Recommending research collaborations. The potential and limits of link prediction techniques


Raf Guns, University of Antwerp

Collaboration in research brings many advantages, including the
potential of saving costs and diffusing insights, expertise and ideas
between partners. Yet it is not always obvious which partners one should
collaborate with. If we consider the network of collaborations among
e.g. persons, institutions or countries, we can employ network-based
techniques known as link prediction to provide recommendations for
collaboration partners.
In this talk I provide an overview of the potential of link prediction
techniques for recommending research collaborations but also discuss
their limitations.

About the speaker
Raf Guns holds a PhD in Library & information science (2012, University of Antwerp) and a Master's degree in Linguistics (2001). He has worked as a research assistant and postdoc at the Library & information science department of the University of Antwerp (Belgium). In October 2015 he started working for the Centre for Research & Development Monitoring (ECOOM), a Flemish interuniversity consortium. His research interests center around scholarly communication, bibliometrics and social network analysis.

Venue: Conference room (room 0.16 & 0.17)

Date: 15 October 2015

Time: 12:30 - 13:30  CET


UNU-MERIT