Code4libBC Day 2: Talk notes

Code4libBC Day 2 talk notes.

CatalogerGPT: an AI powered cataloguing assistant as agentic collaboration

Glen Greenly, Capilano University Library

  • version 1: create a usable catalogue record
  • online version: glengreenly.wixsite.com/catalogergpt
  • version 2 in development
  • design goals:
  • later want to be able to navigate RDA toolkit
  • rapid advancement, so if there’s a limitation, wait 6-12 months
  • GPT-4o (Omni) recently added to the open AI API
  • Reinforcement Learning (RL) to make decisions to improve quality and correctness
  • improvement over time, becomes smarter
  • Retrieval-Augmented Generation (RAG): allow extensive knowledge not embedded in LLM
  • Explainable AI (XAI): provide reasoning behind answers
  • Knowledge graphs: organize and integration info
  • AI agents will get better, can assist in other areas of the library, including identify pain points
  • if don’t have an understanding capabilities, might not know how to apply it
  • conclusion: AI is here to augment, not supplant. You can’t manage or operate without human expertise.

Moving to the Cloud: Better late than never?

Scott Leslie, BC Libraries Coop

  • the question mark is the key point of the talk
  • was formed in 2009, run from colocated downtown
  • BC members are public entities beholden to
  • limited offering cloud servers that was FIPPA compliant, limited offering in Canada
  • people who formed the coop believed in the civic ownership
  • lot has changed after 15 years
  • many choices now
  • during the pandemic, FIPPA itself changed, lessened data crossing border
  • trend has been to move to cloud
  • finding people to maintain server farms requires talented, is expensive
  • been thinking about it since 2018
  • went through review of different providers
  • decided based on privacy, security, level of service
  • timeline based on existing contract
  • started moving in March, end date is probably October, decommissioning into February
  • not convinced it’s right, or the doubts
  • one of the doubts is control, don’t have physical access
  • currently have ASN, at hub; will no longer have a say in representing libraries
  • own vs. rent
  • the more we shift digitally, the physical operations may be seen as less important
  • at the end, it’s hard to argue with the trend, towards tenants, renters of infrastructure, services
  • allow us to focus time on other parts of the tech stack, provide value
  • buys more reliability
  • trend is to outsource
  • think about getting rid of control mean in the long term
  • post-sec have BC Net, don’t lose that

Migration journey: Moving our library to the Folio LSP

Olga Kalachinskaya, Douglas College Library

  • submitted budget request in 2021
  • RFP process and selection 2022
  • start implementation late 2022
  • went live June 2023
  • why Folio?
    • WorldShare:shared environment
    • Alma: expensive, 10 years old
    • lot of EBSCO resources and Primo doesn’t play nice with them
    • front end didn’t change
    • OSS, new, but has good potential
  • implementation: committee (10 people), 7 months, integration with EDS, Banner, etc., setup of LOCATE and Panorama
  • improvements needed: equipment booking, auto-renewals, inventory, stats, some reports can only be done via API
  • modern, does things differently
  • can somewhat participate in development through attending meetings where features are discussed
  • much more community work

OpenAthens and the Ironies of Automation

Ryan Regier, UBC Library

  • UBC migrated to OpenAthens from EZProxy in 2020
  • OpenAthens is SAML-based federation
  • benefits: ease of access, time savings, security, analysis
  • less problems, but more harder problems
  • Irony: making it easier has made it harder
  • issues are less frequent, but the ones that do happen are more difficult to investigate
  • operations have less understanding of how the system works, because they rarely interact
  • skills and knowledge deteriorate over time, and lower confidence
  • very specific user issues that we are unable to recreate or troubleshoot effectively
  • specific vendor website issues that never get fixed
  • outages suggest having same ironies
  • it’s new, will be growing pains
  • number of libraries using it will grow and documentation will get better
  • could be a UBC problem with giant user base and lots of resources
  • can’t teach troubleshooting in a couple of sessions
  • outsourcing too much and losing understanding of it?
  • possible solutions:
    • notification and reporting systems: listserv and resource status page
    • final irony: advance tech might require even more advanced tech to troubleshoot

Title list comparison using Pandas/Jupyter

James Fournie, VCC Library

  • evaluating resources, looking at packages
  • used EDS Package Analyzer
  • wanted to determine usage, overlapping titles
  • download as CSV file
  • populates a jupyter notebook
  • can run queries
  • data is questionable due to the source
  • doesn’t get to coverage dates
  • ISSN matching not great
  • had to match on title, fairly uniform

Telidon Update

John Durno, University Victoria Libraries

  • restoring art made in a now obsolete format called Telidon, from 80s
  • had worked on building a tech stack to display the art
  • turned out it was dozens of artists across the country
  • worked with gallery curator to get grant to convert format
  • web exhibit will open later this year, including art generator

Backup service down

Calvin Mah, SFU Library

  • 2022 Windsor, issue took down the library
  • University will prioritize HR and other key systems
  • bridge the gap between when every down to back up
  • set up a server in the cloud, has EZproxy, invoke handful of resources
  • Adding cloud IP address to range
  • authenticates off of Alma, which is in cloud

End of Day 2 talks

Thanks to all the presenters. Hope to see you next year!

crab