Versions Compared

Old Version 6

changes.mady.by.user E. Lynette Rayle

Saved on

compared with

New Version Current

changes.mady.by.user Simeon Warner

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Date: 

Attendees: Jason, Steven, Huda, Lynette, John, Tim, Simeon

Regrets: 

Agenda & Notes

Review actions from 2020-05-22 Cornell LD4P2 Meeting notes

  • See below

Status updates and planning

  • Enhanced Discovery - WHAM! (see also https://wiki.duraspace.org/x/sJI7Bg and https://github.com/LD4P/discovery/projects/1)
    • See: Organizing doc and Pseudonym thought. Updates also on running notes page
    • Usability - discussions with TIm and Kevin. We have multiple use cases but relatively straightforward features. Astrid will help out too. Document with testing plans
    • Indexing - Huda has been working on pseudonym and see-also cases so that search results from Solr more closely match UI needs. Need to check this in local VM but then need to update main suggest index. Open question about throwing out URIs not currently in catalog then when anything new is added one would have check for it - not sure which strategy is better but perhaps not important to deal with updates now. Related, think that a 2-phase index process is best for now
    • Question about whether to spend time on investigation of stemming and phonetic settings. Tim notes comment from earlier studies about interest in coping with spelling errors. Control scope for now, copy what we do on the main D&A system
    • John notes difference between Blacklight twitter type-ahead (sub library bloodhound) and the type-ahead that Tim wrote. We want to retain the system of headings that Tim developed (which likely rules out bloodhound) but then there is still a need to decide whether to use twitter type-ahead and whether to use jquery or not → suggestion that initially we should try to put in Tim's code directly and consider refactoring for twitter type-ahead later
    • Tim just restarted work on rspec tests for expected results
    • Plan for any index updates next week, user tests the following week
  • Authority Lookups for Sinopia (Lookup infrastructure: https://github.com/LD4P/qa_server/projects/2, Authority requests: https://github.com/LD4P/qa_server/projects/1)
    • QA performance
      • 2020-05-22 Dave is working through LoC authorities with new caching scheme, expect to hear about now
    • QA accuracy
      • 2020-05-29
        • Pretest results (see image in QA/Sinopia weekly notes).  Summary 60 tests run with 26 failing.  Of the 26, 4 were close to passing with the position off by 1 or 2.  The other 22, the expected subject URI wasn't in the results at all.  Geography data is especially bad at providing unexpected results regardless of whether it is coming from the cache or direct from the authority.  10 or the 22 failing tests were searching for a place.
        • Dave has updated the indexing for all LOC, AGROVOC, NALT, Getty, dbpedia, MeSH, RDA, CERL, and Ligatus.  These are ready for a post test run to see what impact it has on the test results.
    • Authorities
      • 2020-05-22
        • Have worked on CERL issues with Sinopia that will be released soon, otherwise mostly working on Exhibits
      • 2020-05-29
        • There are authority requests pending in the QA Issues.  Steven Folsom will work on prioritizing this work.  
        • Interest from Paul Walk fro Antleaf has a client interested in using QA to access Wikidata/Wikibase.  I pointed him to our code in QA for out initial connection to the simple Wikidata API.
  • Linked Data API Working Group
  • LD4P3 Planning
  • Meetings (see LD4P2 Cornell Meeting Attendances)
  • Next meetings
    • ACTION - Simeon Warner - Need to update Zooms for future meetings