search

theodi / schemas

2 stars 0 forks source link

semantics of data/contentLicense #1

Open aisaac opened 9 years ago

aisaac commented 9 years ago

Hi,

I am currently writing some material for the new version of the DPLA/Europeana document on rights that you commented on (thanks again) [1]

There are several reasons for which we will very probably not pick ODRS' contentLicense/dataLicense properties to indicate a link between a License and a RightsStatement derived from it. In the process, a question emerged, about when the ODRS ontology says that these two properties are sub-properties of dct:license [2]. Can one really say that the License is the dct:license of the more specific odrs:RightsStatement, the same way as this License could be the dct:license of an object (or dataset, in the ODRS pattern)? To me it looks like these sub-property axioms amount to a quite different interpretation of dct:license.

[1] https://docs.google.com/document/d/1H6TWxGARqUMxJrc2sXjaBlOsg7UkUTb27rvtS8aC5y4/ [2] https://github.com/theodi/open-data-licensing/blob/master/schema/schema.ttl

ldodds commented 9 years ago

The definition of dct:license merely says that it should refer to a dct:LicenseDocument, it says nothing about the objects to which it is applied.

I've simply defined some extensions to that property that defines the types associated with its range and domain, specifically that they are used in the context of a RightsStatement and an odrs:License.

I'm not reinterpreting the property at all.

Can you elaborate why you think this is incorrect? I don't see any incorrect inferences that could be drawn here.

aisaac commented 9 years ago

Hi Leigh, I don't see any incorrect formal inference. It's all in the human-readable semantics: http://dublincore.org/documents/dcmi-terms/#terms-license says: "A legal document giving official permission to do something with the resource." In this definition, the "resource" will be the subject of the dct:license statement. In the case of odrs:dataLicense in the schema diagram at http://schema.theodi.org/odrs/, we have a Rights Statement as a subject and a License, say, OGL, as object. So reading the DC spec, it seems that you want to license the statement itself (I mean, after infering the dct:license from odrs:dataLicense). One could say that there is a layer of indirection, as sometimes happens in DC records. But that's always cumbersome to defend in a Linked Data context. Also, it's actually quite a different stance from the one that governs the ODRS use of dct:license between the Dataset and the License in the same schema diagram. The latter one is much more "direct" and fit to the definition in the DC spec!