OpenStreetMap
The Belgian community is currently looking at some tools to import house numbers from the AGIV CRAB database. We are in an experimental phase, there is no formal go for the import yet.
Using the tool I saw a street in which I didn’t collect house numbers so far. It was tempting to just copy the numbers. But since I needed to walk the dogs, I decided to pass through that street. So what did I discover during this short survey ? A zone 30, a memorial for Frans Abels (a composer 1899-1962), a missing path and a waste bin. It was just 10 minutes extra compared to our normal walk.
Conclusion ? For me it is not sufficient to just copy numbers from a database. It’s better to go out for a walk with the dogs. Using this method I collect more diverse data en I learn something along the way.
Discussion
Comment from Omnific on 23 October 2014 at 19:47
The local, surveyed input is the best, definitely. But there needs to be a middle ground between no addresses and locally surveyed addresses. Imports make sense to get some data in place, and the error corrections to that data should be left up to the local surveyors.
This highlights one of the issues I have and Steve Coast seems to have recently acknowledged. OSM is a fantastic map in many places visually, but the lack of addresses (there are something like 49 million addresses total worldwide, a tiny fraction of the real number, probably upwards of 2 billion or more) make it a poor substitute for any commercial map. This is especially of note in the US, where, as opposed to Western Europe, there are very few addresses outside of a few major imports (Washington, DC, Seattle, Chicago, NYC, San Francisco). Our contributer base here is also much lower per square mile than in Europe, making surveying hundreds of millions of address by hand unfeasible.
There needs to be more focus on getting accurate and open address data from governments and getting it imported in a timely manner. However, this imported data needs to be tagged in a way to indicate that it is imported and may need a local survey. This tag can be used to make QA tools (i.e. the addresses that have not been reviewed are highlighted by the program). That way, hopefully OSM can have addressing that will be on-par with commercial maps in many places due to imports, but will be superior in that many of the addresses will also be reviewed. Maybe that can strike a balance between the two sides of the addressing argument, importing and surveying.
Comment from Jiří Komárek on 23 October 2014 at 21:48
Not a long time ago the import of house numbers was finished in the Czech republic - http://wiki.openstreetmap.org/wiki/Import_adres_z_RUIAN
Comment from escada on 24 October 2014 at 03:21
I’ll understand that for remote areas where you have to drive for kilometers (or miles :-) ) before finding another house and house number, imports are a nice solution.
But for those who did an import: did you ever try to survey to check the correctness of, lets say, 500 house numbers in an area that was imported ? Was it fun, easy, rewarding (to find 1 or 2 mistakes) ? Did you ever try to survey an area without house numbers ? Was it fun, easy, rewarding to add 500 new numbers ?
Please share your thoughts.
Comment from okilimu on 24 October 2014 at 08:53
In Augsburg, south of Germany, we moved around (approx 5-7) mappers for the last 3 years, started at 5% and we are now at 99%, added more than 40.000 housenumbers (see [1]).
We asked our city government for a list of all housenumbers in Augsburg (about 41.000) and i coded the housenumber evaluation program, referenced in [1]. It’s a good motivation to have a goal in view, even it’s hard at the beginning (and now more at the end at 99%).
In munich, not far away and with 1.2 mio people, i got also the housenumber list and very few mappers mapped nearly 90.000 housenumbers in 15 month!
When i collected housenumbers in streets, where no housenumbers were mapped before, i collected about 60 up to 100 in an hour. Later at home, i needed 2 or 3 times of it to import the data, including some POIs.
Greetings,
Dietmar aka okilimu on OSM
[1] http://regio-osm.de/hausnummerauswertung/auswertung_auswahldetails?land=Bundesrepublik%20Deutschland&stadt=Augsburg [2] regio-osm.de/hausnummerauswertung/historischeentwicklung/grafik?job_id=775
Comment from escada on 24 October 2014 at 12:22
Dietmar, thanks for your reaction
According to Pascal Neis’ tools, I also mapped 35.000 addresses in the past 3.5 years. Your numbers are along the same magnitude. I love collecting new numbers, I don’t see myself verifying the house numbers of an import.
Comment from Sanderd17 on 24 October 2014 at 15:33
I mapped about all housenumbers as far as my dog could walk.
Though, since I live in a rather rural place, that ain’t so much.
It becomes harder when you have to walk (or ride) a longer way before you can start to map.
That’s why I think an import is good. But people should restrict themselves to places where they’ve been before. So they don’t import total rubbish.
And we should certainly have tools ready to find differences between OSM and CRAB. So we can find issues in both databases.
Comment from escada on 24 October 2014 at 15:40
I really hope people would take their time to import the data, really look at strange situations and not just hurry to get all data in.
On weekdays my survey area is rather limited, during the weekends, I don’t have problems to drive an hour or so to make a walk. Of course a walk is not a survey, the prime goal is then different, but I still have collected a reasonable amount of house numbers in (for Belgium) remote areas.
Comment from Warin61 on 25 December 2014 at 00:47
Here in Australia .. there are missing road names.. so numbers are of less concern. Old maps had no house numbers… in some remote parts of Australia numbers are not used .. they use the name of the house/property/station/homestead. Considering that the next place may be some 100s of kms away a physical survey would take some time.
A local physical survey is much better at getting stuff that would be difficult to import. I use the smart phone camera and voice recorder to remember stuff for later entry into OSM. Data imports certainly fill the map quickly and can be mostly accurate. Don’t know the accuracy rate of the data import vs the manual entry from survey, suspect it is similar. But the survey will get a variety of stuff that would need many different data imports .. if they are available.
Comment from escada on 25 December 2014 at 07:17
Warin61, thanks for sharing some problems of the Australian community. Of course the size of your country causes different problems compared to small Belgium.
Probably the quality of a good import is the same as those of surveys. My main critique regarding imports is that sometimes obvious wrong data is imported. Some of those imports have a manual verification step, but that is sometimes rushed, so non-existing buildings are imported or house numbers are placed in the dunes. That something you won’t see with surveys.