Existing metrics to map interdisciplinary research (IDR) have either been purely quantitative (e.g., bibliometrics and social network analysis techniques) lacking an understanding of social dynamics or have been qualitative, lacking replicability for large units regularly. This collaborative project between University Libraries and Computer Science department aims to develop a novel generalizable data-driven and human-centric methodology, combining the advantages of both qualitative and quantitative techniques to map the interdisciplinary research (IDR) within a research unit while devising metrics that could measure the evolutions (across organizational units and time) and identify gaps. This study will use the research collaborations for projects registered under the office of university interdisciplinary programs (OUIP) at NC State University in the past decade (2016-2025) as the case study to develop this methodology.
The project will leverage human-in-the-loop approach with large language models working with respective research unit experts for distilling topics of interest, IDR insights and qualitatively will work with OUIP experts and administrators to (1) identify priority topics that shed light on gaps and overlaps by comparing it to global IDR efforts, (2) identify use-cases, and (3) evaluate analysis insights and fine-tune the results. The methodology facilitates holistic and early (through awarded proposal data) analyses of the IDR profile and growth, and informs formative efforts to strengthen IDR. The project will disseminate the methodology through publications in international journals, the interdisciplinary insights (de-identified of any names) through university level reports and produce an interactive usable website for administrators at OUIP with important insights, that would be automatically updated as new partnerships and research patterns are forged. The tool could also be used by researchers in the University and to look for collaborations motivating intentional interdisciplinarity. The code of the website and the data collection routines would be available on github for other researcher entities to use.