Wikidata:WikiProject LD4 Wikidata Affinity Group/Affinity Group Calls/Meeting Notes/2022-02-22

From Wikidata
Jump to navigation Jump to search

Call Details

[edit]
  • Date: 2022-02-22
  • Topic: LD4P3 Grant work to use Wikidata in knowledge panels in Cornell's library catalog
  • Presenters: Huda Khan & Astrid Usong

Presentation Materials

[edit]

Meeting Notes

[edit]
  • Background
    • Work done as part of Linked Data for Production: Closing the Loop (LD4P3)
    • How to help users better discover items and relationships to other items in the catalog
  • Motivation
    • Ways to support bringing in contextual information about authors and subjects
  • Author panel
    • Additional descriptive data coming from Wikidata
    • Links to related works in catalog
    • Call number link comes from relationship to Library of Congress–id.loc.gov
  • Subject  panel
    • Description comes from DbPedia or if none, then Wikipedia
    • Call number link comes from relationship to Library of Congress–id.loc.gov
    • Focusing on API calls that provided a URI
  • Client side requests
    • Begin with what’s in index (not the identifier, but request to get that put in)
    • Look up URI  in LOC based on text string heading
    • Query Wikidata using the identifier to see what data can get
    • Use QID and label to query DBpedia
    • Call number link depends on retrieving JSON-LD and retrieving classification field
  • Author info slide
    • White boxes literals
    • Black boxes URIs
  • Wikidata and LC Connections
    • 1,340,603
    • LC subjects 51,864
    • LC names 1,289,158
  • Cornell author and subject indices
    • In Cornell’s catalog, authorized headings have info buttons
    • Authorized author headings 4,999,524
    • Authorized subject headings 1,720,035
  • Usability testing
    • 5 undergrads
    • Conversations with user representatives–9 librarians from across the library
    • Showed them demo prototypes and incorporated feedback
    • Went back to show final design mockups
    • Many conversations to coordinate the work
  • Knowledge panel
    • Originally much busier
      • Author panel included image and data from Wikidata and publication timeline, graph of influence relationships
      • Subject panel included more data–temporal and geographic information using LC and PeriodO
    • Generated mock-ups
    • Original author layout better matched catalog
  • Feedback
    • Positively received
      • Greater interactivity with map and timeline, and publication timeline–could be confusing at times for users
    • Influence graph
      • Discomfort putting it in catalog
        • Incomplete
        • Arbitrary
        • Wanted sources/links to articles
    • Publication timeline
      • Visually appealing
      • Wanted more interactivity–being able to click on publications
      • Data incomplete
      • How to align what’s in Wikidata with what’s in catalog
  • Factors supporting integration
    • Process
      • Identify stakeholders
      • Determine how features suggested/improved
      • Identify development processes–teams, sprints, etc.
      • Cornell experimental work had previously been separate from production, so needed to coordinate extensively with that team  
    • Feedback
      • Get clarification, understand their perspectives
      • Iterate based on their feedback
    • Luck
      • Good relationships with user representatives
      • Understood the development process used by production team
  • Experimental index
    • Included LCSH headings and equivalent Wikidata URIs
    • Didn’t make it into production
  • Q&A
    • Subject info boxes don’t show up on item page
      • Want to do that
    • Do you have a sense of what kinds of subjects are well represented in the links and which are not? I can see it working well for something like an event (e.g. a war) as in your example, maybe it’s trickier for more abstract kinds of subjects. Also, is this working off LCSH rather than FAST?
      • FAST used in catalog, but not used for this
      • Which subjects are well-represented
    • Can you control the intake? Can you somehow communicate which names and headings you want enriched in the catalog? (Instead of turning off what you don't want?)
      • Works in the specify what you don’t want direction, but not the other way
    • Are you concerned about response time/reliability of the live queries and API calls?
      • Do have a timeout with DBPedia
      • So far seems not too bad
    • In order to collect info about use of clicks by users and links, it can be necessary to implement cookies
      • Doesn’t believe using cookies, but will check
    • Will you continue incorporating external data as part of the catalog?
      • Yes would like to continue coming up with ideas and it makes sense and stakeholders agree and feel confident about what’s being represented
      • Have had suggestions about adding additional information
        • Subjects
        • Additional authors beyond primary authors
        • Would require analysis and design
    • What kind of UX work already done
      • Huda and Astrid and Tim Worrall did UX work for the grant
      • For each prototype, try to incorporate some usability feedback
    • FOLIO–incorporate the work?
      • Huda thinks would be great, but someone else would have to weigh in on how to make it happen
    • Is it histropedia.com that the publication timeline pulls from?
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:WikiProject_LD4_Wikidata_Affinity_Group/Affinity_Group_Calls/Meeting_Notes/2022-02-22&oldid=1937906807"