Attendees

  • Steven Folsom
  • Simeon Warner
  • Andreas Mace
  • Regine Heberlein
  • Bob Coret

  • Mathis??
  • Lars Svensson

  • Antoine Isaac

  • ?? Center for Life Science, Köln
  • Marcel, Frankfurt research
  • Dmitri Pavlov, Metaphacts
  • Sarven Capadisli
  • Jarmo Saarikko

Notes

At the Finnish National Library that they have a wiki page that says what has changed. Also a mailing list of changes with daily updates. Interested as protocol as way to share updates.

The SOLID Protocol has a notion of subscription to resources and indication of delta. Have looked at RDF diff, SPARQL, SOLID N3 patch, etc.. Discussion of updates in the SOLID has considered SOLID N3 patch because it is used elsewhere in SOLID. The SOLID notification protocol uses a lightweight version of the ActivityStreams protocol as the message.

What about SKOS semantics? The specification does not assume that all authorities use SKOS so it don't use SKOS semantics for updates.

What about notion of successor in the case of deprecate? Steven notes that we had an idea of specify the successor but it didn't cope with all complex cases and we ended up simplifying.

What about scale and usage? Why no push protocol? Working group started by imagining less frequent updates than that, perhaps daily. In Sweden they do updates for local libraries every minute which is faster than appropriate for polling.

When something is Deprecated, then could use `object` and `target` for a move as is done in the IIIF Change Discovery specification. It is noted that an Activity can have more than one `target` (https://www.w3.org/TR/activitystreams-vocabulary/#dfn-target). However, with more than one target the information would no longer be machine actionable if one `target` has to be selected, and the semantics are unclear.

Europeana has use cases such as updated linking that don't need as much data. Steven notes that there are cases such as SKOS vocabularies where just an editorial note changes and nothing else. There might still be a need to update a cache. In general, we haven't assumed how the data will be used. We have attempted to make sure that cataloger activities are covered, there could perhaps be some refinement of types of Update Activity to indicate more but we'd need to better understand the possible uses first.

Has anyone worked with FAST from OCLC? FAST work started before ActivityStreams and OCLC publish spreadsheets of changes with a bit of a taxonomy of change types, along with the MARC authority change files.

Comment that it would be interesting also to have a change set for VIAF. The data is computed so as far as participants know it can just change. There is change data available, for example: https://viaf.org/viaf/9847974/viaf.xml





  • No labels