The Development of the Software Tool for Automated Development of Quasi-ontology using Scientific Texts in the Russian Language

In this project is considered the construction of quasi-ontologies, it is one of the problems of formalizing texts in natural language for applications, such as the automation of design works. A software tool for creating quasi-ontologies is developed to form a network of the natural hierarchy of terms and keywords of input scientific text in Russian. To solve this problem is used a set of software tools Word2Vec from Google able to represent words as vectors and then cluster them. This approach differs from other algorithms realized earlier, because for the first time to solve a similar problem the modificated method of word representations in vector space was applied. Scientific articles and preprints in Russian processed in program were taken from the electronic library databases of the KIAM RAS and KiberLeninka. The texts are related different subject areas, but they have a clear structure meeting the requirements for scientific articles. The proposed method showed the effectiveness and the further directions of approach development are determined. The work consists of an introduction, three main chapters, including a description of the conceptual ideas, existing solutions and implementations, an explanation of the problem formulation and the it's stages separation, the proposal of its own method and its program implementation and a list of literature consisting of 69 items.