Within biology as a whole, BHL probably has closer links to taxonomists than to other disciplines. After all, BHL is full of original descriptions of new names, taxonomic revisions, and keys to flora and fauna. But the last few years have made it clear that there is a great need for structured information about all aspects of an organism’s biology, linked to the source of that information.
The diagram below (taken from a talk at TDWG 2026 by Robert Waterhouse) illustrates this nicely. BHL is potentially a rich source of this information, the challenge is how to extract it. Finding taxonomic information turned out to be relatively “easy” to solve: index the whole corpus by taxonomic name, search for your name, find the oldest occurrence of that name, there’s a good chance you have landed on the original description of that taxon. But what about traits? How do we index BHL by ecology, life history traits, host-parasite relationships, size, etc.? Do we annotate the whole corpus with inferred concepts and relationships? What if we want to search for a concept we haven’t annotated? Do we then have to reindex for this new concept? Or can we search “semantically” where we don’t specify meaning when we index the corpus, but instead process it into vectors such that we can search for text with similar meaning to our query (effectively this is the same approach used to search by images, see Experiments on future ways to search BHL)?
The demand for this class of data represents an opportunity for BHL. How should it respond? Should BHL develop tools and/or interfaces to support discovery of this data? Or should it focus on delivering best quality OCR text and leave processing and indexing to others? How would BHL make that decision?
“Should BHL develop tools and/or interfaces to support discovery of this data? Or should it focus on delivering best quality OCR text and leave processing and indexing to others?” This is my biggest question for BHL as we work on platform modernization. BHL’s area of expertise comes out of libraries and archives. Should BHL’s focus be on the “content in” part of the equation and more robust/reusable “data out”, and let others build the interfaces and pipelines to process and integrate the data into other initiatives?
@Colleen_Funkhouser I agree this the key question. It has implications for what BHL is, but also I suspect what kind of future it has. One concern I have is that because BHL makes all its data open, and all the content is open, there is nothing stopping anyone building their own BHL. Indeed, during the worst of the IA outage and bot attacks I built my own mini-version out of frustration with lack of access. That it’s possible to build a BHL of one’s own is both a testament to BHL’s commitment to open data, but also a potential risk (why fund BHL when there is a newer shinier version?). The library pipeline and expertise can’t be copied in the same way, but that’s probably not a big enough moat to keep BHL “safe”, should someone decide to “clone” it.
Asking ChatGPT, it suggested three possible futures for BHL:
Infrastructure BHL: Focus on digitisation, bibliographic curation and providing high-quality, reusable open data, leaving sophisticated discovery and analysis to others.
Full-service BHL: Combine the core library role with advanced search, extraction, annotation and other tools that make BHL itself the primary place for discovering and using biodiversity literature.
Platform/partnership BHL: Maintain the core data and selected discovery infrastructure while partnering with others to develop new capabilities and integrating their outputs back into BHL.
Where BHL sits is a bit fuzzy as it is certainly infrastructure (arguably if BHL does nothing this is where it will ultimately end up), it does have partnerships which bring it names and articles (and hopefully soon OCR) so is a platform, but having full text search is a service that BHL built, so that’s closer to full service. Perhaps this indicates that the threefold division ChatGPT suggests is a poor fit to BHL, or perhaps it suggests that BHL has not really figured out which of these things it wants to be and hence has fingers in each pie.
What kind of entity BHL wants to be also affects funding. Maybe the mix of membership from libraries with limited (and possibly shrinking) budgets, plus in kind contributions is enough to sustain BHL as infrastructure? But does that route make BHL vulnerable to loss of partners and its content being presented in more engaging ways elsewhere?
Perhaps a more interesting question is to what extent technical choices about how to build the next BHL would constrain these choices. If the kinds of services and interfaces I would love to see are essentially “just add more disk space and RAM, and turn on feature x” then the discussion is moot. But if the technical choices make it harder to add that functionality down the track, the choice of what BHL wants to be matters a great deal.
If anyone has read this far, you deserve a medal for making it through to the end of my lack of coffee, sitting at the airport, barely coherent musings.
