Methodological differences in modeling texts’ statistical structure (on the example of “The tale of the rout of Mamai”)

Three methods of modeling statistical structure of the text are analyzed. The obtained comparative results were derived by the employment of different statistical models to the same material (“The Tale of The Rout of Mamai”). All compared models are designed to separate autosemantic words from synsemantic words of the plot. The results received during models’ testing are provided. The h-point introduced by Hirsch - Popescu is shown to be the most suitable parameter helping to separate content words from structure-class words. The h-point marks the biggest part of non-thematic words for a certain text.