My comments on DUV

Hello all,

Sorry for not being able to join today.

Here is my review for the DUV. Given all the reasons below I recommend not
to publish it yet, even if just for early feedback (at least the different
inconsistencies in the document should be solved first). I didn't have the
time to go through the other documents yet, so for those I just abstain.

- General

A general problem I see is that typical use case is for Datasets to be able
to refer to use examples when we are focusing here on the other way around.
I know reverse queries are still possible but likely far more complicated.
To improve on that side I suggest to

(1) add a note on how to make a reference from datasets to use examples as
well (e.g. using cito:isCitedAsDataSourceBy; dct:bibliographicCitation,
etc.)

(2) include a single "ReuseExample" superclass (with a better name) so one
could easily extend dcat with such reverse property.

Also, definitions reused from other vocabularies are sometimes quoted and
others not. Several links for URIs for ranges, subclasses, etc are missing.

- dcat:Dataset

Should usage be specified at the Dataset or Distribution level? I'm more
for the later.

- duv:Application

Having a look at the description I think that a more general label may be
desirable e.g. Software
Moreover, we may want also to reuse earl:Software instead

- duv:developedBy

I think the purpose here is not to develop a Soft vocabulary and thus
shouldn't go too much into that direction, so I propose to adopt
dct:creator or doap:developer here instead.

- duv:generates

Considering the generation of a dataset a kind of usage sounds quite
strange for me. I think it may be removed.

- duv:Feedback

Having a look at the diagram it is an ao:Annotation subclass, but that was
not included in the description where it is a subclass of rev:Feedback?
Also I think the current definition - which is apparently the rev:Feedback
one, is not suitable for our use case (we don't want to give feedback on
any Review but on Datasets)

Furthermore, I don't understand why to subclass ao:Annotation and then not
to reuse directly oa:hasBody and oa:hasTarget instead duv:hasRating and
duv:endorses respectively (or at least subclass both of them).

- bibo:Document

Current purl namespace is apparently broken.
Why not to use foaf:Document which is equivalent?

- duv:cites

Looks like on one hand we are trying to replicate the cito:CitationAct
model (with cited and citing entities) and on the other we are trying to
bypass it using a direct duv:cites property. I would choose between one or
the other to avoid inconsistent usage.

- duv:hasCitingEntity

Why to subclass this here? also at the diagram we are using
cito:hasCitingEntity directly

- duv:retains

I don't have any idea on what's the purpose of this property.
Feedback has itself as citing entity and at the same time is a set of
provenance descriptions?
No clue, seriously. Maybe just me.

- ao:annotatedBy

Range should be foaf:Agent

- duv:has_datasetCorrection

Not in the diagram.
Inconsistent label in the definition (has_dataCorrection)

- duv:DatasetCorrection

Shouldn't this be a dcat:Dataset as well?

- duv:WebThing

Subclass of duv:Experience that does not exists

- Example 1

I find the use of .configfile as an example of dataset consumption a little
bit confusing (although strictly speaking probably valid).

- Example 2

oa:hasRating does not exists it is duv:
using oa:hasBody and oa:hasTarget here while defining new subproperties for
them as well is quite confusing (although valid again)
See also my previous comment on removing those.

- Example 3

I think that following current model cito:hasCitedEntity should also be
included at :paperA


Best,
 CI.

---

Carlos Iglesias.
Open Data Consultant.
+34 687 917 759
contact@carlosiglesias.es
@carlosiglesias
http://es.linkedin.com/in/carlosiglesiasmoro/en