[Imports] HIFLD

[Mike Thompson]

Hello James,

On Sun, Oct 2, 2022 at 3:19 PM James Crawford <sherbets at disroot.org> wrote:

> The imports listed on the wiki page are the datasets that I have deemed
useful for importing/obtaining metadata for OSM.
Thanks, I missed that.  My bad.  92 is still a large number.  I don't think
the community should have to review all of those.  Why not start with one
and do a really good job with it, and if successful, more on to others?

I do see that "public refrigerated warehouses" are on that list.  You
listed the quality as "good" and "good for import."  As detailed earlier,
that was not my finding, which leads me to believe that we have different
standards for "good", and leads me to question other cases where you said
quality was "good."  Also, as pointed out before, these cannot be verified
from imagery.  How does one tell from imagery if the warehouse is
"refrigerated"?  When you have a point in the middle of the street and
warehouses on both sides of the street, how does one tell if the warehouse
on the one side or the other is refrigerated, or if it is half a mile down
the street (not uncommon in these nationwide datasets unfortunately).

Also, I see "PAD-US" is "in progress." Does that mean that the import has
actually started?  I don't recall seeing it discussed on this list.


> I've been writing specific instructions on an individual dataset basis in
order to clearly translate tags in the database to tags in OSM. If I can
figure out the scripting, I will document it then as well.
There are methods of translating GIS fields to OSM tags that do not involve
scripting.  Whether these will be sufficient for your purposes remains to
be seen.  If they are sufficient, you would need to still document how the
translation was done (along with any configuration files), and provide us
with the translated data for review.  I would not rely on individual
local mappers to do the translation.  We will get 50 (e.g.) different
translations. Translate the data, obtain acceptance from the community,
then, if necessary, parse it out for individual local mappers to import.

Also, where are these "specific instructions"?  I didn't see them on the
wiki page.  Perhaps I missed it.

>
>
>> * I would recommend that where possible, the data be obtained from the
original source (other Federal agency, state, etc.). This ensures that the
most recent data is used (still might not be recent enough), and that
nothing has been lost in the process of merging/translating the data into
HIFLD.   I don't think HIFLD did any "value add" other than aggregating
various sources and perhaps some standardization (e.g. individual features
were not verified).
>
> I can document and do this where possible.
I think this is already documented in the HIFLD data's metadata.  What I
recommend is that we go back to the original source and get their data,
rather than using the HIFLD data.  This will not always be possible, but in
some cases it clearly will be.

Mike
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20221002/996956e2/attachment-0001.htm>