Abstract
The overwhelming majority of words as text elements occur very rarely. Their probabilities cannot, therefore, be estimated from samples. As a consequence, special statistical means are needed which make it possible to consider the frequencies of the text elements without hypotheses about their distributions. A statistically homogeneous text model is presented, which can be used for comparing the frequency of its elements to those in real texts, and its application is illustrated by examples.
Notes
Address for correspondence: Orlov Jurij, Konstantinovič, 380102 Tbilisi 102, Gldani VI Mikrorajon, Korp 20, Kv. 124, Georgia.