Fully automatic extraction of morphological traits from the web: Utopia or reality?

• Published in: Applications in plant sciences Vol. 13; no. 3; pp. e70005 - n/a
• Main Authors: Marcos, Diego, van de Vlasakker, Robert, Athanasiadis, Ioannis N., Bonnet, Pierre, Goëau, Hervé, Joly, Alexis, Kissling, W. Daniel, Leblanc, César, van Proosdij, André S. J., Panousis, Konstantinos P.
• Format: Journal Article
• Language: English
• Published: United States John Wiley & Sons, Inc 01.05.2025; Wiley; John Wiley and Sons Inc
• Subjects: Application; automatic trait extraction; Automation; Collaboration; Computer Science; Datasets; Encyclopedias; Flowers & plants; Large language models; morphological trait matrices; Morphology; Natural language processing; Physical characteristics; Search engines; Species; Taxonomy; large language models; automatic trait extraction; morphological trait matrices; natural language processing; Natural language processing; Large language models; Automatic trait extraction; Morphological trait matrices
• ISSN: 2168-0450; 2168-0450
• DOI: 10.1002/aps3.70005

Premise Plant morphological traits, their observable characteristics, are fundamental to understanding the role played by each species within its ecosystem; however, compiling trait information for even a moderate number of species is a demanding task that may take experts years to accomplish. At the same time, online species descriptions contain massive amounts of information about morphological traits, but the lack of structure makes this source of data impossible to use at scale. Methods To overcome this, we propose to leverage recent advances in large language models and devise a mechanism for gathering and processing plant trait information in the form of unstructured textual descriptions, without manual curation. Results We evaluate our approach by automatically replicating three manually created species–trait matrices. Our method found values for over half of all species–trait pairs, with an F1 score of over 75%. Discussion Our results suggest that large‐scale creation of structured trait databases from unstructured online text is now feasible due to the information extraction capabilities of large language models. However, the process is currently limited by the availability of textual descriptions that cover all traits of interest.