is the possibility of
_merging_ the two datasets. That is, writing an automated script that
does the following for a given region:

1. Look up all named OSM ways. 
2. For each way:
 - Finds the corresponding GeoBase segments with the same name and the
same proximate positioning. Obviously some cleverness will be required
here. I've written a geocoder which should be able to help with the
naming bit. The positioning part is a computational geometry problem
which can be solved in a number of ways -- I'd need to do some
experimenting to see what works best.
 - Merge any tags from the existing OSM way (e.g. cyclespeed) which
cannot be expressed with the GeoBase road gml files.
 - Replace the OSM nodes with GeoBase ones.

As a disclaimer, I've never written any code which actually interacts
with the openstreetmap server (I've only done operations on raw OSM
files), so there might be a step or two I'm missing. Other than that,
can anyone see a reason why this wouldn't work? Ordinarily this is
something I'd just go ahead and try (I'm planning to give it a shot next
week), but I thought I'd bring it up here now since the topic is up for
discussion.
-- 
William Lachance <wrlach at gmail.com>