Abstract:

This article discusses the application of vector models for the preliminary analysis of students' free-form answers. Vector representations of words and documents were obtained using word2vec, doc2vec, and BERT models. The similarity between the answer given by the student and the correct answer was determined using the cosine measure. It was found that vector models allow identifying obviously incorrect answers with sufficient accuracy. For answers that are close in wording, an additional verification step is proposed. Using word2vec, binary classification of answers to certain questions was performed, and accuracy, precision, recall and F1-measure estimates were given.

Abstract:

Cloud platforms and services have become an important factor in the explosive development of the digital economy in the last decade. The ability to quickly scale the service, coupled with a reduction in investment costs at the start of projects within the framework of the Iaas, PaaS, SaaS approaches, gave positive results and formed the basis of new business models and corporate solutions.

 

In this article, we discuss the reasons for the importance of multicloud and explore approaches to integrating cloud services in a multicloud architecture. The article also proposes a way to solve the problem of cloud migration - developing a system for migrating cloud resources between cloud services.

Abstract:

The Central Sikhote-Alin Fault (CSAF) is one of the largest tectonic structures in the continental part of the south of the Russian Far East. Its modern geodynamic activity is still very poorly understood, especially by remote sensing methods. The available geological estimates of movements along the CSAR reach several mm/year, however, the available scattered modern geodetic data allow us to state that no displacements exceeding a few mm/year have been detected in the central part of the fault.

In the present work, based on the data of GNSS observations carried out in the first two decades of the 21st century, the first quantitative estimates of the secular (plate) movements of the Earth's crust in the vicinity of the village of Kievka (Primorsky Krai), located at the southern end of the CSAR, are obtained. Also, in the study area, numerical modeling of the effect of geoenvironment disturbances (co- and postseismic displacements) initiated by the catastrophic Tohoku earthquake on March 11, 2011, Mw 9.1, was evaluated and performed. The results obtained are in good agreement with the available models of the earthquake source and models of postseismic viscoelastic relaxation of the geoenvironment.

Abstract:

The COVID-19 pandemic makes the problem of monitoring and analyzing the movement of people indoors more urgent in order to timely identify those who have been in contact with the sick and prevent further spread of the infection.

The article proposes one of the ways to solve this problem - the development of a system for determining and saving the history of the location of people inside the premises. The article also discusses methods, parameters and technologies that can be used to solve the problem of indoor localization.

Abstract:

This paper considers the questions about the possibility of using data about the students presented in electronic form to build their digital portraits.  A set of characteristics necessary for its construction is proposed, a data model is designated.

Implemented tools for collecting data about students from social networks and other Internet resources. Algorithms for constructing a digital portrait are proposed. The application of machine learning algorithms for these tasks is illustrated. Examples of the use of digital portraits in education are given.

Abstract:

In the context of ever-increasing informatization, automation and digitalization of business, new schemes of unfair actions by both legal entities and individuals are emerging. In this regard, there is an acute problem of quick, effective and high-quality identification of information about a potential or current counterparty, which will allow you to quickly make the right management decisions.

The article describes one of the ways to solve this problem - the development of a system of information monitoring of counterparties, which will allow you to quickly identify and analyze information about their activities.

Abstract:

