We propose a method to characterize and classify complex networks based on the time series generated by random walks and different node properties. The analysis of the fluctuations of the time series reveals the presence of long-range correlations, and allows to define, for each network, a set of characteristic exponents that capture its essential structural properties. By considering a large data set of real-world networks, we show that the characteristic exponents can be used to classify complex networks according to their function, and are able to discriminate social from biological and technological systems.
Characteristic exponents of complex networks
Vincenzo Nicosia, Manlio De Domenico, Vito Latora