About
The Premodern Concordance is a prototype for a larger project that will use computational methods to link named entities — people, plants, substances, places, diseases, and concepts — across multilingual early modern texts relating to natural and scientific knowledge. The goal is to make the unstable, polyglot terminology of premodern natural knowledge searchable and comparable across languages, centuries, and traditions, and then to explore new research possibilities and questions unlocked by this.
Our methodology is described here, and a full data set is available here (in progress).
Personnel
Benjamin Breen
Department of History, UC Santa Cruz
Mackenzie Cooley
Department of History, Hamilton College
Colophon
Typography
- System UI — body text
- UnifrakturMaguntia — logotype, blackletter display
- EB Garamond — early modern serif
- Space Grotesk — geometric display
Framework
- Next.js 16 — React framework
- Tailwind CSS 4 — styling
- Vercel — hosting
Research models
- BGE-M3 — fine-tuned cross-lingual embeddings for entity matching (current model was fine-tuned on 500 matched multi-lingual pairs of early modern natural concepts/terms)
- Google Gemini 2.5 Flash Lite — entity extraction, enrichment, and verification
- OpenAI text-embedding-3-small — semantic search index
Site code
Built with assistance from Claude Code (Claude Opus 4.5).
mercury · mercurio · azogue · vif-argent · hydrargyrum