[OSM-talk] Tag for true OSM data?

[Martijn van Exel]

On Thu, Mar 31, 2011 at 1:22 PM, Jukka Rahkonen <
jukka.rahkonen at latuviitta.fi> wrote:

> M∡rtin Koppenhoefer <dieterdreist <at> gmail.com> writes:
>
> > 2011/3/31 Martijn van Exel <m at rtijn.org>:
> >> Having said that, there's the important issue of human contributors
> >> adding value to the imports: adding metadata, removing stale data,
> >> improving geometry. Those are real human contributions that would
> >> not have occurred if it weren't for the import happening first.
>
> > +1
> > And with the time, the original imports will more and more (given an
> > active community) fade towards cloudbased human mapping.
>
> > I also don't share the fear that there will on the long run a growing
> > amount of imports but rather guess that imports will become less,
> > because it becomes always more complicated to import stuff, when half
> > of it is already present in OSM, while it was easy to import into an
> > empty map.
>
> Data which are significantly enhanced by the users (I know, it
> is hard to define what is significant) would be valuable for me
> as well. Basically I do not care about rendering but I have an
> interest in plain data. I am after interoperability without a
> need to import everything into OSM first. I would like to do
> spatial queries from the database and through web services and
> see what features and what tags OSM users have recorded around
> a given place and do comparisons with data from other data sources.
> But I fear no single tag nor a simple tag combination can ever be
> used for finding the real user contributed data and I will need
> to preprocess OSM data somehow.
>
> -Jukka Rahkonen-
>

One thing that will  be helpful is the full history dumps that are made
available on an irregular basis. From that data, if you find a way to import
that full history back into a database,  you could derive how much of a
feature is made up of human contributions and how much is import, in terms
of metadata (tags), geometry, number of human contributors to the feature.
It is not straightforward but playing around with the full history may give
you a feeling for how to tweak those parameters to get the right filter.

This is something I am thinking about a lot too, but currently I am still
finding out the best way to deal with the huge chunk of data that is the
full history of OSM...

Best,
-- 
Martijn van Exel
http://about.me/mvexel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20110331/a9ab7407/attachment.html>