Identifying talent within scholarly networks

Selecting scientific talent for positions and grants is a complex phenomenon. Where peer review traditionally played a dominant role in the past, increasingly, this is replaced by expert (but non-peer) committees who decide partly based on bibliometric indicators.
From both an evaluation and analytical perspective, this is not very satisfying. In talent selection, many criteria are deployed (implicitly or explicitly) that ultimately determine who gets a job or a grant and who doesn’t. As current research mainly focuses on traditional bilbiometric indicators, we (i) do not understand very well the process of talent identification very well, as many aspects remain invisible, and as far as the indicators used are (normatively) used, we (ii) misinform decision makers by providing them with the wrong (or one sided) indicators.
For selecting scholars for jobs, we showed elsewhere (Van den Besselaar et al 2012; Van Balen et al, 2012; Bornmann et al 2010; Van den Besselaar et al 2009) that bibliometric indicators actually do not work very well, and that we are in the need for indicators that show candidates’ scholarly independence. Other dimensions of independency should be included, such as roles in program committees of important conferences or the independence of the scientist’s academic network. The aim of this project will be the development a new more dimensional concept of independence that combine bibliometric data with other web based data (social media, information access tools (Mendeley), and conference program committees (EventSeer)). In earlier work, we explored the combination of formal and social (research blogs) scholarly media (Groth & Gurney 2010). This will lead us to a better understanding of talent selection in science networks.
In terms of computer science, we are interested in the challenging of acquiring multilevel semantic networks from these data source for use to build the indicator.
This Project builds on the previous Academy Assistants’ Project (Groth &Van Den Besselaar 2012), Where we started with the integration of heterogeneous data for science policy studies and research evaluation.

Article: Dumitrache, A., Groth, P. and Besselaar, P. van den, 2013: Identifying Research Talent Using Web-Centric Databases

Student Research Assistants:

Ravindra Harige
MSc in Artificial Intelligence
E-mail: ravindra.harige(at)

Anca Dumitrache
MSc in Artificial Intelligence
E-mail: anca.dumitrache(at)


Dr. Paul Groth
Knowledge Representation & Reasoning Group
Artificial Intelligence Section
Department of Computer Science
VU University of Amsterdam

Prof. Peter van den Besselaar
Department of Organizational Science
VU University Amsterdam