Applıcatıon of Paragraphs Vectors Model for Semantıc Text Analysıs Irina Gruzdo 1[0000-0002-4399-2367] , Iryna Kyrychenko 1[0000-0002-7686-6439] , Glib Tereshchenko 1[0000-0001-8731-2135] , Olga Cherednichenko 2[0000-0002-9391-5220] 1 Kharkiv National University of Radioelectronics, Kharkiv, Ukraine {irina.gruzdo,iryna.kyrychenko, hlib.tereshchenko}@nure.ua 2 National Technical University "KhPI", Kharkiv, Ukraine olha.cherednichenko@gmail.com Abstract. The paper examined a model of paragraph vectors, as well as its methods of distributed memory and distributed bag of words. The peculiarity of this model lies in the definition of the objective functions of individual sentences and their representation in the form of some local vectors, on the basis of which a global vector is constructed, which determines the semantic component of the text as a whole. Various aspects of the application of distributed memory and distributed bag of words methods were considered, as well as the sets of algorithms of the underlying distributed memory and distributed bag of words methods, which allow obtaining distributed vectors of text parts to solve the problem of determining similar articles, where the search will be carried out key words, annotations, and articles of various sizes. It was experimentally established that Doc2Vec and its Bag-of-Words method, the most complete, allows you to determine borrowing and analogues depending on the structural elements of the text, in accordance with the review and the task. Also Bag-of- Words allows the user to make an exact picture of the lexical meaning of a word and its semantic relations in language and texts. Keywords: Text Meaning Definition, Semantic Analysis, Latent-Semantic Analysis, Experiment, Textual Information, Model, Semantic Analysis Library, Text Analysis, Text Fragment. 1 Introduction At the present stage of development of information technologies, both worldwide and in Ukraine, the tasks related to the processing of textual information for solving a num- ber of tasks such as plagiarism detection, text recognition, highlighting the structural blocks of text, analysis and issuance of recommendations, etc. [1, 2, 3]. Among all these tasks, one of the essential problems, which has been solved for more than 60 years and is the “cornerstone”, is the problem of semantic analysis of the text [1, 4, 5]. In [9–15], approaches to checking semantic correctness are shown. During the analysis of the pri- mary sources of the first works devoted to semantic analysis, a tendency was observed Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).