Spanish-language text classification for environmental evidence synthesis using multilingual pre-trained models

20990 6 111706 2 337657 document 111706 s13750-025-00370-9.pdf application/pdf e6148a1e515d57b964e075b2d363b33c MD5 2064507 2025-11-17 09:03:08 https://pure.iiasa.ac.at/id/eprint/20990/1/s13750-025-00370-9.pdf 20990 1 1 application/pdf text en public cc_by_4

s13750-025-00370-9.pdf

published 111719 2 337687 document 111719 lightbox.jpg image/png ecab1ce5d02f01f4c24548157b8d4515 MD5 47111 2025-11-17 09:13:11 https://pure.iiasa.ac.at/id/eprint/20990/2/lightbox.jpg 20990 2 2 image/png other Thumbnails conversion from text to thumbnail_lightbox en public

lightbox.jpg

preview.jpg

medium.jpg

small.jpg

indexcodes.txt

http://eprints.org/relation/isVersionOf https://pure.iiasa.ac.at/id/document/111706 http://eprints.org/relation/isVolatileVersionOf https://pure.iiasa.ac.at/id/document/111706 http://eprints.org/relation/isIndexCodesVersionOf https://pure.iiasa.ac.at/id/document/111706 archive 353 disk0/00/02/09/90 2025-11-17 09:04:48 2025-11-17 09:04:48 2025-11-17 09:04:48 article show Berdejo-Espinola V. Hajas Á. Cornford R. 2422 0000-0002-9963-3603 Ye N. Amano T. Spanish-language text classification for environmental evidence synthesis using multilingual pre-trained models pub BNR BEC Artificial intelligence (AI) is increasingly being explored as a tool to optimize and accelerate various stages of evidence synthesis. A persistent challenge in environmental evidence syntheses is that these remain predominantly monolingual (English), leading to biased results and misinforming cross-scale policy decisions. AI offers a promising opportunity to incorporate non-English language evidence in evidence syntheses screening process and help to move beyond the current monolingual focus of evidence syntheses. Using a corpus of Spanish-language peer-reviewed papers on biodiversity conservation interventions, we developed and evaluated text classifiers using supervised machine learning models. Our best-performing model achieved 100% recall meaning no relevant papers (n = 9) were missed and filtered out over 70% (n = 867) of negative documents based only on the title and abstract of each paper. The text was encoded using a pre-trained multilingual model and class-weights were used to deal with a highly imbalanced dataset (0.79%). This research therefore offers an approach to reducing the manual, time-intensive effort required for document screening in evidence syntheses—with minimal risk of missing relevant studies. It highlights the potential of multilingual large language models and class-weights to train a light-weight non-English language classifier that can effectively filter irrelevant texts, using only a small non-English language labelled corpus. Future work could build on our approach to develop a multilingual classifier that enables the inclusion of any non-English scientific literature in evidence syntheses. 2025-11-12 published doi:10.1186/s13750-025-00370-9 3568 public Environmental Evidence 14 1 e21 TRUE 2047-2382 info:eu-repo/semantics/article