Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4968101 | Journal of Informetrics | 2017 | 12 Pages |
Abstract
The evolution of a collaboration network is to some extent steered by the network topology itself. This is the reason behind the success of network evolution models and approaches to link prediction. At the same time, some changes are due to exogenous factors (i.e., factors external to the network itself). In this paper, we explore changes in the collaboration network of African countries (2000-2014), with a focus on detecting emergent links beyond endogenous network features. Using link prediction and machine learning, we generate an 'expected' (predicted) collaboration network based on past data and compare it with the actual network that evolved in later years. The results show that the intensity of collaboration with non-African countries is higher than expected, especially for countries that are scientifically more active. To a lesser extent, we also find an increase in collaboration within Africa, which seems mostly due to the scientifically less developed countries. Emergent collaborations are mostly found in the first half of the studied period; in the second half the network structure appears to stabilize.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science Applications
Authors
Raf Guns, Lili Wang,