Generating pseudo-random texts based on the frequency characteristics of texts in natural languages

Abstract

The paper discusses generation of pseudo-random texts based on frequency characteristics of texts in natural languages. The follow frequency characteristics of texts and their values for the Russian and English languages are considered for generation: the distribution of unigrams and bigrams over frequency of occurrence in texts, the distribution of words over the length. Based on the considered frequency characteristics, an algorithm for generating pseudo-random texts is suggested. Texts generated according to the algorithm are studied in experiments of language recognition in texts.