Abstract

In many ways, it is a great time to be a researcher or a research manager. On one hand we have a wealth of comprehensive data on research outcomes (publications cataloged in e.g. ORCID, Google Scholar, NCBI, etc.) as well as both past and future funding calls (e.g. EU Funding Tenders Portal). On the other hand, powerful AI and data science approaches can be used to make predictive analysis based on such data, for example by using text analysis via natural language processing, large language models, and clustering algorithms. Finally, the tools to harvest and analyze the data have been largely democratized - no supercomputers or large programming teams are needed. Instead, this can be done in-house, with existing digital resources, by empowering and modernizing research support departments.

The synergy between the available data and analysis opportunities is already enabling new ways to find collaborators, identify relevant funding calls, understand research trends, and inform university policy. In our talk we will highlight ways to combine AI and data to design and advance institutional and research agendas, both by presenting the best practices in general and by providing concrete examples. Of course, AI is not a magic bullet, and we will also discuss the appropriate, responsible, and transparent ways to use it, especially on already available, open data. This includes GDPR considerations, which are significantly alleviated through the use of data already open and available online.

To illustrate our points we will showcase Connect by YERUN, a multi-university collaboration and funding platform we have developed and deployed. It was built for and is used by the 23 universities of the Young European Research Universities Network (YERUN). Over the past year we built a database with over 35,000 researchers and close to 1 million publications, which allows us to automatically provide AI-driven recommendations for possible collaborators, relevant funding, as well as a synthesis of trends in past research for individual universities and the overall network.

Our main takeaway messages focus equally on the needed digital tools, skills and resources, as well as the organizational and soft-skill that must accompany them. In terms of data sources and data tools, where we recommend on striking a balance between using tried and tested packages (e.g using Django for our databases) while constantly keeping an eye out for new tools, testing and adopting them quickly (e.g. new, open large language models, beyond ChatGPT). Beyond digital aspects, we recommend fully embracing a bottom up, co-design approach, with users at the center of our platform. We extensively discussed with and consulted both researchers and research managers about what kind of platform would best fit their needs. Running multiple, iterative design workshops and creating efficient lines of communication between our team and the platform users was instrumental in maintaining interests and driving the early adoption of our tool. We continue to build on the support of motivated researchers and managers who act as best ambassadors for any digital platform.

The overall goal of this talk is to provide a succulent roadmap for colleagues who wish to benefit from available AI tools to open data in research support, without overwhelming them with technical details. We strongly believe that underlying principles and approaches we have used in our Connect by Yerun platform can return control to universities and researchers themselves, helping them improve research outcomes and access to funding. Our experience can be both replicated and scaled for other institutions and we hope this talk is the first step towards broader use of AI and open data in research management.