Too big to see: exploring proxies of structure in a real large-scale university-industry cooperation network

TL;DR

This paper analyzes the structure of a large-scale university-industry cooperation network using proxies, revealing complex topology features, temporal stability patterns, and parallels with other real-world networks.

Contribution

It introduces the use of statistical proxies to study large-scale network topology and compares structural features across different real-world 2-mode networks.

Findings

Firm node degrees follow a power law distribution.

Network stabilizes after a short period of high clustering.

Structural parallels with other large-scale networks.

Abstract

We investigate static and dynamic topologies of a 2-mode real world university-industry cooperation network. Due to its large size and complex structure, we choose to use statistical proxies for this goal. Among the findings, we shall call attention to the rank-size distribution of the firm node degrees with power law signature log-linear behavior. Which invokes hints of a complex network architecture, as a counterpoint to the random case. We compare furthermore the rank-size distributions of both modes with other real-world 2-mode large-scale networks and draw parallels for their causes. Moreover, we investigate structural change in the network by computing Robins-Alexander clustering coefficients in a rolling window fashion in order to capture topological change in the network temporal evolution. Findings suggest that network stability is achieved after a short transition phase of high clustering and cross-correlation.