Have you looked at this? They seem to have an overall plan and some software for aiding conflating points with buildings and POI's
https://wiki.openstreetmap.org/wiki/Seattle_Import On Tue, 2018-07-10 at 21:22 -0400, Leif Rasmussen wrote: > Good ideas! > In my previous address import, the Orange County NC address import, I > did not initially merge the addresses with POIs and buildings. > Instead, I am slowly merging the addresses with other features > manually in small changesets, mostly via ID. I have already merged > hundreds of addresses. I could potentially merge the data > differently in this next import if that is easier. I don't know a > good way to do that without over complicating the import though. > Whatever ensures that the data in OSM is good in the end works for > me! > Thanks, > Leif Rasmussen > > On Tue, Jul 10, 2018, 7:56 PM James Umbanhowar <[email protected]> > wrote: > > Hi Leif (replying on list so we don't get duplicates.) > > > > Ok, it's good that we won't get duplicate addresses. > > > > One thing that many address imports have done is to merge addresses > > onto preexisting buildings. Have you considered this? I'm > > guessing > > that right now the points are at the centroid for the property and > > in > > many cases they are pretty far from existing drawn buildings. > > > > With that said, I'm pretty sure that Durham has building data that > > it > > might be worthwhile to include in this import. > > > > Similarly, if there is an existing POI (say a restaurant) that > > doesn't > > have address tags, it would be very good to merge the imported > > address > > with the POI, either manually or some other way. It would be a > > pity to > > have an address point and a POI and not have them together if they > > are > > equivalent. > > > > You may want to ask some other groups who have done similar imports > > how > > they managed all these tasks. I think the Seattle building import > > had > > great experience and several other groups have used their mode. You > > could probably get a few volunteers (I'd be one) to do some of the > > manual work if that is necessary. > > > > James > > > > On Tue, 2018-07-10 at 15:57 -0400, Leif Rasmussen wrote: > > > Thanks for the email! > > > I have already removed duplicates of existing addresses from the > > > dataset so that only missing addresses will be added. I used a > > > custom made Java program that finds the duplicates to do it. It > > is > > > available on Google drive. The link for the program is included > > in > > > the wiki page. I will not edit any existing addresses. As for > > the > > > apartment complexes, good point! I will fix those by merging the > > > addresses into one and removing addr:unit. If you see anything > > else, > > > please let me know! > > > Thank you, > > > Leif Rasmussen > > > > > > On Tue, Jul 10, 2018, 3:16 PM James Umbanhowar <[email protected] > > m> > > > wrote: > > > > Importing these addresses may be useful, but the current > > proposal > > > > does > > > > not look ready to go, yet. > > > > > > > > There doesn't appear to be any plan to conflate the data with > > > > existing > > > > osm objects. There are many existing buildings and POIs in the > > > > data > > > > base and you will need to find some way to add addresses to > > > > existing > > > > buildings, merge addresses with existing POI's, and remove > > > > duplicate > > > > addresses. > > > > > > > > In addition, there are some spatial issues with addresses, > > > > especially > > > > multiple addresses in an apartment complex where they are laid > > out > > > > in > > > > grid, rather than at a specific location. > > > > > > > > I haven't done a complete examination of the data/plan, but > > these > > > > two > > > > major issues stick out already. > > > > > > > > James > > > > > > > > On Tue, 2018-07-10 at 13:58 -0400, Leif Rasmussen wrote: > > > > > Hi everyone, > > > > > I am now planning a third address import in North Carolina. > > The > > > > > import will include 140,000 addresses in Durham County, North > > > > > Carolina USA. I will follow the same procedure as the Orange > > > > County > > > > > Address Import to avoid import disasters. The import will > > take > > > > place > > > > > next weekend. If anyone has any concerns, please let me > > know! > > > > As > > > > > always, I am open to discussing anything at all! > > > > > > > > > > Wiki Page: https://wiki.openstreetmap.org/wiki/Durham_County_ > > Nort > > > > h_Ca > > > > > rolina_Address_Import > > > > > Transformed Data: https://drive.google.com/drive/folders/1Wbj > > wEvv > > > > IcqA > > > > > G_K1O2q0g1jansczWFlBi > > > > > > > > > > Also, I am currently waiting for a reply from the Chatham > > County > > > > GIS > > > > > department. I could not find the license for the addresses > > on > > > > their > > > > > website, so I sent them an email asking about it. As soon as > > I > > > > am > > > > > 100% sure that I can legally import the Chatham County > > addresses, > > > > I > > > > > will add them. I am pretty sure that the addresses are > > Public > > > > > Domain, but if anyone can find the license on the website > > (https: > > > > //gi > > > > > sopendata- > > > > > > > > > > > chathamncgis.opendata.arcgis.com/datasets/7f7881ca05a94fa1ac6fe618e > > > > 84 > > > > > fb725_0), please let me know! The license is all I need now > > to > > > > > import the 40,000 addresses. This is a separate import from > > the > > > > > Durham County Address Import, which uses the Open Database > > > > License > > > > > 1.0. > > > > > > > > > > Thank you, > > > > > Leif Rasmussen > > > > > _______________________________________________ > > > > > Imports-us mailing list > > > > > [email protected] > > > > > https://lists.openstreetmap.org/listinfo/imports-us _______________________________________________ Imports-us mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports-us
