[Talk-us-massachusetts] Import etc.

[Angela Morley]

(Did you know this listserv won't accept messages greater than 40kb?)

Very nice! I'm going to have to check that plugin out. I'm curious at what stage of the import that plugin would be most helpful.

I've been struggling to learn all of these steps involved in this process. It's just one thing at a time to get through it all. I finally figured out how to get JOSM to download large areas. You have to enable Expert Mode in the bottom left corner of the Preferences dialog, and then click the regular download button on the hotbar, and a new tab for Overpass API appears. I was able to download full town sized blocks by doing a wizard query like " building=* in "georgetown massachusetts" ".

I've been playing around with sandbox imports of the process I discovered with QGIS and JOSM with conflation plugin. I personally couldn't get the python script someone referenced to work -- I'm on Windows, and I barely got Python installed. Now I'm getting errors about GDAL missing, and all the guides online are about linux. Frankly, it's frustrating me too much.

Back on topic... I'm getting pretty close to working out the kinks with this process. I've discovered that occasionally Conflation is assigning the address points to the closest building node instead of to the building area itself, and I'm not sure how to restrict that just yet. I've taken a close look at the conflicts I've come across. They seem to be occasions where people were adding addresses to buildings, like shopping centers, that don't match the state's documentation. You can individually review each conflict within the Conflation plugin and decide what to do for each one. In Georgetown, I found five or six conflicts out of 2,600~ buildings in my testing.

What we should determine here is a standard for conflation/conflict resolution. Here's what I propose:
Importing addr:housenumber and addr:street
if addr:housenumber of A is different than B (
if A and B are sequential/odds/evens ( Concatenate to A;B )
if A and B are completely different ( assume manually mapped data is accurate, ignore import )
)

if addr:street of A is different than B (
if A and B are similar with misspelling ( Correct with import data )
if A and B are completely different ( assume manually mapped data is accurate, ignore import )
)

Thoughts?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us-massachusetts/attachments/20180721/3eaf6ec2/attachment.html>