ИСТИНА

The power-law or approximately power-law degree distributions are often observed in experimental networks. There exist a variety of stochastic network models, which reproduce this behavior (see, e.g., [1,2]. However, most of these models have been invented and studied for the case of undirected and unweighted networks, and it is interesting to adapt existing models for the case of directed networks. Here we consider a dataset [3] of the Florida University network of free associations in English language. This is a weighted directed dataset. The out-degree distribution of this network (in terms of the number of edges) is approximately normally distributed, while the in-degree distribution has a power law tail with exponent close to three. We generalize the embedding procedure described in [4] in order to embed this directed network into the hyperbolic space/ Also, we construct random networks with similar characteristics (number of nodes and edges, in- and out-degree) using the preferential attachment model [5] and the popularity vs similarity model [6], and compare their structural characteristics to the characteristics of the original dataset.