Hi again, just checked Seltjarnarnes. Took me not even 10min to check the data. Well, this was especially easy as the data in OSM was present and much more consistent and accurate. So I just needed to delete all the OFP streets... ;-)
I´m going to clean up more asap. kv. Christoph Am 17.01.2010 22:13, schrieb Christoph L. Hess: > I want to say something to this, too. I´m using JOSM and as described by > Ævar before the import, this was a easy to check all the data with the > help of the color set. Thanks go to the authors. > > I spent the last two days cleaning up Easticeland and parts of the > eastern mid-highland. OK, these were two hard days but at least there is > quite a lot new data now. A lot of ways that no one was driving before. > Check the ways in the 900 area (besides that, am I the only one adding > and caring the road system website on Icelandic wikipedia?!) > > There is in fact a lot of useful information in the data itself. And the > connecting of all the ways and reconnecting is a quite big amount of > work, but I´m sure you get through this. > > Anyways after spending some 35 hours the last two days I can tell one: I > deleted around 90% of the data because it was already on OSM (shows the > progress in OSM and the quality of the OSM data is IMHO in 95% of the > ways I checked far superior), 10% is a good deal so far. And talking > about the "5%". But there were a lot of highland tracks with great > accuracy in the OFP data. I think this is not a perfect solution but all > these roads that possibly might be taken from a map I marked with a > FIXME for resurvey. Especially roads that are just 5 points on 10km seem > quite suspicious to me. But anyways, who knows, maybe they are just very > simplified? OSM is a dynamic project that will never end or be finished, > so we always have the chance to kill hese resurvey ways. And for 95% of > the ways I added there is still the source=ourfootprints tag to find the > ways. > > And while checking the data I added and corrected some OSM data and > relations as well. > > Thank you again to Ævar for the import. > > Greetings from the far east to the south... > > Christoph > > > Am 17.01.2010 21:27, schrieb Ævar Arnfjörð Bjarmason: >> On Sun, Jan 17, 2010 at 20:29, Karl Palsson<tw...@tweak.net.au> wrote: >>> Ævar Arnfjörð Bjarmason wrote: >>>> Tom að mestu rakið vegi bæði frá egin GPS ferlum og u.þ.b. 30 annara >>>> sem hafa sent honum gögn en hann segir að um 95% af kortinu sé búið >>>> til svona, 5% er teiknað eftir lokuðum kortum (LMÍ kortum geri ég ráð >>>> fyrir), ár, vötn, jöklar og strandlengjan eru svo teiknuð eftir Google >>>> Satellite myndum. >>> >>> How did you go about excluding the 5% that were copied from other maps? >>> You do >>> know that importing this has totally corrupted any claim OSM might have on >>> being >>> completely clean data? >> >> 5% of Tom's source dataset is partially derived from Google Maps / LMÍ >> but none of what just got imported on OpenStreetMap. >> >> What the ourFootPrints import imported was a subset of the >> ourfootprints.de dataset, it was just the GPX traced ways& none of >> the Google Satellite traced waterways, POIs or various other things >> that were either unfit for importing or we'll possibly do in the >> future. >> >>> Further, what possibly motivated you to import this over reykjavik? We now >>> have >>> a great big pile of vomit, with no real known source or age, whose only real >>> qualification is that a bunch of Germans made it before OSM existed. Good >>> for >>> them. >> >> When you do any sort of data import into an area with existing data >> you're going to have conflicts. There are a few ways to deal with >> that: >> >> * Delete all the existing data& Import the new one, obviously not >> applicable here but this was done e.g. in Gaza >> * Try to automatically or manually detect conflicts before >> uploading. This was done for the Corine import& requires a lot of >> programming / debugging / testing work or manual labour by a limited >> group beforehand. >> * Import the data on top of the existing data& clean it up later >> >> I went with the last option after it became clear that most of the >> users doing regular edits (who replied on the list) were willing to >> help clean it up afterwards. The whole import is only around 8000 ways >> which isn't too big for such manual cleanup work. >> >> I'm sorry for any inconvenience this has caused and in the areas such >> as Reykjavík where this import affects most users (especially those >> using Potlatch such as yourself) I'll try to cleanup the data very >> soon. I probably would have done so already for the capital area >> except I've been fixing up the import slightly by deleting source=* >> tags on all the nodes I imported. >> >> Most of the OFP data in Reykjavík and its surroundings will be >> deleted, except some residential ways which haven't yet been covered >> by us (like some in Kópavogur) and a few tracks. >> >>> While this surely has some very positive outcomes in unmapped areas, in >>> other >>> areas we now just have a mess that someone has to go out and _resurvey_ >>> just to >>> try and work out which one is right. >> >> How is this a bad thing? Once we're done with the cleanup we'll have >> data in areas where we had none before. Sure that data may be out of >> date or incomplete but it's still *some* data instead of none at all. >> >> We may need to treat ourfootprints data as implicitly needing >> confirmation or resurveying of course, but in the meantime we at least >> have something. >> >>> Further, large chunks of the OFP import in reykjavik isn't even >>> _semantically_ >>> correct, let alone geographically correct. Roads that overlay, but don't >>> intersect, Roads drawn as straight lines, which may have been straight on >>> original city plans, but clearly are not when you stand on the street, >>> roads or >>> connections that don't exist anymore. >> >> Yes a lot of the OFP roads don't connect with each other (although >> some do). Errors like that can be easily fixed with the validator when >> the data is imported. >> >> As for the accuracy see the point above. >> >>> In some other parts of the country, this was _possibly_ a good idea, but I'm >>> sorry, I really think this was a bad idea to import the whole lot. >> >> Out in the country it was definitely a good idea. The import greatly >> expands our coverage of the national highway system& various offroad >> tracks. The quality of those ways is on par with or even surpassing >> our own existing ways. >> >> _______________________________________________ >> Talk-is mailing list >> Talk-is@openstreetmap.org >> http://lists.openstreetmap.org/listinfo/talk-is > > _______________________________________________ > Talk-is mailing list > Talk-is@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-is _______________________________________________ Talk-is mailing list Talk-is@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-is
