[Talk-ca] Importing geobase data, learning from everyone

[Sam Vekemans]

Cool, i'll add the point about the NID's to the wiki, as thats a
universal principle for all imported data, to trace it back to the
source.

The issue here is that OSM is a sandbox, where all points get moved,
changed, added, & ommited (in the case of adding roads that exist
already).
The question is: how to deal with the duplicates? And how to deal with
geobase updates?

(I mentioned on an earlier message.)

lets have a 2 part process..
1 load all the data, with its id's, skipping those most mapped areas.
2 -create a geobase plug-in, which shows all the geobase data as a
distinct, updatable, layer. -  so for the almost covered areas, it can
be traced. Can we expand on this idea?

Hope it makes more sence.
Sam

On 11/27/08, Michel Gilbert <michcasa at gmail.com> wrote:
> 2008/11/27 Dale Atkin <datkin at ibycus.com>
>
>> I agree. This isn't an easy problem to solve (assuming I'm seeing the
>> right
>> problem). I think you'd probably be best off to look at passing through
>> the
>> NIDs that geobase uses in to your internal database, and using that as a
>> cross reference.
>>
>> Dale
>
>
> Right Dale. The NIDs is the key here to join with Geobase attributes. The
> NIDs will not change from one version to the next. So it will be easy to
> detect the changes. Geobase provides (as a file) the changes between the
> last 2 versions. Also, the NID will be necessary to add Road name and
> address ranges when they will be available for the provinces that have only
> basic attributes for now.
>
> Michel
>
>>
>>
>> -----Original Message-----
>> From: Dave Hansen [mailto:dave at sr71.net]
>> Sent: Thursday, November 27, 2008 2:30 PM
>> To: Sam Vekemans
>> Cc: Corey Burger; Dale Atkin; Jason Reid; Michel Gilbert; Richard Weait;
>> simon at mungewell.org; talk-ca at openstreetmap.org
>> Subject: Re: [Talk-ca] Importing geobase data, learning from everyone
>>
>> On Thu, 2008-11-27 at 11:20 -0800, Sam Vekemans wrote:
>> > (aside: What about converting rejected data to 'highway=road'?)
>> > when new geobase data is available, the script would be the same, only
>> > look at the road name and road class in close proximity, and add whats
>> > missing.
>>
>> This is a very easy thing to say, but exceeding hard to accomplish in
>> practice.  We've thought about it extensively for the newer TIGER
>> revisions, but haven't come up with anything that we're happy with.
>> There are just plain too many corner cases to deal with.
>>
>> -- Dave
>>
>>
>