A clustering coefficient structural entropy of complex networks

Abstract

The structural entropy is a quantification of the topological structural complexity of the static complex networks, which is defined based on the structural characteristics and the Shannon entropy. For instance, the ’degree structural entropy’ is based on the network’s ’first-order’ topological properties: the degree of each node. The ’betweenness structural entropy’ is based on the betweenness centrality of nodes, which is a global topological structural property of static complex networks. The two different structural entropy give two completely different views of the network’s topological structural complexity. However, a ’mesoscopic’ structural entropy is still missing in the network theory. In this work, a clustering coefficient structural entropy of complex networks is proposed to quantify the structural complexity of static networks on the mesoscopic scale. The effectivity of the proposed ’mesoscopic’ structural entropy is verified in a series of networks that grow from two different seed networks under the Barabási–Albert and Erdős–Rényi rules. We also find that the quantification of structural entropy effectively reflects the impact of structural heterogeneity on the growth rule in the early stages of seed network growth. Finally, we observe that the structural ratio of the clustering coefficient structural entropy and degree structural entropy remains stable and unchanged when network growth reaches maturity. We also note that the convergence rate of the network’s structural entropy ratio varies under different guiding rules. These findings suggest that the differences in structural entropy can serve as a novel tool for measuring the stability of complex networks and provide fresh insights into achieving a ’balanced’ state in the dynamic evolution of complex networks.