2025-07-23

Housekeeping

Melanie

Action items from last meeting:

Announcements:

• Call for TCDL planning committee is open! Reach out to Elliot is you or someone you know is interested. Call for Committee Members #TCDL2026

• Yogita will not be able to attend for the moment since she has a new commitment with an SAA group, but Josh and Katie can hopefully check in and share things as relevant.

• LD4 conference is next week! https://2025ld4.sched.com/ LD4 Community Site - Conferences Sessions on Wikidata, Arches, Alma Primo

  • Free and may be recorded - gather links

  • Put on agenda for next month meeting!

Updates from the field

From our last meeting:

• Library of Congress June 2025 BIBFRAME Update Forum Monday, June 30, 2025, 1:00 PM EDT - Recording link

  • any highlights/takeaways folks want to share?

  • Elliot attended, but nothing jumped out as key to share with the group

• ALA Linked Data sessions (June 26-30) (Paloma attended, will report out in August meeting)

  • OCLC cataloging community meeting – new developments in OCLC cataloging, including plans for BIBFRAME

  • The Mothman Properties: Exploring linked data at West Virginia University

  • BIBFRAME in Production: Practical Insights and Case Studies from UC Davis and Penn Libraries

  • Making It Real: Preparing for BIBFRAME in Public Libraries

  • PCC and Linked Data: Successes and Challenges of Implementing Linked Data

• ELUNA conference session highlights

  • Benn Chang attended and his notes are below with links to info:

  • Linked Data and AI: Harnessing the power of both for transforming Library Services

    • Checkout knowledge graphs with AI and RAG pipeline 

    • Model Context Protocol (MCP AI Agents) bypass RAG pipelines. OpenAI, Langchain agents, Google Vertex AI 

    • 8 billion agents in a couple years according to IBM. ExL should create an agent registry / catalog for reference and validation. 

    • ExL will eventually explore LD or Knowledge Graphs in the AI Research Assistant (this feature is no longer in Beta, but UTL has not deployed it yet). 

    • ExL metadata assistant may involve LD and URI insertion. Ideally cataloging can become LD 

    • Recommended embedding strategies can feed LLM

  • Navigating Challenges and Embracing Opportunities: Advancing Linked Open Data Initiatives in Libraries

    • Bunch of initiatives, products and services and workflows we have to live with ExL, OCLC, LC/PCC, ShareVDE/LD4P, EBSCO (other vendors) 

    • Build a roadmap 3-5 years 

    • Introducing Bibframe records, no good argument yet for big institutions to move to. Folks don’t know the records exist or how to make use. 

    • Very few Proofs of Concept out there. No support from Admin for mythical metadata fantasy. Dean says to make it low priority. 

    • Who’s going to go first? What’s next? Lack of direction, values gained. 

    • All these URIs created by different institutions all pointing to same entity. 

    • Chicken and egg. 

    • Author card in Primo is step in right direction.

• OCLC cataloging community meeting - Elliot attended, heard about effort to transform Dewey into linked data, for concepts

  • taking base numbers and using that as a conceptual category, giving those concepts a URI so that could be used for something other than books

    • Filling in gap in range of things where people are doing linked data

    • Higher level, broader conceptual

    • Public and school library communities as primary audience?

    • OCLC manages Dewey so that may be a factor

Continued discussion from last meeting

• Primo Person cards/pages

  • response from Misu Kim at UT Dallas:

    • not as exhaustive or formal discussion as we have had at UTL

    • their library is more exploratory, and usually activate new features unless there's a compelling reason not to

    • as part of their library’s Primo group to review new features for activation, she recommended turning on the person/author cards

    • has not gotten positive or negative feedback yet, BUT one request to correct a name displayed in person info card

    • wants to see if there is usage data about clicks on person cards

      • current MixPanel data (new analytics for Primo that is part of the new Next Discovery Experience) shows users are clicking creator links more than subject, etc. - could person cards be a factor in that?

  • shared Irene’s report with UTL Discovery Services Advisory Group - well received, concerns resonated

    • UTL response - not ready for deployment, so many cards don’t have substantial information

  • Added to Ex Libris Idea Exchange for Primo: Melanie submitted!

    • add an expandable disclaimer option in the Person Info Card panel with text that institutions could control

    • add an option for users to request more information about a person, giving user communities agency to prioritize authority work by information professionals - related to NACO “Lite” and EMCO

    • Sometimes ExLibris picks up these ideas for development beyond enhancements that are voted for each year

    • noting that UT Dallas person cards also have Wikipedia links

    • noting there is an active LoC bot adding LC ID statements to to Wikidata records (Josh mentioned LD4 session next week)

• Entity reconciliation for place names and venues

  • Ella shared Ransom Center efforts last month, nothing new to report this month, but happy to continue conversation in future

  • Devon has an update - UT has Spanish terms translation project with University of Florida and UC San Diego called PANA (Pan American Auhorities) - LoC source code, planning to work on an API to be able to reconcile terms to PANA, which will take time

    • recently met with UCLA and they shared terms, project called HAPI (recently lost NEH grant)

      • lots of crossover, so may be opportunity for merger or cross-citation!

  • May be good to continue conversation in future meeting!

• Partnering with UTL Research Data Services on Linked Open Data efforts

  • Katie - geospatial datasets as foundational for us - done win collaboration with Michael Shensky

  • Some institutions have focused efforts on faculty publications - probably way to get institutional buy-in

Analysis of Wikidata item changes

Josh Conrad

Presentation, link: Wikidata analysis 2021-2024.pdf | Powered by BoxPreview

• Josh gave previous presentation in 2021 to this group: 2021-07-28Preview

  • gave presentation at Digital Frontiers in 2021

  • modeling people, buildings, events

  • found Wikidata structure works for

  • gazetteer that has data extracted from several collections so far

  • have been working on project over time for years

  • now focused on getting all data ready for ArchivesSpace, which will hopefully facilitate getting additional or all collections represented in Wikidata

  • record of uploaded batches to Wikidata thus far - over 6900 records

  • 6009 have been edited, 903 unedited (this number of edits was unexpected)

  • Field edited the most: given name, likely because we included names in title, but didn’t always specify given name

  • Edits: lots of bot edits, lots adding translations of titles and descriptions, a lot adding IDs, some adding geographic data (parsing geodata into separate fields)

  • Manual superusers: Fralambert and Thierry Caro - both very active with heritage data throughout the world

  • Analysis: SPARQL query - March 2024 - all Wikidata items that have statements that cite AAA

    • Downloaded edit history from MediaWiki API

    • Want to rerun script and see what updates have been made over the last year

  • Virginia San Fratello - one of top records with changes

    • many IDs, including art and architecture specific as well as LoC, VIAF, etc.

  • So far, have not found edits that he finds problematic, they are all potentially valuable, BUT

    • Sex or gender - one statement that has been added, many times by bots based on names

      • HRC has legacy metadata about sex/genders as well as ethnicity and nationality of subjects of works - likely added without knowledge of the self identification of the subjects

      • Devon has best practices for trans/gender diverse resources guide - they can share!

        • long and short - not to record it and remove info if it was assumed without clear/obvious notation

      • Suggestion to request sex or gender property in Wikidata be split apart

  • Ella - have you found instances where you find records that exist but they are not structurally or conceptually what you are are wanting?

    • We definitely have instances where we see records for individual architects and related firms or a building / company that occupied the building being confused

    • It would be great to have some follow up discussions!

Wrap-up

Next meeting is August 27th