A GOAL-ORIENTED CHATBOT BASED ON MACHINE LEARNING
T.M.T Nguyen, M.V. Shcherbakov
Nowadays chatbots are becoming very popular in many areas, such as business, banking, healthcare, study, travel tips, etc. The popularity of messaging platforms such as Telegram, Messenger, Whatsapp, and others has made chatbots not only popular but also become a trend in the future. Since the end of December 2019, the onset of the COVID-19 pandemic has brought about a major global health crisis. Therefore, it is extremely important to provide information about the epidemic to all people. Many governments and organizations have launched chatbots to inform the public about COVID-19. However, these chat rules are limited as they understand a limited set of questions entered by users. Thereby, creating a chatbot based on machine learning for coronavirus information is an urgent task. The purpose of the study is the development of a chatbot for searching for information about COVID-19 coronavirus infection. The method of designing and developing a chatbot on the RASA framework, as well as testing of the developed prototype, are described. Three chatbot models were created: the baseline model (B), the baseline model with synonyms (BS), and the baseline model with synonyms and noises (BSS). The effectiveness of three models was evaluated based on the following indicators: accuracy, precision, and F-measure. The analysis results showed that the BS and BSS models are better than the B model.
Keywords:chat bot, natural language processing, serverless, intent, entities, RASA, COVID-19.
ADAPTATION TO USER ENGAGEMENT IN AN ADAPTIVE LEARNING GAME
O.A. Shabalina, A.V. Kataev, A.A. Voronina
User engagement is one of the key indicators of the quality of interactive software (software), which is characterized by intense user interaction with the system. Training software belongs to the category of products, which, by definition, are based on interaction with the user, so the user’s involvement in the process of his interaction with the training system directly affects the quality of the system. Modern trends in the development of educational software are associated with the personification of the processes of user interaction with the system, which has led to the emergence of adaptive educational systems that can monitor user actions and adapt to their capabilities and needs. User involvement has a significant impact on learning and directly affects the result, therefore, the level of user involvement in the process of its interaction with the training system, as an indirect assessment of the user’s knowledge level, is applicable as a characteristic of the adaptation model in the development of adaptive learning systems. The results of the engagement analysis can be used to adapt the system aimed at retaining and increasing user engagement in the process of its interaction with the system, and thus improve its quality. The paper considers methods for assessing involvement and the possibility of their application to assessing the quality of educational software at different stages of its life cycle. The features of the use of online-assessment of engagement to adapt the learning process to the user in adaptive learning games are shown, related to the need to distinguish between involvement in the game and involvement in the learning process, and correlation of involvement and success in mastering knowledge in the game. Some possible combinations of assessments of the involvement and effectiveness of the user’s knowledge level in the process of interaction with the educational game and their possible interpretations are proposed.
Keywords: engagement, engagement assessment, educational software quality, learning system, engagement online-assessment, adaptive learning game.
THE FUZZY APPROACH FOR CLUSTERING BORROWERS OF MICROFINANCE ORGANIZATIONS
UDC 51-74: 004.9
The development of information technology is accompanied by a comprehensive transformation of the service sector, including microcredit. This sector of the Russian financial market shows steady growth annually. However, amid the high debt load on the Russian population, the availability of microcredit for most citizens, including online, has led to a high share of default disbursements of microloans in MFIs. Pressure from the regulator and a decrease in the income of Russians led the majority of MFIs to bankruptcy, while the remaining players in the microfinance market led to lower interest rates, and as a result, their margins decreased significantly. In this regard, MFIs have an urgent need to develop a scoring model that would be able to identify high-margin borrowers at the stage of applying for a microloan and “cut off” potentially defaulted borrowers. As part of this work, a methodology is proposed for clustering borrowers based on the fuzzy criterion “level of financial responsibility” and assessing the effectiveness of microfinancing based on the profitability of the loan portfolio depending on the proposed classification of borrowers.
Keywords:microfinance, fuzzy modeling, clustering techniques, risk management, classification of borrowers.
