I would really like to at least further pursue the feasibility of
getting the POI info in there now in one import step instead of multiple
ones. Its definitely going to involve manual review - as will doing it
in multiple steps - so seems to me, especially since its such a large
number of features, we pull in as much data as reasonably possible on
the first go. What makes this really nice in Miami-Dade is the POIs and
addresses have a common key for joining - folio (PIN). That is a big deal.
Postgres / postgis is ideal for the data joining, spatial analysis, some
text cleanup, etc and Levente looks like you are already most of the way
there with earlier code. Leif - your approach is cool too, though!
I think hopping on a slack channel would be great. I've got some
graphics I've been putting together showing different scenarios of
positional accuracy, completeness, etc for the different data sets and
would be good to discuss these issues in a more real time / chat session
type scenario.
Brian
On 9/17/2018 2:16 PM, Levente Juhász wrote:
This sounds great so far. I like the idea of doing three separate
imports (address, building, and poi). Of course, as Frederik
mentioned, the Wiki should reflect these changes in the original plan
if we decide to implement them.
As for the community engagement part, I am down for bringing Maptime
Miami alive again! We used to organize these events monthly and we
reached a couple students from the greater Miami area. We also had
quite a few participants from other groups, such as Code for Miami. I
too think that it would be beneficial to reach as many people as we
can and introduce them to OSM. Maybe we can also come up with a more
beginner friendly way of dividing the county into TM tasks. Currently,
they might be too big for someone new to OSM.
I believe we could use my old tool [1] to pre-process the buildings.
In addition to the initial list Leif started to put together in his
other mail (height conversion, overlap checks, etc), we will also need
to look for other things (i.e. buildings overlapping with the
road/railway network). Other quality checks should include eliminating
geometry errors and others. For example, the Large Building dataset
from 2 years ago contained several self intersecting buildings that
need to be fixed. The tool [1] should be able to catch/fix these
issues and is also expandable in case we need to implement something
else. I'm happy to do this part.
I have not given the POI import much thought yet but it sounds good if
we can come up with a feasible plan.
It seems that the original plan changed quite a bit. I'm wondering if
we should regroup and discuss everything before moving on to the next
step instead of jumping right in the process? Maybe we should have an
informal discussion on Slack or something before coming back to the
mailing list with a final plan.
Cheers,
Levente
[1] https://github.com/jlevente/MiamiOSM-buildings
On Mon, Sep 17, 2018 at 2:54 AM Frederik Ramm <[email protected]
<mailto:[email protected]>> wrote:
Leif,
On 16.09.2018 21:47, Leif Rasmussen wrote:
> Import wiki
> page:
https://wiki.openstreetmap.org/wiki/Miami-Dade_County_Address_Import
It appears this page has been copied-and-pasted from a "Durham
Country"
import page, as a couple of references to Durham remain.
Can you fix that, and make really sure that the rest of the page
actually reflects your planned process and doesn't contain other
left-overs of the copy-and-paste process that are not applicable
to the
actual import?
In your email you write
> it is technically possible to just import the addresses without
any manual checks.
What does that mean exactly - are actually planning to import the
addresses without any manual checks, or why do you write that? The
wiki
page speaks of "merging addresses with amenities and buildings
manually".
For an import to be imported without manual checks, one would expect a
number of additional controls to be carried out, like checking if
one of
the address points to be imported sits in the middle of a road or
waterway which would then point to either the address point or the
other
feature being inaccurate, etc.
Generally, when someone proposes an import, what I love to see is
a plan
of community engagement, i.e. do you have any plans to involve (other,
in case you live in Miami-Dade) local mappers in the import? You write
that "the tasking manager will be used" but that doesn't say much, for
all we know it could be a lone Californian clicking through square
after
square ;) - if you have anything to say about this, expand the section
"Team Approach" on the wiki page.
If you decide to follow some of the ideas that Brian has mentioned in
his other post, of course then we're talking a wholly different import
and the documentation should be expanded accordingly, especially with
regards to any methodology applied to matching data from different
sets.
Bye
Frederik
--
Frederik Ramm ## eMail [email protected]
<mailto:[email protected]> ## N49°00'09" E008°23'33"
_______________________________________________
Imports-us mailing list
[email protected] <mailto:[email protected]>
https://lists.openstreetmap.org/listinfo/imports-us
--
Levente Juhasz
https://blog.jlevente.com/
_______________________________________________
Imports-us mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports-us
_______________________________________________
Imports-us mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports-us
