I’ll be presenting some of my colleagues’ and my work for the UF CTSI Network Science program on Wednesday, October 26, at the UF Informatics Institute (UFII), with Dr. Valerio Leone Sciabolazza. More details and RSVP are here and here.
Most of the talk will be about the use of different data sources to map networks of scientific collaborations at a research university. Faculty and researchers at a university work together in many different ways, such as writing articles and grant proposals together, participating to the same awarded grants, and serving on the same graduate committees. All these types of collaboration often leave a track in some form of bibliographic or administrative records available on campus.
We extracted, visualized and analyzed longitudinal networks of scientific collaboration at the University of Florida over the last few years. Network analyses and visualizations show the evolution of collaboration patterns at UF over time. We propose a method to identify research communities in these longitudinal networks. Part of our work also consists in conducting ethnography and qualitative data collection to understand how the identified communities are reflected in the everyday perceptions and professional practices of UF faculty and researchers.
We also used network statistical models to understand what makes collaboration more likely. We are particularly interested in factors that predict interdisciplinary collaboration between different communities. We found that common institutional affiliation (department, college) matters, as well as spatial proximity. More importantly, we found that the services provided by research institutes such as the UF CTSI can increase the likelihood of interdisciplinary collaboration. This also shows that network-based methods can be used evaluate the role of specific research institutes, such as the UF CTSI, in promoting interdisciplinary team science at a research university. We discuss ways to use this knowledge to intervene on collaboration networks and create better research teams.