LLM-Based Tool for Standardising Climate Research Metadata

This technical report presents a proposal for developing a tool that utilizes Large Language Models (LLMs) to standardize metadata across climate research repositories. The objective is to enhance data interoperability and facilitate interdisciplinary studies by addressing inconsistencies in metadata, such as varying parameters and definitions. The proposed tool aims to automate the extraction of relevant metadata from datasets, thereby improving the usability of climate data. It will also support the creation of a unified metadata schema aligned with the FAIR principles, which advocate for Findability, Accessibility, Interoperability, and Reusability of data. The report outlines the challenges faced in managing large datasets and the potential of LLMs to resolve ambiguities in natural language descriptions. The implementation of this tool is expected to significantly enhance the integration of datasets from major repositories, ultimately aiding in comparative analyses and predictive modeling related to climate change and biodiversity loss. Future developments include scaling the tool for broader applications and creating a user-friendly interface for metadata mapping.