Clustering of Russian-language texts by styles based on the random walk model

摘要

The work is devoted to the analysis of Russian-language texts using the random walk model. The frequencies and order of the vowel letters are compared to the migration of quasiparticles that is characterized by the diffusion coefficient and the deviation of the mean squared displacement of the quasiparticles from the Einstein relation. These parameters are determined for six subcorpora of texts, differing in functional styles. The dispersion of the parameters inside each of the subcorpora is described by the gamma distribution. The distributions for the differing subcorpora are shown to have statistically significant difference. It is found, particularly, that religious and poetic texts stand out in separate groups. On the basis of the obtained data, features of the internal structure of texts, defined by the functional style, are discussed.