Serge Wroclawski <emac...@gmail.com> writes: > On Mon, Feb 21, 2011 at 3:58 PM, Peter Budny <pet...@gatech.edu> wrote: > >> If automated edits are problematic, it's not because the robot >> apocalypse is coming. It's because automated edits are hacked together >> due to a lack of tools and support in OSM for doing anything other than >> manual editing. This isn't a problem with automated edits; it's a >> problem with OSM. > > Peter, you've done an excellent job of identifying problems with OSM. > Do you have any suggestions on how to resolve these issues?
I would be naive to think that I have all the answers, or even some of them. It's a lot easier to point out problems than to solve them. However, you were kind enough to ask, so let me try to do some brainstorming: One big issue is that there's no way to combine two disparate sets of data any way but manually. Suppose TIGER 2015 data comes out and has centimeter-accurate centerlines, but still has all the typos and other nasty things that come with TIGER data. It would be great to use the centerlines but keep OSM's edited data for everything else. Or suppose I somehow get a database of every McDonald's location in the world, complete with addresses, phone numbers, etc. Rather than asking editors to pore over the whole map manually, it would be better if there were some kind of merge toolkit that can select portions of multiple data sets and combine them intelligently, automating the process when there are no conflicts (like a CVS merge). Of course the process needs to be guided by humans, but they shouldn't have to select or trace every single road. Here's another issue that ought to be much easier to solve, but hasn't been: the relation analyzer. http://ra.osmsurround.org/analyze.jsp?relationId=36947 That particular relation follows the "correct" practice of using role=(blank)/forward/backward, but even this isn't recognized by the analyzer. What it ought to show is just two chunks: from point A to B, and from point B to A. With the current display, how is one supposed to figure out if relation is completed or not? What about GPS tracks... how does one deal with those? The only interface I know is to click on "GPS Traces" from the slippy map, but that just gives me a list, and shows every trace in the whole world. I only want to see traces for the region I'm looking at, like "show me every GPS track within 100m of this highway". I see there's a wiki page on change monitoring, but it doesn't look like there's anything very sophisticated. Perusing the history when there are so many edits that are -180 to 180 is a big turn-off. If mappers are supposed to watch their hometown for changes, shouldn't they be given tools to do that with? Like I said, this is just brainstorming. I'm sure other users also have ideas for tools that would make the editing process a lot less tedious, without even touching on the topic of automated edits. Tools like these can benefit everyone (unlike a discussion on how to tag lumber yards or hotdog carts or whatever the question du jour is). -- Peter Budny \ Georgia Tech \ CS MS student \ _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk