[OSM-talk] Announcing Remap-a-tron
Martijn van Exel
m at rtijn.org
Sat Sep 1 04:17:51 BST 2012
I want to add The Remap-A-Tron to the ever growing list of tools
designed to support the ongoing remapping effort.
Although this tool currently only covers the US, it would be
relatively straightforward (given proper server resources) to deploy
it for other regions or even worldwide, which is why I include talk@
What it does is pretty straightforward - at least for the user. It
serves up a Mapnik map with an overlay showing one redaction-deleted
major road (motorway - tertiary). By looking at the Mapnik map you
should be able to quickly determine whether that way has already been
remapped. If it is, hit the 'flag as remapped' button and you're on to
the next deleted way. If it isn't, you can either remap it straight
away - there's links for JOSM as well as Potlatch - or skip it and
leave it for someone else. There's buttons as well as shortcut keys to
make cycling through the deleted ways as efficient as possible.
The selection of deleted ways is pseudo-random so you might find
yourself looking at the same way twice in one session.
But wait - all deleted ways? Aren't many of them already remapped?
Yes, they are. This is where my Fairy Dust alogorithm comes in. It
detects with some accuracy if a way has already been remapped so you
should see mostly non-remapped ways. If enough people (currently 3)
agree that a way has already been remapped, it won't show up again.
Some more background on this tool is here -
In the two days since I first soft-launched this through my blog, IRC
and twitter, the number of non-remapped major class ways in the US has
gone down from 1863 to 1692, and my hope is to get that number down to
zero in a number of days. If and when that happens, I can easily tweak
the tool to include non-major roads (there are some 15K of those not
remapped in the US alone, or so Fairy Dust tells me). The tool can
also be repurposed for other (re-)mapping tasks altogether - let me
know your ideas.
Let me know if it's useful / how it can be improved. The code will be
on github soon but I can tell you now it's not very pretty.
martijn van exel
More information about the talk