Data-driven modeling of collaboration networks: a cross-domain analysis

EPJ Data Science

Table 4 Modular properties for the aggregated R&D network, the six largest sectoral R&D networks, and the six representative co-authorship networks

	Clusters	Q	\(\boldsymbol{Q^{\mathrm{rand}}}\)
Aggregated R&D network	3,561	0.679	0.570 ± 0.001
Sectoral R&D networks
Pharmaceuticals (SIC 283)	860	0.607	0.438 ± 0.002
Computer hardware (SIC 357)	783	0.623	0.502 ± 0.002
Communications equipment (SIC 366)	749	0.653	0.461 ± 0.002
Electronic components (SIC 367)	302	0.502	0.311 ± 0.002
Computer software (SIC 737)	354	0.531	0.333 ± 0.002
R&D, laboratory and testing (SIC 873)	256	0.527	0.317 ± 0.003
Co-authorship networks
Quant. mech., field theories, spec. relativity (PACS 03)	3,029	0.779	0.2344 ± 0.0004
General relativity and gravitation (PACS 04)	1,207	0.795	0.128 ± 0.016
Optics (PACS 42)	2,853	0.794	0.195 ± 0.002
Electronic transport in condensed matter (PACS 72)	2,411	0.832	0.2609 ± 0.0004
Superconductivity (PACS 74)	1,663	0.769	0.208 ± 0.003
Other applied and interdisciplin. physics (PACS 89)	966	0.920	0.395 ± 0.001

For all domains, we consider the respective cumulative networks, i.e. the networks obtained by keeping all the links at any time. For each network, we report the number of clusters detected by the Infomap algorithm, the modularity score Q of the network, and (as robustness check) the modularity score \(Q^{\mathrm{rand}}\) obtained in a set of 100 randomly generated networks with the same size and degree sequence as the network under examination.