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

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.

• Publication year

  2025

• Venue

  Environmental Evidence

• Publication date

  2025-11-12

• Fields of study

  Medicine, Computer Science, Environmental Science

• Identifiers
  DOI 10.1186/s13750-025-00370-9PMID 41225673PMCID 12613578
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar, PubMed

• No claims are published for this paper.

• No concepts are published for this paper.

• Assessing diverse values of nature requires multilingual evidence

  2025cited by this paper
• Careful design of Large Language Model pipelines enables expert-level retrieval of evidence-based information from syntheses and databases

  2025cited by this paper
• Testing the utility of GPT for title and abstract screening in environmental systematic evidence synthesis

  2025cited by this paper
• OpenAI's 'deep research' tool: is it useful for scientists?

  2025cited by this paper
• Large language models help facilitate the automated synthesis of information on potential pest controllers

  2024cited by this paper
• Language inclusion in ecological systematic reviews and maps: Barriers and perspectives

  2024cited by this paper
• Testing the reliability of an AI-based large language model to extract ecological information from the scientific literature

  2024cited by this paper
• Large language models overcome the challenges of unstructured text data in ecology

  2024cited by this paper
• Performance of two large language models for data extraction in evidence synthesis

  2024cited by this paper
• Human-AI collaboration to identify literature for evidence synthesis

  2024cited by this paper
• Methods for using Bing's AI‐powered search engine for data extraction for a systematic review

  2023cited by this paper
• Data extraction for evidence synthesis using a large language model: A proof‐of‐concept study

  2023cited by this paper
• Multilingualism for pluralising knowledge and decision making about people and nature relationships

  2023cited by this paper
• The role of non-English-language science in informing national biodiversity assessments

  2022cited by this paper
• Growth of non‐English‐language literature on biodiversity conservation

  2022cited by this paper
• Culturally diverse expert teams have yet to bring comprehensive linguistic diversity to intergovernmental ecosystem assessments

  2021cited by this paper
• More Than Words: Towards Better Quality Interpretations of Text Classifiers

  2021cited by this paper
• Tapping into non-English-language science for the conservation of global biodiversity

  2021influential reference
• Machine learning reduced workload with minimal risk of missing studies: development and evaluation of a randomized controlled trial classifier for Cochrane Reviews

  2020cited by this paper
• Poor availability of context-specific evidence hampers decision-making in conservation

  2020cited by this paper
• Making Monolingual Sentence Embeddings Multilingual Using Knowledge Distillation

  2020influential reference
• Ignoring non‐English‐language studies may bias ecological meta‐analyses

  2020cited by this paper
• Array programming with NumPy

  2020influential reference
• Exploration of Concerns about the Evidence-Based Guideline Approach in Conservation Management: Hints from Medical Practice

  2020cited by this paper
• mT5: A Massively Multilingual Pre-trained Text-to-Text Transformer

  2020cited by this paper
• 5分で分かる!? 有名論文ナナメ読み：Jacob Devlin et al. : BERT : Pre-training of Deep Bidirectional Transformers for Language Understanding

  2020cited by this paper
• Fast, scalable, and automated identification of articles for biodiversity and macroecological datasets

  2020cited by this paper
• An open source machine learning framework for efficient and transparent systematic reviews

  2020influential reference
• Toward systematic review automation: a practical guide to using machine learning tools in research synthesis

  2019cited by this paper
• Unsupervised Cross-lingual Representation Learning at Scale

  2019cited by this paper
• Explaining individual predictions when features are dependent: More accurate approximations to Shapley values

  2019influential reference
• Building a tool to overcome barriers in research-implementation spaces: The Conservation Evidence database

  2019cited by this paper
• Predicting the time needed for environmental systematic reviews and systematic maps

  2018cited by this paper
• Automatic differentiation in PyTorch

  2017influential reference
• Imbalanced-learn: A Python Toolbox to Tackle the Curse of Imbalanced Datasets in Machine Learning

  2016influential reference
• FastText.zip: Compressing text classification models

  2016cited by this paper
• Languages Are Still a Major Barrier to Global Science

  2016cited by this paper
• An Introduction to Information Retrieval

  2013cited by this paper
• Scikit-learn: Machine Learning in Python

  2011influential reference
• Steven Bird, Ewan Klein and Edward Loper: Natural Language Processing with Python, Analyzing Text with the Natural Language Toolkit

  2010influential reference
• Imbalanced text classification: A term weighting approach

  2009cited by this paper
• On strategies for imbalanced text classification using SVM: A comparative study

  2009cited by this paper
• Language bias in randomised controlled trials published in English and German.

  1997cited by this paper
• Developments in Automatic Text Retrieval

  1991cited by this paper

• No citing papers are available for this paper.