Hi, we recently got the ok from a local authority to add street names to an area. I am aware that you can go a change manually via the website but that’s arduous and will take much longer.
I am aware there is an api which brings back the nodes, ways etc. I am not familiar at all with the OSM terninolgy. Just what i read about it the last few days. I come from a database background, so SQL expert but i also understand javascript, C#, xml, programming in general. Ideally i would like to download the xml, load it into a DB, say postgres, SQL, amend the street names on the DB and then upload it into open street naps, I would also like to test it first, of course! i understand there is a copy of OSM maps as another api call? I read that you support postgres and mysql. Can someone please explain how this might be possible and can i use a DB to save the xml data there, test then load. Or will that overcomplicate the process? What do the experts here recommend. Thanks very much.
Before we even get onto anything remotely technical, please follow this guide.
@SomeoneElse Yeah, technically, but honestly I would say this before proceeding:
Show the data. All of it.
People will find any issues before you’re even halfway into writing an import proposal.
This. We had an issue where someone imported data before properly discussing about it and now it’s a continuous operation of cleaning up (duplicate data or improper info).
@Panos_Stavroulis As it was said, as you gonna do such import, better follow the guidelines, just to avoid any future issue with those data.
In other words, as I assume from your edits so-far it will be about street names in Kilkis area, better post the proposed import to the Greek community so to better check if everything is ok.
Importing street names into OSM from an external source is hard. So hard that in Serbian community we decided to leave the process (semi-)manual, i.e. left to individual mappers but assisted by tools. Here’s the list of roadblocks we encountered:
Data issues
- Street names in the official dataset were spelled UPPERCASE. Easy for them, but unsuitable for OSM. We auto-converted them to Title Case, but that still does not conform with orthographic rules, so we checked and edited the list manually.
- We decided to drop the default word “ulica” (‘street’) from names, but it’s used somewhat haphazardly in the dataset. But then, we want to keep it for “17. ulica”. So again, manual checking.
- Numbered streets or those named after dates are a pain, since some are officially spelled with a numeral (“11. oktobar”), some with a word (“Jedanaesti oktobar”), sometimes in nominative case, and sometimes in genitive case (“Jedanaestog oktobra”).
- There were a few obvious misspellings in the official dataset.
Conflation issues
Even when you overcome the data issues somehow, the hardest part is conflation of official data with the existing OSM objects. The issues include:
- One OSM highway is divided into several segments, all sharing the name and official ID. You have to assign the name to all of them.
- One OSM highway is actually drawn over two or more differently named streets. You have to split it before assigning names.
- There are missing OSM highways that should be added with help of official geometry.
- There are “phantom” government streets (and addresses assigned to them) which are drawn between some houses, but there really is not a public highway. So we want to keep those out.
- En route, we can use the opportunity to review and retag many OSM highways (between
unclassified
,service
,residential
andtrack
), split some, merge some…