Technical Metadata Call 2015-04-28

Time: 8:30am PDT / 11:30am EDT

Call-In Info: Google Hangout: https://plus.google.com/hangouts/_/event/cpt6do5qc6l6mt4g853rqolk5tk

Moderator: Aaron Coburn (Amherst College)

Notetaker: 

Attendees:

• @Former user (Deleted) (Amherst College)

• @Juliet Hardesty (Indiana University)

• @Corey Harper (NYU)

• Hugo Manguinhas (Europeana)

• @Juliet Hardesty (Artefactual Systems)

• @Sharon Farnel (U. of Alberta)

• @Jon Stroop (Princeton)

• Dallas Pillen (U. Michigan)

Regrets:

• @matienzo (DPLA)

• @Nick Ruest (York University)

Agenda:

1. Review proposed technical metadata application profile

   1. premis vs. ebucore for equivalent properties

   2. Additional baseline properties

      1. premis:Agent / prov:SoftwareAgent

      2. nfo:fileUrl / rdfs:seeAlso / edm:isRepresentationOf

   3. Specialized properties (more on this later, after the baseline is complete)

      1. dct:extent / exif:width / ebucore:width

      2. exif:bitsPerSecond

      3. ...

2. Review sample mappings: europeana, others?

3. Recommendations for workflow tools

   1. FITS

   2. Europeana: https://github.com/europeana/contrib/tree/master/ntua/mediachecker 

4. NonRDF Technical metadata and PCDM (via pcdm:hasRelatedFile)

   1. structural predicates

      1. edm:isDerivationOf / prov:wasDerivedFrom

5. ...

Actions

• ACTION ALL: Capture questions / comments in the google doc.
   https://docs.google.com/document/u/2/d/1SZCpSIdlGfXgoYrAnW2eRKlIt6O-1ADIDDhmLrvxeLc/edit#heading=h.a8hurtypz8qi

• ACTION: All look at some edge cases. A Web Archive. 

• ACTION: Jon: Images. Hardware & Sofware Questions.

  • Diff between image production, image capture, and image software. 

• ACTION: Aaron/Jon/Julie: Note to the sprint group by Thursday giving update:

  • Here's what we've done

  • Here's our open issues

  • Here's our next steps

Notes:

• First cut at Application Profile:
  https://docs.google.com/document/u/2/d/1SZCpSIdlGfXgoYrAnW2eRKlIt6O-1ADIDDhmLrvxeLc/edit#heading=h.a8hurtypz8qi

  • Baseline core properties

  • Based on Justin's spreadsheet for documnt, media, AV

  • LDCX, people looking at NFO & Pronom

  • PBCore combining with ebucore. This also met Europeana use cases, so baseline draws on EBUCore.

  • AP also includes equiv props from other namespaces

  • Q from Aaron: Premis already supported by and used in Fedora. Do we want to privledge one or the other of Premis v. EBUCore

    • Julie: potentially better to do Premis if already used. Maybe focus those on preservation?

    • Jon: Another approach is to find the vocab closest to having everything we need, but causes least domain strife

    • Consensus seems to be using EbuCore as much as possible, with understanding that some Premis should map to equiv.

  • Corey: Needs to be possible to override a property choice.

  • Q from Jon: How to implement over-ridability. 

    • Corey: I suggest that the property definition lives in Hydra. 

    • Jon: But silly to recreate Fedora-wheels. Maybe map properties to Fedora Internals. Acknowledge overlap.

  • Q from Hugo: Being aware of type inconsistencies. Example, ebucore:fileSize specified as xsd:double, though Europeana's profile expresses as bytes, which are xsd:long

  • Name of original file? Where does this go? What is ebuCore:filename? How does map to Premis:hasOriginalFile. Property to map full path? 

    • ACTION ALL: Capture questions like these as comments on the google doc. 

• Specialized properties:

  • These should end up covered by EbuCore 

  • We can probably persue those more slowly

  • Jon: Make sure there's not a breaker at a lower level. 

  • ACTION: All look at some edge cases. A Web Archive. 

  • ACTION: Jon: Images. Hardware & Sofware Questions.

    • Diff between image production, image capture, and image software. 

  • Justin: Question about predicates defined specific to types, or just a long list of predicates. See barmintor's PCDM PR.
    https://github.com/duraspace/pcdm/pull/2

  • Justin: Define best practices per doc type

• NonRDF Tech Metadata. 

  • Tools that Generate non-RDF: FITS, etc.

  • How to do this

  • pcdm:hasRelatedFile for things that are related but not the object itself. 

  • Aaron looking to other vocabs to structure it more. Multiple related files. Example of a file with & without a color target. Tech MD about each.

  •  edm:isDerivationOf; prov:wasDerivedFrom.

  • derivatives are different from tech MD generated from an object. Perhaps we grab an "isMetadataFor"

  • Possibly rdf:Description, or ore:describes, (or even foaf:focus?)

    • Meh, ORE Describes has domain of a resource map...

    • http://lov.okfn.org/dataset/lov/terms?q=describes

• Other General Topics:

  • Jon Q: Property that might be missing. Byte Order.

  • PBCore is working with EbuCore -- may be able to get stuff added. 

  • Sweet from Jet Propulsion Library has Byte Order. Jon also comes back to it for representations of units of measure.
    http://sweet.jpl.nasa.gov/2.2/reprDataFormat.owl#byteOrder
    https://sweet.jpl.nasa.gov/sweet2.2

    • Maybe good to find another source, since we don't want to import giant vocab for one property.

  • EBUCoreHash vs. a Premis Proeprty

  • Convo about Provenance -- Defer to next week? Software agent, etc. 

• Do we need to get everything to them by the end of this week? 

  • ACTION: Aaron/Jon/Julie: Note to the sprint group by Thursday giving update:

    • Here's what we've done

    • Here's our open issues

    • Here's our next steps