Our plan is to follow in the footsteps of the NOLA building/address
import, which is currently in Task Manager.
http://tasks.openstreetmap.us/job/41
How/when can we get a similar task setup in Task Manager?
We will use their same exact process, and in fact I did a ward in the
NOLA task to test out how it would work, which was awesome!
http://wiki.openstreetmap.org/wiki/New_Orleans,_Louisiana/Building_Outlines_Import/Contributor_Guide
Currently we are working with the city gov to get into and will conflate
the buildings and address data together, similar to this:
https://github.com/mtoupsUNO/nola-buildings
We'll put our final code on Github too.
Since we'll be importing it in small chunks manually with JOSM, then
verifying the data by hand, hopefully we can catch most of the error
that may exist.
We have checked some of the areas qualitatively that we know that have
existing OSM data. Areas of the Downtown, Highlands and Jtown. No one
has found any issues yet. And while we won't be overwriting any OSM
building footprints from users, it does appear that the city data is
better in every case we can see (likely due to access to higher-res
satellite photography).
http://wiki.openstreetmap.org/wiki/Louisville,_Kentucky/Building_Outlines_Import/OSMvsGISOverlapExamples
For updatability, we will be adding a unique ID that comes from the
source data and will remain the same in updates. But really any other
updated data will be imported using this same method: creating a Task
and dividing it out into chunks to manually compare and validate using
our local volunteers.
I hope we are on the right path with all this. Any more questions or
advice is welcome and thanks for everything so far! We hope to have the
data ready and Task Manager setup by early next week.
*Michael Schnuerle* | Code for America Louisville Brigade Captain
/e/ [email protected] | @CivicDataAlly
<http://www.twitter.com/CivicDataAlly>
CivicDataAlliance.org <http://www.civicdataalliance.org/> | Open Data
Portal <http://data.civicdataalliance.org>
On 1/31/15 9:46 AM, Sander Deryckere wrote:
Hi,
I think a more elaborate documentation is needed before it can
actually be judged.
The workflow is unknown. What tools will you use to split up the task,
what tools will you use to detect conflicts, how will you handle
conflicts (f.e. the same address on a different position when you
compare OSM with city data)...
Next to that, you also need to do some research about data quality.
The building outline precision is hard to check with amateur hardware,
thus only a quick check is useful there. However, address databases
may contain many different logical mistakes. In our Flemish dataset,
I've encountered housenumbers being assigned to the wrong street
(sometimes a neighbouring street, sometimes a street with a similar
name), numbers where no building exists, new houses that were not yet
numbered, spelling mistakes in streetnames, overlapping housenumbers
that should go on different houses... You should do qualitative
research on an area you know, or that's well mapped in OSM already.
Quantitative research is hard here, because there's no data to compare
it too.
As a final note, you must also think about updatability. Once the data
is in, people will build on it, and you can't just swipe it out and
replace with a newer dataset once that's released. So if you import a
2012 dataset now, what will you do when the next dataset comes out?
You say that the housenumbers are updated daily, but how will you
handle that? As imports require a smaller community than normal
mapping, updatability is quite important when doing an import, since
there's about nobody left to do the job.
Updatability can be achieved by adding an id to the imported objects
(that id must be permanent in the source DB), but it's also possible
without id when you develop algorithms to compare OSM data with the
official data.
Thanks for working on this,
Sander
On 1/28/15 4:36 PM, Clifford Snow wrote:
> Are you going to conflate the addresses with the buildings ahead of
time? I usually take the following approach:
> o If one address inside a building outline, conflate the address to
the building outline
> o Multiple addresses inside individual buildings are kept as
individual nodes. One city I'm working on now actually include unit
numbers and main building number. In this case the main address is
conflated with the building outline and the units are kept as nodes
> o everything else is kept as delivered.
> Sounds like a good project. Good luck,
> Clifford
2015-01-28 20:55 GMT+01:00 Michael Schnuerle
<[email protected] <mailto:[email protected]>>:
Hi all,
Louisville KY's Code for America brigade has volunteered to help
the American Printing House for the Blind work with the OSM
community to import the building footprint and address data properly.
I'm coordinating the effort and wanted to reach out to this list,
per the Import Guildelines. I've created some OSM Wiki pages as a
start to compile all the info publicly.
Main Import page:
https://wiki.openstreetmap.org/wiki/Louisville,_Kentucky/Building_Outlines_Import
Added to this list:
https://wiki.openstreetmap.org/wiki/Import/Catalogue
We've got a number of members who have made some OSM edits over
the last few years. We are reaching out to more active local OSM
editors too. If there is anyone on this list who is an active
editor locally we'd love to have your support and help!
You can read more about the project here on our forum:
http://www.civicdataalliance.org/forum/?place=msg%2Fcivicdataalliance%2FJDkgh-uc2so%2FFoQZuypzZZEJ
I'd like to ask for help with making sure we do everything
correctly, and next steps. We want to follow in the footsteps of
NYC, LA, Chicago and NOLA who have all recently done something
similar.
We have a technically competent team of volunteers (programmers,
mappers, GIS, coders), and our short term goal is to get things
setup for a hackathon we are hosting Feb 21 called CodeAcross. At
CodeAcross we'll have a day to really get people focused on
loading the data in chunks into OSM.
I'm also sending this to the imports-us@ list. If I should focus
my communications there just let me know.
Thanks,
--
*Michael Schnuerle* | Code for America Louisville Brigade Captain
/e/ [email protected]
<http://[email protected]> | @CivicDataAlly
<http://www.twitter.com/CivicDataAlly>
CivicDataAlliance.org <http://www.civicdataalliance.org/> | Open
Data Portal <http://data.civicdataalliance.org>
_______________________________________________
Imports mailing list
[email protected] <mailto:[email protected]>
https://lists.openstreetmap.org/listinfo/imports
_______________________________________________
Imports mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports
