[Talk-GB] Adding a further 250, 000 UK roads quickly using a Bot?

Kai Krueger kakrueger at gmail.com
Fri Feb 4 01:14:20 GMT 2011

Ed Avis wrote:
> I think automatically importing the OS data for areas where OSM currently
> has
> little to no coverage - or coverage merely traced from Yahoo imagery - is
> a great idea.  Many people don't want bots to trample on their 'patch'
> which they
> have carefully surveyed, which is fine.  But there is still a huge part of
> the
> country which is very under-mapped.  Let's get it complete as soon as
> possible
> using the best data we have.  (It is my opinion that the OS data is often
> of
> higher quality than a typical OSM ground survey, although of course
> multiple
> surveys by different people combined with the use of Bing imagery will
> give the
> best possible map.)

I fully agree on this. Many of the problems people complain about with
respect to bots is that they degrade very well mapped areas, as e.g. they
added an old road back in that was deliberately left out in the ground
survey as it no longer exists. Adding new data in areas that are not mapped
at all is much less problematic, as we can be sure it hasn't made the
situation worse, just not made it immediately as good as it can be after
many iterations of improvements.

OS Streetview data is very high quality data with good positional and
labeling accuracy. Sure it has errors, is out of data and doesn't have all
POIs and other metadata. But it is better than most first pass ground
surveys done by OSM contributors.

According to Muki Haklay's Paper "How Many Volunteers Does it Take to Map an
Area Well? The Validity of Linus’ Law to Volunteered Geographic
Information", on average OSM only reaches its quality only once it reaches 5
- 15 contributors per square kilometer.

Does that mean one shouldn't go to mapping parties in different cities or
map during one off visits or holidays? After all, one might leave dead data
behind without building a community? No, OSM has always been do as good a
job as you can now and some one will improve things later.

If we can't build the right tools and services to attract people into areas
that already have a base coverage, to update and improve things, OSM will be
dead data anyway eventually. We need to face these issues sooner rather than
later and having an (imported) base map to build upon will allow us to get
to this stage quicker. It allows us to focus on these issues of how to
attract people into maintanance and gardening data that others have entered
before they joined.

Also looking at Germany, they don't actually seem to have too many problems
attracting new people into  
data dense areas. 

Ed Avis wrote:
> I also think that what is good for ITO is good for the map in general - if
> there
> is a need to have a complete map for some government contract or whatever,
> then
> that is a need that exists in the real world.
Also, we are getting some of the most comprehensive on the ground
verification and improvement reports from applications like the Sat-Nav
Skobbler bugs with MapDust. The more complete the map is, the more people
will use things like that and give feedback on errors. Currently you still
get a lot of errors like "Leeds is missing 4000 roads" or "No roads in the
next 10 km" which are useless, but in areas that were traced from OS data,
one can get some good on the ground feedback about missspellings, wrong
connectivity, missing POIs and other valuable data. Way more than we have
gotten in those areas in the last 6 years.

(Yes, MapDust still has a big problem with Signal-to-Noise ratio, but thanks
to the high volume there is actually quite a lot of good and helpful data,
unless the town just simply doesn't exist.)

View this message in context: http://gis.638310.n2.nabble.com/Adding-a-further-250-000-UK-roads-quickly-using-a-Bot-tp5986539p5991203.html
Sent from the Great Britain mailing list archive at Nabble.com.

More information about the Talk-GB mailing list