Many cultural institutions have made large digitized visual collections available online, often under permissible re-use licences. Creating interfaces for exploring and searching these collections is difficult, particularly in the absence of granular metadata. In this presentation, Tilton and Arnold introduce a method for using state-of-the-art multimodal large language models (LLMs) to enable an open-ended, explainable search and discovery interface for visual collections. Through a case study using a collection of 170,000 documentary photographs from the Farm Security Administration (FSA) and the Office of War Information (OWI) in the United States, Tilton and Arnold provide examples of their work and initiate a roundtable discussion.

The talk will be conducted in English.

• Lauren Tilton, University of Richmond: https://laurentilton.com
• Taylor Arnold, University of Richmond: https://math.richmond.edu/faculty/tarnold2/