comments on DUV and some proposals

Dear All,

When analyzing the current version of the Data Usage Vocabulary, I found the
need to propose several changes, so I attach here a diagram with my proposal
for DUV.

I hope that you will agree with me that, while we have a good start, we are
not ready to go FPWD with DUV.

One general remark is that the current version relies on a number of
vocabularies (some of which are not stable like bibo, cito). I think we
should minimize those dependencies as much as possible, and replace them
with more stable and well-known vocabularies, more specifically I mean DCMI
Metadata Terms (dct [1], which is already used in DCAT), and the W3C PROV
Ontology (prov [2]). I am not sure about the status of  Open Annotation and
how we should we use (details below). Further, there are some fragments
which do not need to be ³reinvented", so we can just reuse. This is to me
the case with Application.

I see three clear modules for the DUV effort:
* A) Data usage representation (in the sense that a software agent is
dependent on the dataset)
* B) A particular kind of usage which is citation
* C) User feedback/rating representation
Detailed comments below:

With respect to A, I propose to use prov:SoftwareAgent instead of
Experience/Application/WebOfThing. I propose to change duv:consumes to
duv:consumed. Further, I think that we need to be able to reify this
Consumption (or Usage if we prefer that other term). This would be a
solution similar to what PROV does with respect to attribution. It includes
a simples wasAttributedTo, but this can be qualified with an Attribution
resource, which reifies the attribution and allows for more information
about the attribution to be captured. I think we should do this here because
A is actually the core of DUV, and is currently only one property in the
whole thing.

With respect to B, I don¹t think we need to enter the business of specifying
the range of cito:hasCitingEntity, specially not with bibo. So, in my
opinion, we should say nothing about this, and therefore introduce no
dependence to bibo. With respect to cito, the only thing we would be doing
is subClass cito:CitationAct into duv:DataCitationAct (I propose to rename
duv:Citation, because since we are using cito, we should use their
convention in the nominalization which is better and clarifies that Citation
here is an Act.) We should examine this solution altogether, because citing
data is already possible without us doing any specialization of the
vocabulary. So, B, could be altogether left out from DUV and we could just
give examples of how to cite data using cito (and even bibo). If we decide
to include duv:DataCitationAct then there should be some reason further than
constraining cito:hasCitedEntity to have a range of dcat:Dataset.

With respect to C, if we go with Open Annotation, then we could call what is
currently called duv:Feedback as duv:DataRatingAnnotation. However, note
that Open Annotation suggests that we do not subclass oa:Annotation because
of particular motivations for annotation but instead use SKOS and create
instances of oa:Motivation. In this case, we should eliminate duv:Feedback
altogether, and just understand User feedback/rating as a new instance of
oa:Motivation (e.g., oa:rating). (see current list at [3], which does not
include in my opinion something like oa:rating). What currently is
duv:has_rating would be some subclass of oa:hasBody (or it would just be
oa:hasBody). This requires more discussion as I found the current examples
unclear, which bodies of the annotations that are just text.

So, perhaps, all our effort with respect to ³C², would just be in creating
conventions to use Open Annotation and not a specialization thereof.

I did not understand duv:retains.

I am sorry I have to send regrets for the meeting today.

Best regards,
João Paulo

[1] http://dublincore.org/documents/2012/06/14/dcmi-terms/
[2] http://www.w3.org/TR/prov-o/
[3] http://www.openannotation.org/spec/core/20130208/core.html