Classification of documents with the assignment of classifier codes is a traditional way of systematizing and searching for documents on a specific topic. The Universal Decimal Classification (UDC) underlies the systematization of knowledge presented in libraries, databases and other information repositories. In Russia, UDC is an obligatory attribute of all book production and information on natural and technical sciences. The choice of classification codes is associated with the analysis of the structure of the classifier tree and is traditionally decided by the author of a scientific article. This article proposes a solution for automating the assigning the UDC classification code for a mathematical article based on a special resource – the OntoMath^PRO ontology for professional mathematics, developed at Kazan Federal University. An approach to solving the problem is to create "code maps" for each classifying code in the UDC tree in the field of mathematics. Under the "code map" is meant a weighted set of all extracted, with the help of OntoMath^PRO ontology, mathematical named entities from the collection of articles with a given UDC code. The creation of "code maps" is based on the hypothesis that the choice of the UDC code is determined by a certain set of classifying features that can be represented by classes from the OntoMath^PRO ontology. The proposed hypothesis was tested and confirmed in the paper. The hypothesis was tested on a collection of mathematical articles An approach to solving the problem is to create "code maps" for each classifying code in the UDC tree in the field of mathematics. Under the "code map" is meant a weighted set of all extracted, with the help of OntoMath^PRO ontology, mathematical named entities from the collection of articles with a given UDC code. The creation of "code maps" is based on the hypothesis that the choice of the UDC code is determined by a certain set of classifying features that can be represented by classes from the OntoMath^PRO ontology. The proposed hypothesis was tested and confirmed in the paper.  The hypothesis was tested on a collection of mathematical articles published during 1999-2009 in the "Izvestiya VUZov. Mathematics" journal. 

Abstract:

Model robustness to minor deviations in the distribution of input data is an important criterion in many tasks. Neural networks show high accuracy on training samples, but the quality on test samples can be dropped dramatically due to different data distributions, a situation that is exacerbated at the subgroup level within each category. In this article we show how the robustness of the model at the subgroup level can be significantly improved with the help of the domain adaptation approach to image embeddings. We have found that application of a competitive approach to embeddings limitation gives a significant increase of accuracy metrics in a complex subgroup in comparison with the previous models. The method was tested on two independent datasets, the accuracy in a complex subgroup on the Waterbirds dataset is 90.3 {y : waterbirds;a : landbackground}, on the CelebA dataset is 92.22 {y : blondhair;a : male}.

Abstract:

The article is devoted to the creation of an effective solution for user segmentation. The article presents an analysis of existing user segmentation services, an analysis of approaches to user segmentation (ABCDx segmentation, demographic segmentation, segmentation based on a user journey map), an analysis of clustering algorithms (K-means, Mini-Batch K-means, DBSCAN, Agglomerative Clustering, Spectral Clustering). The study of these areas is aimed at creating a “flexible” segmentation solution that adapts to each user sample. Dispersion analysis (ANOVA test), analysis of clustering metrics is also used to assess the quality of user segmentation. With the help of these areas, an effective solution for user segmentation has been developed using advanced analytics and machine learning technology.

Abstract:

Today, users of mobile applications in different areas leave a huge amount of digital footprint. The main types of digital footprints are text, photos, videos, audio, and current location. To assist the teacher in horizontal learning, a mobile application that collects all of the above types of digital footprint was developed as well as web application that analyzes it.

Abstract:

The article is devoted to automation of the software design stage. In the course of the study, the reasons for the high importance of this stage and the relevance of its automation were analyzed. The main stages of this stage were also considered and the existing systems that allow automating each of them were considered. In addition, an own solution was proposed within the framework of the problem of class structure refactoring based on the combinatorial optimization method. A solution method has been developed to improve the quality of the class hierarchy and tested on a real model.

Abstract: The article presents significant changes in the field of education, taking place in connection with the active and systematic application of digital information technologies. The main directions of development of informatization of domestic education in conditions of realization of opportunities of digital information technologies are justified and described. The high-level characteristics of the main components of traditional pedagogical science (including didactics) and pedagogical science in conditions of informatization of education of the period of digital technologies have been revealed. The definition of didactics of digital information technology period is presented.

Abstract:

The growth in the number of IT products with machine-learning features is increasing the relevance of automating machine-learning processes. The use of MLOps techniques is aimed at providing training and efficient deployment of applications in a production environment by automating side infrastructure issues that are not directly related to model development.

In this paper, we review the components, principles, and approaches of MLOps and analyze existing platforms and solutions for building machine learning pipelines. In addition, we propose an approach to build a machine learning pipeline based on basic DevOps tools and open-source libraries.