On text author identification in network communication

The article deals with general and particular patterns of solving the identification problem of attributional linguistics. Experimental research is carried out on the material of the texts of written network communication. The problem is solved for several discursive areas: online literature, corporate e-mail, short comments on posts on the entertainment portal. These discursive spheres correlate with several functional styles of speech: fiction, official and colloquial, which makes it possible to identify the trends in textual attribution within the material of Internet communication. A holistic approach with linguistic modelling as its means, is used for the analysis of speech material. The problem of determining text authorship is solved by comparing a disputed text, the author of which, according to the conditions of the experiment, is unknown, with texts, the authors of which are known. The application of specialized software “KhoRom”, has enabled the researchers to create mathematical models of the individual styles of the compared texts authors. The contrast of the resulting models is performed by considering Pearson linear correlation coefficient, coefficient of determination, and Students t-criterion. On the basis of these models and their study, a number of conclusions about statistical patterns of an authorship study in various discursive spheres of electronic communication are made. The revealed general and particular statistical regularities are represented in score tables.