Your Name and Title: Noelle Leong, Reference Librarian
Organization Name: California State University, East Bay CORE Library
Co-Presenter Name: n/a
Area of the World from Which You Will Present: SF Bay Area, CA
Language in Which You Will Present: English
Target Audiences: Those interested in natural history collections, scholarly communications, OER, linked metadata
Short Session Description: Possibilities with Training AI to Identify Natural History Specimens and Link Article Metadata with the Biodiversity Heritage Library
Full Session Description (as long as you would like): Publishing a niche ecological study can feel like it's entering a circle of the same papers and citations at best, and a scientific vacuum at worst. With the help of Drexel University and the Smithsonian Libraries & Archives, I had the opportunity to try building on past work to build a tool and train an name-entity recognition to eventually be able to recognize specimen ID numbers in PDFs with the eventual purpose of being able to link papers to their specimens. What could it mean for this type of linked database with overlapping metadata to be used in tandem with GIS mapping? With journal databases? For new researchers studying these interconnected relationships? AI has its limits, but with intention, could help create a more connected research landscape.
Websites / URLs Associated with Your Session: [Github link forthcoming]
Replies