[NLP2RDF] NIF and provenance

[J. Fernando Sánchez]

Hello,

This topic is really interesting. There are many ways to use PROV and NIF
together, so the challenge is defining the best practices. For instance,
the aggregation of annotations from different tools in NIF falls under the
use cases of PROV, but its approach is completely different.

I can tell you about our experience in a European project using NIF with
PROV-O for a specific scenario: sentiment analysis services. We focused on
ease of use for service developers, so we used json-ld output and one of
the priorities was making this json-ld as readable and usable as possible.

Simply put, each call to the NLP service generates and returns a
prov:Activity. That call generates the necessary nif:Context (also a
prov:Entity), and adds sentiment information too. The Sentiment Analysis is
a prov:Activity itself.

You can read about it here:
http://eurosentiment.readthedocs.org/en/master/format/servicesformat.html
And a working example: http://demos.gsi.dit.upm.es/onyxemote/

Then again, our use case is special because contexts are tied the service
call, we don't have to deal with versioning and aggregation.

Regards.

On Thu, Dec 18, 2014 at 3:12 PM, Rob H Warren <warren at muninn-project.org>
wrote:
>
> Philip,
>
> What would be the advantage of adding to NIF versus reusing the PROV
> ontology for this purpose? I have used this for some projects and have not
> seen a corner case yet.
>
> -rhw
>
> On Dec 17, 2014, at 5:02 PM,
> nlp2rdf-request at lists.informatik.uni-leipzig.de wrote:
> >
> > Dear all,
> >
> >     as I mentioned to Sebastian briefly, we are working on adding a
> > provenance layer to NIF. And we would like to talk to you.
> >
> > I sort of agreed with Sebastian that we could meet in Leipzig to discuss
> > this.
> >
> > However, I think it is more efficient if we exchange material beforehand
> > and have a telco early January to discuss the material.
> >
> > Sebastian: you mentioned that you have some proposals on how to
> > represent provenance of NIF annotations. Can you share that proposal
> > with us?
> >
> > Peter: can you circulate our draft to the people in Leipzig (i.e.
> > Sebastian and Martin).
> >
> > We can then start from there and organize a telco in January.
> >
> > Best regards,
> >
> > Philipp.
> >
> > --
> > --
> > Prof. Dr. Philipp Cimiano
> > AG Semantic Computing
> > Exzellenzcluster f?r Cognitive Interaction Technology (CITEC)
> > Universit?t Bielefeld
> >
> > Tel: +49 521 106 12249
> > Fax: +49 521 106 6560
> > Mail: cimiano at cit-ec.uni-bielefeld.de
> >
> > Office CITEC-2.307
> > Universit?tsstr. 21-25
> > 33615 Bielefeld, NRW
> > Germany
> >
> >
> >
> > ------------------------------
>
> _______________________________________________
> NLP2RDF mailing list
> NLP2RDF at lists.informatik.uni-leipzig.de
> http://lists.informatik.uni-leipzig.de/mailman/listinfo/nlp2rdf
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.informatik.uni-leipzig.de/pipermail/nlp2rdf/attachments/20141218/3d5e7067/attachment.html>