Table 1 Characteristics of the Largest Connected Component (LCC) co-authorship network in 2006, 2010, and 2014. The values in parentheses correspond to the proportion of each quantity falling within the LCC as a fraction of the entire co-authorship network (e.g. for 2010, there were 294,181 papers forming the LCC, which is 0.48 of all Computer Science papers available in that year). The communities were detected with the Label-propagation algorithm [24]. Information about the growth of these metrics per year is given in Section S1

Number of papers

194,114 (0.43)

294,181 (0.48)

369,304 (0.52)

Number of nodes

249,797 (0.47)

407,532 (0.54)

566,835 (0.57)

Number of edges

292,336 (0.22)

1,453,217 (0.29)

1,042,623 (0.32)

Density

9.37e−06

1.75e-05

6.49e-06

Clustering coefficient

0.78

0.99

0.89

Mean degree

4.97

13.12

6.48

Mean weighted degree

5.99

14.33

9.44

Mean strength degree

1.73

1.78

1.8

Number of communities (≥3 researchers)

24,470

39,998

54,655

Number of researchers in communities (≥3 researchers)

249,797

407,532

566,835

Number of internal papers (all the authors within

86,354

128,415

189,072

the same community)