How open is research at your institution? And how could you measure that without months of manual analysis?

This poster explores a new idea being developed at the University of Amsterdam: using Large Language Models (LLMs) to automatically detect Open Science practices across research outputs.
Open Science encourages researchers to share data, software, and research results so that science becomes more transparent, reproducible, and reusable. However, tracking whether these practices are actually being followed is surprisingly difficult. Evaluations are often time-consuming, expensive, and rely on manual review.
Recent advances in LLMs create new opportunities. By analysing publications, datasets, software records, and other research outputs, LLMs may be able to identify signals of Open Science practices at scale.

At this poster, we will tell you about this approach and what it might mean for universities, research support staff, and faculties themselves. The goal is to develop a practical toolkit that faculties, research support teams, and of course, central university departments could use directly to monitor Open Science practices within their own units.

You will learn:
• How LLMs could help identify practices such as open data sharing, software availability, preregistration, and other Open Science activities.
• How these insights could help faculties and institutions understand trends in Open Science adoption.
• How a toolkit like this could support research assessment, recognition and rewards, and Open Science policy development.
Stop by to exchange ideas and explore how tools like this could help institutions better understand and support Open Science.