Development of a Software Tool for the Semantic Ambiguity Resolution of Natural Language Texts

In order to improve and spread the use of machine learning for natural language processing new methods of generating very accurate digital representations of words are required. Nowadays, the most popular methods for constructing such representations do not take into account the diversity of the meanings of the same word depending on the context, which leads to a decline in the quality of the models created on their basis. The purpose of this paper is to develop software that is able to transform text into vector space with the resolution of its semantic ambiguity for the Russian language. To achieve this goal, existing practices of creating a digital representation of the text, as well as methods for resolving ambiguities in the text were studied, an adaptation of the AdaGram method for working in Python was proposed, and also tested on three different tasks. The results of this work can be used for further analysis of methods for resolving the semantic ambiguity of the Russian language, as well as to improve the accuracy of the text analysis algorithms. The paper consists of 49 pages, contains 4 figures, 4 tables, 5 listings and 11 applications. Keywords: machine learning; data mining; natural language processing; semantic ambiguity resolution.