Dixon M1, Elstermann E1, Brown J1, Naim F1
1Curtin University, , Australia
Biography:
Melanie is the Extension Coordinator for the DataHarvest Project, in this role she collaborates with researchers in the Australian grains industry to improve their research data management skills and practices. She holds a Master of Data Science and is committed to supporting the agricultural industry by promoting best-practice research data management, ensuring that the data generated, along with the time and effort invested in it, delivers maximum value.
Abstract:
Discovering and reusing research data depends on the quality of the metadata that describes it. Drafting clear, consistent dataset titles and descriptions is subjective and effortful, requiring researchers to summarise complex technical information accurately. As generative artificial intelligence (AI) tools become increasingly accessible, there is an opportunity to support the drafting of high-quality metadata that meets domain expectations.
DataHarvest is a co-investment from the Grains Research and Development Corporation (GRDC) and Curtin University, designed to strengthen research data management capacity among partners. Through this work, the creation of descriptive metadata has been identified as a high-effort step in preparing records for the GRDC Data Catalogue. In response, the project explored how AI could support the creation of structured, accessible metadata.
A general, reusable platform-agnostic prompt was developed to assist in drafting consistent dataset titles and descriptions. The prompt was designed as a guided, conversational process that gathers key dataset details while allowing users to control what information is included. Development followed an iterative cycle of design, testing, and refinement with researchers across the grains industry, using feedback to improve usability across varied dataset types and experience levels. The prompt can be further optimised by users to suit other disciplinary needs and metadata conventions.
I will present our approach to designing and optimising the AI prompt, as well as key findings from the development process. The prompt reduces the researcher's burden of metadata creation, supports positive user experience, and improves findability consistency, leading to higher-quality metadata for the Australian grains industry.
