Skip to content
Hunt Institute for Botanical Documentation
A Research Division of Carnegie Mellon University

Hunt Institute Archives Text Discovery Platform

Search a large and growing portion of our online collections, including handwritten documents.
THIS IS A PROTOTYPE
This application lets you search the textual contents of our Archives’ online collections. Both handwritten and printed or typed materials were transcribed using a cutting-edge vision-language model (VLM), so you can search across both. In practice, this means nearly everything we have made available online is now searchable. Because this is a prototype, functionality and quality will continue to evolve and improve.
How to use it
  • Search behavior: Results include pages containing all your terms. Use quotes for an exact phrase. Substring matching is supported (e.g., "aceae" finds plant families; "and grass" may match and grasslands or wetland grass).
  • Open a result: Click the larger “View item” link (e.g., “View item · DO #987 · 319_Love_Bx2FF32r · page 50”) to see metadata and the PDF. Then scroll the page (not just the PDF frame) to the Transcript section with your terms highlighted. This may help you locate them in the original document.
  • ArchivesSpace: The other links (Collection, Item/Folder, Digital Object) take you to the corresponding record in our Archives Collections Database (ArchivesSpace). Collection Dates reflect the entire collection, not the specific item or page.
  • If a PDF does not load: On the detail page, use the Digital Object link, click “Go to file” in ArchivesSpace, and navigate to the page number shown in this interface next to Digital Object (e.g., “View item · DO #1257 · 53_Arber_AN22r · page 77”).
Current limitations
  • Due to the nature of today's VLMs, some transcriptions may contain partial text or unintended repetition.
  • Handwritten text recognition (HTR) has improved greatly, but transcriptions will still have errors. We will continue trying to improve the models and data.
  • Search is currently a straightforward keyword/phrase match (AND semantics with substring support). Future versions may add finer-grained controls and additional search modalities.
  • As an active prototype, the service and interface may change frequently, and the site may occasionally go down for a few minutes at a time.

Search across all page transcripts. Supports exact phrases in quotes (ex. "Allium giganteum").