[ontoiop-forum] "Dataset description" paper about the OntoIOp registry submitted: questions about license and API4KB

[Christoph LANGE]

Hi all,

[@Tara plus maybe @Elisa: something specifically for you about API4KB 
below]

at 
http://www.semantic-web-journal.net/content/ontoiop-registry-–-dataset-supporting-ontology-model-and-specification-integration-and 
(interesting, a URL with a Unicode "em dash") there is a paper 
submission by me, Till and Oliver, describing the OntoIOp Registry from 
a linked dataset perspective.

It was a last-minute effort before an individually extended deadline, so 
we had no chance to have a deeper interaction in this community _before_ 
submission.

However, I believe that two issues are of more general interest, so why 
not discuss them now.  Hoping that the paper will not be rejected (I am 
reasonably confident about it, was we have "ticked" many of the "boxes" 
by which a linked dataset is measured), we will then still have the 
chance to "fix" them in the next round of the submission process.

1. License of the dataset: A linked open dataset should have a license, 
which you usually choose from a small set of well-known licenses 
depending on your requirements.  After some initial research I found the 
following ones appropriate.  BTW whatever the license, no one will be 
able to _mess_ with _the_ _real_ registry at its official URI=URL 
anyway, as we own the http://purl.net/dol redirect.  I think the "reuse" 
scenario of incorporating variants of data from the Registry, or the 
republication of dumps of the registry (possible in alternative formats) 
is the more likely scenario.  I believe that a share-alike license is 
too restrictive, so two reasonable ones remain:

http://opendatacommons.org/licenses/by/ (summary at 
http://opendatacommons.org/licenses/by/summary/) – permits pretty much 
anything but requires attribution of the original source (I guess 
something like "derived from the OntoIOp Registry, (c) the OntoIOp WG").

http://creativecommons.org/publicdomain/zero/1.0/ – public domain = 
anything allowed.  Just that there is a more water-tight legal 
framework, i.e. the CC0 license is safer than just attaching a string 
"public domain".

Do you have any opinions?  I hope this won't be too hard a decision, but 
should be taken by consensus.

2. API4KB's OntoIOp Terminology ontology 
(https://github.com/API4KBs/api4kbs/blob/master/ontologies/OntoIOpTerminology.rdf) 
and its relation to LoLa: LoLa plays a role in the paper in that it is 
the vocabulary of the "OntoIOp Registry" dataset in its current state. 
The "OntoIOp Terminology" ontology (@Tara: Till made me aware of it, and 
I've had a first look) appears to be very similar – but may have been 
designed to be different in order to support different use cases (of 
which I am not fully aware, but happy to learn).  In the paper I 
promised (and thus have to do it within a few days, before the reviewers 
are assigned) that I would, because of this similarity, align LoLa with 
the OntoIOp Terminology.  I'm planning to drop several "LoLaConcept 
EquivalentClass OntoIOpTerminologyConcept" links into LoLa and then 
point you to them as a basis for a further alignment discussion.  Does 
this make sense?

Cheers,

Christoph

-- 
Christoph Lange, Enterprise Information Systems Department
Applied Computer Science @ University of Bonn; Fraunhofer IAIS
http://langec.wordpress.com/about, Skype duke4701

→ SEMANTiCS conference: Transfer–Engineering–Community.
   Leipzig, Germany, 4–5 September (workshops 1–3 September).
   Including Vocabulary Carnival, LOD for SMEs, Linked Data Quality.