There is a general sense of fragmentation when one compares European science to the American one. We set out to test the extent of fragmentation in the European network science field.
With Marco Scotti and Mariya Ivancheva we mapped the co-authorship network of European network science. We included all scientists with a European affiliation who presented a paper at the INSNA Sunbelt conferences or the NetSci annual conferences between 2005 and 2008 - 532 scholars. We looked up the top 5 most cited publications of these scholars, and included their co-authors in a dataset, that ultimately contains 3543 persons authoring 1689 publications.
The first chart shows the co-authorship network, where colored circles are authors, and white squares are publications. This chart is colored by discipline.
The general sense one gets by looking at this chart is that co-authorship is relatively fragmented - there are many disconnected islands of network components. But is this fragmentation significant, compared to some baseline expectation? One way to answer this question is to construct networks that represent a baselline expectation. We simulated scenarios when European authors are free to choose any co-author, from any country or field. The only contraint that we kept is that the number of authors, the number of publications, and the distribution of authors per publications needs to stay the same.
This next figure shows the oucomes of our simulations. The top panel of this figure shows the distribution of the number of components. In the one thousand simulations the average number of components was 139, with a range of 98 to 166. The observed co-authorship network has 240 components, a high number that is not likely to arise by chance. The bottom panel shows the relative size of the largest component to the size of the network. In a fragmented system the largest component does not gather a large fraction of the network. In our simulations the largest component on the avearage gathers 91.2% of all nodes, with a range between 89.1% and 93.4%. The observed proportion of the largest component is 18.6%, way smaller that we would expect in an "unbiased" system.
Beyond fragmentation we started to investigate predictors of academic success - the number of citations that a publication gathers.
This third figure shows a preliminary result - a relationship between component size and citation counts. It seems that the larger the component the higher the citation score. This suggests that if European network science grew more cohesive (with larger components), the scientific impact might benefit.