Possibilities to enhance the semantic navigation of BHL: further developing the 'unit of work'

What this was about

Nicky Nicolson, starting from Kew's IPNI (funded by Darwin's legacy) which indexes names to the exact page, argued that BHL has been built primarily as a library (volumes, pages, then articles via BioStor), whereas as a biodiversity data source its pages are richly structured, containing people directories, multilingual glossaries and gazetteers that are not best reached via articles. She contrasted bottom-up entity discovery across 64 million pages with top-down, community-driven subsetting, and proposed that BHL learn from GBIF's hosted portals: branded, pre-filtered views that let communities and funders see their slice (e.g. DiSSCo UK), which could also address contributors like Madrid who keep content out of BHL's viewer for branding reasons. Comments from Donat Agosti (GBIF hosted portals for journals) and a TaxonPages developer (need for BHL API endpoints) followed.

Why it matters. Rethinking BHL's units of navigation beyond volumes and articles could expose the structured entities in literature to the communities that need them.

In the room

• IPNI indexes to the exact page and structured data, not just the article.
• BHL was populated as a library: scanned volumes, OCR, then post-accession article identification.
• Name indexing is to name strings, not the structured entities they represent.
• Pages often function as printed databases: people, glossaries, gazetteers.
• GBIF (25) vs BHL (20): GBIF used later years to stitch data together (e.g. clustering) and build hosted portals.
• Funders want acknowledgement ('does it acknowledge my investment?') and users ask 'is it relevant for me?'.

Notable moments