search

ld4lt / linguistic-annotation

Towards a consolidated LOD vocabulary for linguistic annotations
Other
15 stars 3 forks source link

Ability to express claimant or responsible person/organization #3

Open Arithmeticus opened 4 years ago

Arithmeticus commented 4 years ago

I'll follow the lead of @balmas and suggest it important to be able to express what person(s) or organization(s) is (are) responsible asserting the claim behind a given linguistic annotation.

chiarcos commented 4 years ago

Absolutely. I extended https://github.com/ld4lt/linguistic-annotation/blob/master/survey/required-features.md#b16-provenance-and-confidence accordingly. As this requirement is documented now, shall we close the issue or leave it open for further discussions?

balmas commented 4 years ago

How deep the requirements are meant to go at this stage and to what extent will other ontologies be used? Provenance can quickly become an immense topic. For example, "tools involved" could be expanded to include algorithms used for tokenization, etc. But this is probably a place where it makes sense to support ontologies like Prov.

(in case it's helpful as a reference, this poster gives some examples of considerations for using Prov for linguistic annotations https://www.rd-alliance.org/sites/default/files/Krohnposter.pdf , linked from https://www.rd-alliance.org/groups/research-data-provenance/wiki/rdaus-scholars-program-internship-abstract-and-poster.html/ )

Arithmeticus commented 4 years ago

Ah, I misunderstood the meaning of "provenance." Is this nestable? X claims that Y claims that Z claims that...[linguistic annotation]

On Wed, Nov 25, 2020 at 12:31 PM Bridget Almas notifications@github.com wrote:

How deep the requirements are meant to go at this stage and to what extent will other ontologies be used? Provenance can quickly become an immense topic. For example, "tools involved" could be expanded to include algorithms used for tokenization, etc. But this is probably a place where it makes sense to support ontologies like Prov.

(in case it's helpful as a reference, this poster gives some examples of considerations for using Prov for linguistic annotations https://www.rd-alliance.org/sites/default/files/Krohnposter.pdf )

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/ld4lt/linguistic-annotation/issues/3#issuecomment-733847729, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAQD53SYZ2YTZVDV3KBR5OLSRU5OXANCNFSM4UCUDLLA .

-- Joel Kalvesmaki kalvesmaki.com

chiarcos commented 4 years ago

@balmas : Goal of requirement collection at the moment is to decide whether to start working with NIF or Web Annotation as a basis (starting spmething from scratch is probably not an option), and whether any of these need resivion or extension in that case and if so, how to organize the discussion of required features for developing a vocabulary for linguistic annotaitons. Not too deep thus ;)

chiarcos commented 4 years ago

@Arithmeticus : I guess the standard solution in an RDF context would nowadays be to resort to Prov-O (https://www.w3.org/TR/2013/NOTE-prov-primer-20130430/). Prov-O is not directly nestable (provenance is defined in terms of RDF properties only), but it is possible to use RDF reification.

chiarcos commented 4 years ago

In any case, we should formulate an explicit recommendation on how to model provenance and confidence, preferrably using external vocabularies if anything established can be found.

pennyl67 commented 4 years ago

Confidence can become quite complex on its own: how it has been decided, tool/algorithm used to ascertain it, person who may have added it (if added by a person), etc. Do we want to go into such details?

chiarcos commented 4 years ago

@pennyl67 I think we can basically delegate this to Prov-O, but we should check that this really works.