[OSM-talk] Adding Wikidata tags to 70k items automatically

Simon Poole simon at poole.ch
Sat Aug 30 11:23:09 UTC 2014


There are three aspects to your question.

1) wikidata is licensed on CC0 terms, essentially that boils down to no
restrictions on use at all. Looking at it from the WMFs position we can
link to wikidata data as much as we want, however on the other hand the
WMF does not guarantee or warrant in any form that the contents of
wikidata are free of any rights of third parties.

2) My understanding of the WMFs position is that they state that there
is no copyright on facts and that is it. This leaves the onus of
determining if a certain item in wikidata is problematic (for example
for use in OSM) clearly in the lap of the user. There was, when I was
talking to the wikidata people in May, some discussion of clearly
identifying the source of an wikidata statement for example in the case
of OSM derived data, but I don't know if that has actually happened.

3) To get back to your example, assume that the WMF got hold of a
geo-referenced list of UK post codes per building and imported it in to
wikidata and we, either linked in lieu of the post code to wikidata
(addr:postcode:wikidata=Q....) or extracted them and applied them
directly to OSM objects. I suspect (this can naturally only be
speculation) that in both cases we would be in big trouble and there is
legally no argument to prefer one over the other.

Simon

Am 30.08.2014 12:02, schrieb Robert Whittaker (OSM lists):
> On 27 August 2014 22:06, Andy Mabbett <andy at pigsonthewing.org.uk> wrote:
>> There is no license issue. No copyright-protected data is being added
>> to OSM. No cordinates from Wikidata are being added to OSM. No text
>> from Wikidata is being added to OSM.
> 
> While there may not be a problem in this case, and I'm actually quite
> supportive of the proposed addition of the wikidata tags, I'm not
> convinced that this argument is completely correct.
> 
> Take UK postcodes for example. Suppose that postcodes themselves are
> non-copyrightable identifiers, but their locations are subject to
> database rights. Suppose we have a copyrighted database of each
> address with it's location and postcode. For each address we could
> search OSM for objects that may represent houses/businesses in the
> vicinity of the location in the database. In cases where there is
> exactly one match within say 10m of the location, we could safely
> assume that the postcode belongs on that object.
> 
> Suppose we automatically add the postcodes to these objects. We
> arguably haven't added any copyrighted data to OSM, and certainly
> haven't added any coordinates from the copyrighted database. But what
> we'd end up with in OSM is a list of postcoded objects with OSM
> coordinates that are pretty much the same (within 10m anyway) as those
> in the original database. It seems to me that the owner of the
> postcode database would have cause to complain that the OSM data is a
> derived work of their database.
> 
> The use case for wikidata is slightly different, since (as I
> understand it) the location data is being used more as a filter than
> as a primary matching tool. But I think we may need to be a bit more
> careful. Perhaps LWG should be asked for their opinion...
> 
> Robert.
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20140830/60f1f8f1/attachment.sig>


More information about the talk mailing list