Dr Berenice Talamantes1, Dr Sachit Rajbhandari1, Dr Miwa Takahashi1, Dr Denis C. Bauer2,3,4,5, Katherine Tattersall1
1National Collections and Marine Infrastructure, CSIRO. , ACTON, Australia, 2Biosecurity, CSIRO, Adelaide, Australia, 3Applied BioSciences, Faculty of Science and Engineering, Macquarie University, Macquarie Park,, Australia, 4University of Sydney, School of Medical Sciences, Department of Biomedical Informatics and Digital Health, Sydney, Australia, 5The University of Adelaide, Australian Institute for Machine Learning, Adelaide, Australia
Biography:
Berenice Talamantes is a Senior Technical Officer at CSIRO. Her work focuses on environmental DNA (eDNA) data publishing, metadata standards and the development of interoperable workflows to support FAIR data practices. She is currently leading the design of a modular eDNA publishing platform for Australia, integrating workflow mapping, stakeholder engagement and emerging artificial intelligence technologies to improve metadata harmonisation and support operational eDNA data publishing.
Abstract:
Environmental DNA (eDNA) is increasingly used for biodiversity monitoring, biosecurity and environmental management. While laboratory and analytical methods are relatively well established, the preparation and storage of consistent, structured metadata requires further work to facilitate the reuse and combination of datasets across projects and repositories.
This project aims to support good data governance by improving metadata harmonisation during eDNA data publishing. To achieve this, we are developing a Human-Artificial Intelligence framework that focuses on the standardisation and transformation of source files through mapping to common standards such as FAIRe and Darwin Core. As an initial proof of concept, A large language model will be evaluated as an artificial intelligence-assisted metadata harmonisation component, supporting human review prior to deterministic transformation.
Here, we propose using artificial intelligence as an assistive layer rather than as an autonomous decision-maker. We aim to maintain transparent reporting, reduce manual effort in data harmonisation, and maintain scientific rigour and the quality of outputs.