[Talk-ca] Scripts to import GeoBase/GeoGratis into OSM

Richard Degelder rtdegelder at gmail.com
Wed Dec 10 03:40:17 GMT 2008


We are hearing a lot about the scripts that will will be used to import
the data into OSM from GeoBase/GeoGratis but at no point do I recall
anyone pointing to such a script, or a number of scripts, that will be
used to do the work.  And unless I am writing it, or someone else is
willing to write it or provide it, I cannot be sure what is possible or
not.

So we have to look at what is possible.  Is it possible, and how is the
best approach to adding the data from GeoBase/GeoGratis into areas that
already have some OSM data, such as roads, within them?  This question
has to be answered for whatever density of roads already exists, be it a
single road or a massive number for places like Toronto, because the
conflict resolution just has to work.  How did other imports, such as
the TIGER import, deal with data that was already there?  What did they
do to match existing data within OSM when it was imported?  And how did
they add nodes to existing roads when a new road was added from the
import?

To what extent do we want to import data into OSM from
GeoBase/GeoGratis?  A very basic import that we agreed would be useful,
and a good start to test the import, would be geopolitical boundaries.
I am also pretty certain that adding the roads from GeoBase/GeoGratis is
also a pretty high priority, especially where the street names and house
numbers are also included.  If those are the basic initial needs we can
apparently get the most current data from GeoBase so lets concentrate on
using it for now.

Is a script going to be able to determine that a roadway within
GeoBase/GeoGratis is the same street as within OSM or if it is a
different street?  To some degree, and at this point I am only guessing,
we can get a script that will be able to determine that a short straight
street could be the same but what if the street is longer and not so
straight?  And what happens when something has happened, such as it was
lengthened to the or partially relocated, since the data was input into
GeoBase/GeoGratis?  What kind of things are going to appear within OSM
then?  What happens if a street was closed since the data was entered
into GeoBase/GeoGratis?  Are we going to get a new street there again?

At this point we are going to have to either develop a script, or more
likely a set of scripts, to perform the actions we want in order to
import the data into OSM or we are going to have to acquire, with the
legal right to do so, the scripts we need.  That means that if it was
done in the past by someone then we have to have the legal right to use
their scripts, and that means a license even if it is something like the
GPL, and to modify it as needed or write our own.

To some extent our wishlist, get everything we can for OSM, has to be
tempered with what is realistically possible, we can decide to take a
longer term approach to adding things that are going to be more
difficult so we can develop our expertise in importing those features
later.  And what is realistically possible is going to be determined by
the scripts that we develop or acquire to import the data and to the
extent they must massage the data to conform to OSM.  Other projects had
the benefit of a clean slate to start with and so they did not have to
import data into areas that already contain data.  So how did large
imports of data, like the TIGER import in he United States and and AND
import into the Netherlands, do it?  What limitations did they run into
and how did they resolve importing data into areas that were extensively
done?

Richard Degelder
rtdg





More information about the Talk-ca mailing list