AI Subject Generation: Alma AI Metadata Assistant vs. UCF’s Custom AI

AI Subject Generation: Alma AI Metadata Assistant vs. UCF’s Custom AI

In response to the integration of Artificial Intelligence (AI) into library cataloging and metadata, the University of Central Florida (UCF) Libraries has created a custom programming script to automate subject generation. This presentation updates and expands on the session delivered at ELUNA 2025, providing an overview of the project and presenting new findings from evaluations using additional datasets to compare UCF’s approach with the Alma AI Metadata Assistant. The custom script generates both Faceted Application of Subject Terminology (FAST) terms and keywords using the OpenAI API for traditional and digital collections, initially incorporating OCLC's FAST Reconciliation Service and later transitioning to a custom FAST headings vector database for term validation. The updated experience with both UCF's custom AI approach and the Alma AI Metadata Assistant will be shared, focusing on the evaluation of newly generated terms, features, advantages, and limitations of each method. This session will also examine their impact on metadata workflows, internal and external collaborations, and key lessons learned since the initial presentation. Attendees will gain valuable insights into integrating AI into cataloging practices and the latest practical outcomes of adopting these innovative approaches. 

Deng, Sai (Sai Sophie)

Piascik, Jeanne

ELUNA Learns 2026

ELUNA

2025-12-10

application/vnd.openxmlformats-officedocument.presentationml.presentation

eng

Text; Presentation