University of Florida

The University of Florida will pilot a human-guided, Large Language Model-assisted bulk-work environment. The project will focus on building features to assign subject terms for text-based content. The project will add new access points to approximately 20,000 newspaper issues published in Pensacola and St. Augustine. The project will deliver a functional bulk work environment powered by NavigatorAI and HiPerGator and a website documenting the major technical decisions, as well as tested prompts and corresponding results. The project seeks to address critical needs for improving the accessibility of our digital collections at scale. This includes features such as generating transcripts for audio and video, creating alt-text for images, and providing content summaries for all file types.