[open-bibliography] Having a license (and attribution) travel with data (was: Re: Announce: Open Bibliography (JISCOBIB))

Mon Jun 28 15:35:40 UTC 2010

On 25 June 2010 20:18, Karen Coyle <kcoyle at kcoyle.net> wrote:
> Quoting Jonathan Gray <jonathan.gray at okfn.org>:
>
>>
>> Just to add: I think it would be *fantastic* if LT released data under
>> an attribution license! ;-)
>
>
> What would the logical mechanism be for having the license travel with the
> data? For example, OCLC proposed a MARC field for license information, but
> such a field is easily removed, on the one hand, and is also necessarily
> lost when one transfers data to any format that does not have an analogous
> license field. And that only works for records... if you begin to develop a
> linked data model, you would need a license on each statement.

You raise an interesting question here Karen -- and one that is not
specific to bibliographic data of course. I think there are various
possible answers:

a) You could try to associate license (or attribution) information at
some fairly low level (just as you would do for general provenance
info -- i.e. where this came from, who's edited it etc). I
specifically say low-level because they could be some flexibility
here. For example, it could be at the record level or named graph
level rather than the level of a field or individual triple.

b) You could (assuming a set of compatible licenses) simply provide an
overall citation/license for a dataset with that citation/license
encompassing the datasets included therein. So, for example, if
dataset A were used in the construction of dataset B you wouldn't cite
the individual field/records/statement from B that came or derived
from A but would simply say: "Dataset B uses material from dataset A"
and then include an appropriate citation.

There are also half-way houses between the two approaches: for example
you indicate that a particular segment of your data came from a given
source.

Regards,

Rufus