michaelp2: Topic: Recap of DC Task Group session
michaelp2: Kai: People were interested in task group.
michaelp2: … Topic of metadata provenance was perceived as relevant
michaelp2: … Workshop itself was split in two parts:(1) presentation of several aspects, (2) open discussion of use cases.
michaelp2: ACTION: Add workshop slides to wiki
michaelp2: Kai: One of the central question was on how to model metadata provenance information vs. modeling “provenance as process” paradigm
michaelp2: … For the second part, Antoine Isaac did lightning presentation of provenance in Europeana project / EDM.
-
kai: See Slide 12 for the Proxies
kai: Michael: Europeana reinvented the metadata record by means of ORE procies
DGarijo> kai pls mute yourself
kai: … Concerns about such complicated, non-standard models, that hide the information from the user by not making them explicit
michaelp2: … No modeling of change information.
michaelp2: … EDM “sort of” applies FRBR through the backdoor. That the “manifestation” is modeled as ore:proxy is problematic in my view.
michaelp2: Kai: I liked the approach. If you have this information in your domain model, model it as first-class data.
michaelp2: s/this information/this kind of information
michaelp2: … OPM is very complicated, but if my metadata processes are that way, it might be a good fit.
michaelp2: … Idea in the discussion: Is it possible to implement a subset of a rich data model like OPM or EDM in a simple model ar metametadata?
michaelp2: … Can we do it in the spirit of Dublin Core, ie., extract the nuggets of core information and present it in a simple way?
michaelp2: Michael: We need a clear path back to OPM.
-
michaelp2: Daniel: Luc's charter for Prov is in draft. It is not clear if OPM is going to be the main vocabulary.
michaelp2: … There was a feeling the group the the OPM community has not made the effort to understand PML.
michaelp2: Michael: But people are very clear that they want to start with _a_ vocabulary.
michaelp2: Daniel: Question for Kai: How Do you mean to extract metadata from EDM to metadata prov metadata?
michaelp2: Kai: EDM needed to distinguish providers, so they extended the model.
michaelp2: … I think an alternative would be to introduce a metalevel, to distinguish between data about resources and data about “triples” data.
michaelp2: … This data can then be presented in a “Provenance Core”.
michaelp2: Daniel: What if the original data already contained metalevel information?
michaelp2: Kai: We have to think about that. No clear answer at the moment.
michaelp2: … What happens if metalevel information becomes part of the RDF standard and people start to use it for different things.
kai: Michael: Provenance Information has to be marked as such by classifying the graph as provenance graph (in case of named graphs)
michaelp2: … Let's think about that in terms of a domain model over the next couple of weeks (with visualization!).
michaelp2: … Provenance graph: Meant as attaching a provenance account resource to the standard named graph.
michaelp2: Kai: Can we start with translating the EDM use case?
michaelp2: Michael: I think we are close to a draft of a domain model.
michaelp2: ACTION: Work on first draft of domain model (low-key approach)
michaelp2: ACTION: Daniel to talk about some of his OPM use cases.
michaelp2: ACTION: Translation of Antoine's EDM ore:Proxy approach to metametadata
michaelp2: Daniel: We should highlight benefits of metamodel. Help seperating sources, how does this affect trust, etc.
michaelp2: Kai: Let's start working on Google Docs. Michael starts drawing with Inkscape.
michaelp2: … Meeting adjourned.