Abstract:

Subject ontology in the context of this article is understood as a set of key concepts related to a certain field of science, with their semantic connections, supplemented by indexes of various classification systems describing this scientific field. Subject ontologies are a necessary component of each subspace that is part of the Unified digital space of scientific knowledge (DSSK). This article presents the results of research related to the construction of subject ontologies based on the created automated system for supporting terminological dictionaries and suggests a methodology for identifying new key terms in a particular field of science. The proposed methodology is based on the use of existing classification systems in conjunction with citation databases, such as Web of Science and Scopus for English–language publications and the Russian citation index for Russian-language publications. The methodology involves dividing the scientific field into a number of sections in accordance with the selected classification system, extracting from the CSB the core of articles related to each section, and from the articles - new author's keywords, which should constitute, in combination with the corresponding sections of classification systems, the basis of the subject ontologies of this scientific field.

Abstract:

This study focuses on the application of machine learning methods to improve the quality of test items. The research includes a review of the subject area and the implementation of two enhancement methods: similar question retrieval and distractor quality assessment. The first method involves testing five transformer-based models for generating text embeddings and six clustering algorithms. The second method uses the same transformer models in combination with three classification algorithms. Experimental results demonstrated the high effectiveness of the proposed approaches in solving both tasks.

Abstract:

The results on the creation of methods for managing mathematical knowledge in the context of digital mathematical libraries are presented. The software tools developed on the basis of these methods are part of the OntoMath digital ecosystem, within which they interact. A brief description of the architecture of the OntoMath ecosystem is given, the levels of subject ontologies and external ontologies are highlighted, as well as the level of software tools and services. Semantic services are separated into a separate category. This term denotes software tools, in the functionality of which queries to subject ontologies are used to ensure the management of knowledge objects. General descriptions of developed subject ontologies are given: educational mathematical ontology OntoMath^Edu and ontology of professional mathematics OntoMath^PRO. The development of educational ontology is reflected in the direction of including educational prerequisite links between classes. Among the software tools of the digital ecosystem, search services for mathematical electronic collections, a service for semantic annotation of mathematical documents, tools for semantic marking of educational mathematical documents, as well as a system for automatically generating testing tests in mathematical educational disciplines are highlighted. As part of the OntoMath digital ecosystem, special-purpose recommender systems are being developed. The current version of the ecosystem includes a recommender system for generating a list of related articles based on the OntoMath^PRO ontology, a recommender system for appointing experts to support the scientific review process, and recommender systems for selecting subject classifiers UDC and Mathematics Subject Classification codes for mathematical documents. The results are also presented in the direction of creating a digital library metadata factory, which includes services and tools for extracting, refining, replenishing and normalizing the metadata of electronic mathematical collections. Note that the OntoMath ecosystem is being developed as the technological basis for the Lobachevskii Digital Mathematical Library.

Abstract:

The video editing process involves numerous labor-intensive operations for sorting and preparing footages, requiring significant time investment. This article describes the development of a software solution that uses machine learning technology to automate these processes.

The primary focus is on creating a system capable of classifying and sorting media files according to the screenplay text, thereby increasing the efficiency of material preparation for editing. The system includes modules for speech recognition, audio and video classification, and algorithms for determining screenplay compliance.

Testing showed that the proposed system correctly classifies media files in most cases, significantly reducing rough-cut editing time.

Abstract: In the article presents the classification of problems by differential geometry, which is based on the nature of the relationship between the elements of the problem and the relationship between the reproducing and creative activity of students in their decision. It is shown that an important source for the choice of texts of problems and methods of their solution are the works of scientists – creators of classical differential geometry. Work with the corresponding scientific text allows the student to master such an educational strategy as methodological reduction.