“iBiogeography: aproveitar o poder dos Big Data para a monitorização biogeográfica” iBiogeography: harnessing and measuring the power of big, unstructured data for biogeographical monitoring

Financiador

Organização

Publicações

Who is reporting non-native species and how? A cross-expert assessment of practices and drivers of non-native biodiversity reporting in species regional listing

Publication . Castro, Andry; Ribeiro, Joana; Reino, Luís; Capinha, César

Each year, hundreds of scientific works with species' geographical data are published. However, these data can be challenging to identify, collect, and integrate into analytical workflows due to differences in reporting structures, storage formats, and the omission or inconsistency of relevant information and terminology. These difficulties tend to be aggravated for non-native species, given varying attitudes toward non-native species reporting and the existence of an additional layer of invasion-related terminology. Thus, our objective is to identify the current practices and drivers of the geographical reporting of non-native species in the scientific literature. We conducted an online survey targeting authors of species regional checklists—a widely published source of biogeographical data—where we asked about reporting habits and perceptions regarding non-native taxa. The responses and the relationships between response variables and predictors were analyzed using descriptive statistics and ordinal logistic regression models. With a response rate of 22.4% (n = 113), we found that nearly half of respondents (45.5%) do not always report non-native taxa, and of those who report, many (44.7%) do not always differentiate them from native taxa. Close to half of respondents (46.4%) also view the terminology of biological invasions as an obstacle to the reporting of non-native taxa. The ways in which checklist information is provided are varied, but mainly correspond to descriptive text and embedded tables with non-native species (when given) mentioned alongside native species. Only 13.4% of respondents mention to always provide the data in automation-friendly formats or its publication in biodiversity data repositories. Data on the distribution of non-native species are essential for monitoring global biodiversity change and preventing biological invasions. Despite its importance our results show an urgent need to improve the frequency, accessibility, and consistency of publication of these data.

2023-05Artigo científico

Acesso aberto

Ver mais

Large language models overcome the challenges of unstructured text data in ecology

Publication . Castro, Andry; Pinto, João; Reino, Luís; Pipek, Pavel; Capinha, César

The vast volume of currently available unstructured text data, such as research papers, news, and technical report data, shows great potential for ecological research. However, manual processing of such data is labour-intensive, posing a significant challenge. In this study, we aimed to assess the application of three state-of-the-art prompt-based large language models (LLMs), GPT-3.5, GPT-4, and LLaMA-2-70B, to automate the identification, interpretation, extraction, and structuring of relevant ecological information from unstructured textual sources. We focused on species distribution data from two sources: news outlets and research papers. We assessed the LLMs for four key tasks: classification of documents with species distribution data, identification of regions where species are recorded, generation of geographical coordinates for these regions, and supply of results in a structured format. GPT-4 consistently outperformed the other models, demonstrating a high capacity to interpret textual data and extract relevant information, with the percentage of correct outputs often exceeding 90% (average accuracy across tasks: 87–100%). Its performance also depended on the data source type and task, with better results achieved with news reports, in the identification of regions with species reports and presentation of structured output. Its predecessor, GPT-3.5, exhibited slightly lower accuracy across all tasks and data sources (average accuracy across tasks: 81–97%), whereas LLaMA-2-70B showed the worst performance (37–73%). These results demonstrate the potential benefit of integrating prompt-based LLMs into ecological data assimilation workflows as essential tools to efficiently process large volumes of textual data.

2024Artigo científico

Acesso aberto

Ver mais

Entidade financiadora

Fundação para a Ciência e a Tecnologia

Número da atribuição

PRT/BD/152100/2021