TDWG2026
Home / Talks / FORUM
talk · Thursday 24 September · 14:02–⁠14:14 · FORUM

The future is the future - thought about access to data in publications

Donat Agosti · Planning the Libroscope: creating research ready biodiversity from scientific publications

Recording time 5:41:53–5:53:16Open on Vimeo ↗

Summaries and key moments are machine-produced. Check the recording for the speaker’s exact words.

The short versionLiberating data from literature works, but the future lies in use-case-driven, digital-first publishing that also targets the undescribed majority of species.

Overview

What this was about

Donat Agosti introduced the Libroscope and the Disentis roadmap goal of making 100% of major biodiversity publications open, FAIR, machine-actionable and AI-ready by 2035, using a use-case-driven approach. He described Plazi's current pipeline, in which PDFs or XML are processed so that data reach GBIF, Zenodo and dashboards within a day of publication, with provenance down to the exact position in the text, and noted 1.2 million treatments extracted and a steady ~10,000 new species a year across roughly 200 journals processed. He argued the community should also move beyond extracting from PDFs toward digital-first description of unknown and dark taxa.

Why it matters. Literature holds much of what we know about biodiversity yet remains the least open data type; this talk frames a roadmap for making it trusted, AI-ready data.

Key ideas

In the room

  • Estimates: 500 million printed pages, ~2,000 taxonomic journals, 20-50 million treatments, and only 10-30% of species data accessible.
  • The 2024 Disentis meeting produced a roadmap to make all major biodiversity publications open and FAIR by 2035.
  • The Libroscope is use-case-driven: target specific data rather than everything at once.
  • Around 30 GBIF hosted portals are dashboards from publications; material citations link back to exact text positions via persistent identifiers.
  • Zenodo communities (e.g. for fruit flies) and the Biodiversity Literature Repository collect publications for processing.
  • Lag from publication to GBIF has fallen from years to under a day in the best case.
  • 1.2 million treatments extracted; ~200 journals processed over 10 years show a flat rate of ~10,000 new species a year.
Jump in

Notable moments

In their words

Transcript

Automatically generated captions can contain mistakes, especially in names and technical terms. Times are relative to the room recording.

Read the transcript ↓
Loading transcript…
Keep exploring

Related talks.

See the map ↗
talk · 24 Sept · 14:14–⁠14:26 · FORUM

Biodiversity literature as FAIR and AI-ready data by design (and how to get there)

Don't just liberate biodiversity literature, publish it liberated.

Chris Le Coquet
talk · 24 Sept · 14:38–⁠14:50 · FORUM

Building digital-native species: Transitioning from PDF liberation to semantic-by-design publishing

Keeping data linked and structured from specimen to publication yields FAIR, trusted data on the day of publication.

Laurence
talk · 24 Sept · 14:26–⁠14:38 · FORUM

Realtime Access to Data in new Taxonomic Publications

Plazi can deliver interlinked, machine-actionable data from new taxonomic papers within hours, and XML-first publishing removes most of the costly effort.

Guido Sautter
Programme abstract

The future is the future - thought about access to data in publications

Donat Agosti

Scientific publications contain a large share of humanity’s knowledge on biodiversity and are becoming increasingly important as training and reference sources for Artificial Intelligence applications. Yet most biodiversity literature remains difficult for machines to access and interpret. Estimates suggest a corpus of hundreds of millions of printed pages, with 50,000-200,000 biodiversity-related articles and around 2,000 taxonomy-relevant journals published annually. Resources such as the Biodiversity Heritage Library and Plazi demonstrate both the scale of the challenge and the feasibility of extracting structured biodiversity knowledge from publications.

Biodiversity publications not only describe taxa and synthesize research, but also cite specimens, observations and datasets, thereby linking natural science collections, occurrence records and scientific interpretations. However, most publications are still produced primarily for human reading and lack machine-actionable semantic structure. Unlocking this knowledge at scale therefore requires strategic prioritisation.

Within the framework of the Disentis Roadmap, which aims to liberate biodiversity knowledge from the literature by 2034, key questions arise: should limited resources focus on continued digitisation of historical print literature, semantic enhancement of legacy publications, or the promotion of future-ready publishing workflows that produce structured and FAIR-compliant data from the outset, as demonstrated by initiatives such as Pensoft and CETAF?

This contribution argues that a practical compromise is the development of a Biodiversity Libroscope: a framework to systematically digitise, extract and publish semantically standardised biodiversity knowledge from well-defined corpora of literature.

Submitted to the conference programme before the talk, so the talk itself may differ.