DECISION-MAKING ALGORITHMIZATION IN THE MANAGEMENT OF THE EDUCATIONAL SYSTEM OF DUAL TRAINING OF INFORMATION AND COMMUNICATION COMPLEXES’ PERSONNEL
K.I. Lvovich, Y.P. Preobrazhensky
The article is focused on one of the urgent problems arising in the course of active digitalization in production, social and economic systems – the increase of the number and volume of tasks within the framework of infocommunication systems. These tasks are related to the digital transformation of big data and the transfer of digital resources in the management and physical environment. The importance of staff in achieving these goals within the framework of human-machine system is significantly increased. A practice-oriented training mechanism is required to ensure high performance and error-free tasks. The formation of such a mechanism requires an optimization approach in managing the effectiveness of the personnel training system. Thesis discusses the formation of the algorithmic support of decision-making in the management of the educational system of dual training of staff information and communication systems. It is shown that the structure of the algorithmic procedure is determined by the sequence of optimization problems associated with the management in the development of the educational program on a variety of thematic modules that ensure the implementation of educational and professional standards, and forms of implementation of dual training. The peculiarity of extreme problems is their belonging to the class of discrete programming. The integrated multistage scheme of randomized search with certain conditions of transition of the iterative decision of optimization problems in sequence of reception of the optimum variant of the educational program is offered. The final management decision is determined by the evaluation, which is carried out by seeking the agreed opinion of the expert group.
Keywords:management, educational system, dual training, optimization, discrete programming, randomization, expert evaluation.
MODELING AND DEVELOPMENT OF A SYSTEM FOR IMPLEMENTING AND RECOGNIZING LATENT IMAGES
O.I. Maslova, G.V. Shagrova
If earlier the problem of authenticity concerned only printing products (securities, documents, tickets, money, etc.), today the protection of documents presented in digital form (scanned documents, photographs and other multimedia documents) is no less relevant. Therefore, the creation of systems for embedding hidden information in secured documents by various methods is an urgent task, because It will allow you to protect documents, as well as in disputed situations, confirm their authenticity or copyright. In this work, a mathematical model of the system for embedding and recognizing hidden images is developed, which differs from its analogues in the possibility of expanding the many used functions for embedding and recognition. In contrast to the well-known unified model of a system for embedding information in digital signals, a model is proposed that allows implementing the modular principle of constructing a system for recognizing images containing hidden information introduced by various methods. Also, on the basis of the constructed system model, an information system for the implementation and recognition of hidden images, consisting of three interconnected software modules, was developed. This system is a graphical shell for models developed in MATLAB. This approach allows you to integrate the system, add new ones or make changes to existing modules, make changes to the developed model.
Keywords:latent image, wavelet analysis, recognition of hidden images, decomposition by singular numbers, image embedding system.
RECOMMENDATION SYSTEM’S DEVELOPMENT FOR PLANNING TOURIST ROUTES BASED ON OPTIMIZATION MODELS
I.Y. Kotsyuba, A.E.Nazarenko
The article discusses the relevance of intelligent IT’ use in planning tourism activities for both individual and collective use. In this regard, this article is aimed at revealing a set of problems in the use of recommendation IT for generation of tourist routes based on optimization models. The process of describing information about tourist attractions and the process of generating a tourist route are presented in form of a formal model of linguistic descriptions for IT system and decision-making model. The methodology proposed by authors is based on methods of multicriteria optimization of discrete programming and methods of object-oriented programming. The article presents a functional model of developing a tourist route’ process as well as models of system and software architectures, a data model of developed software package. The presented methodological and software allows to reduce the complexity of tourist routes’ design, analyze expert opinions of various categories of experts. The specificity and results of experimental testing of development’s use for educational tourism are given.
Keywords:IT in tourism, educational tourism, multi-criteria optimization.
QUANTITATIVE EVALUATION OF THE LIVING STANDARDS OF THE RUSSIAN FEDERATION REGIONS
N.A. Shchukina, A.V. Golub
The study is devoted to the issue of assessing the living standards and quality of the Russian Federation regions in 2010-2018. An integral indicator is formed on the basis of the values of 33 socio-economic indicators and acts as an indicator of assessing the living standards. Selected for the study indicators are combined into seven groups: the income level of the population, the level of development of the consumer market, the standard of housing and quality of housing conditions, level of development and availability of health and education, demographic indicators, employment and unemployment, as well as the environment. The information base of the study consists of official statistics for 2010-2018. Based on the results of the integral indicator calculations the distribution of Russian regions by the living standards is obtained. The changes dynamics in the average Russian integral indicator indicates a decrease in the living standards of over the period under review. To conduct a comparative analysis of changes in the living standard in the Russian Federation regions for each region, the total increments of the integral indicator and its components were obtained. These values formed a feature space for identifying homogeneous groups of regions by the total increment of each of the seven indicators using cluster analysis methods. As a result of the stable classification procedure, the Russian Federation regions were divided into three homogeneous groups and 13 atypical regions were identified. Atypical regions differ in subindex increments that are not typical for the selected groups. A significant disparity in the rate of change in the living standard was revealed. This characterizes the lack of effectiveness of state planning and implementation of social programs at the regional level.
Keywords: living standards, integral indicator, ranking, classification of regions, cluster analysis.
SITUATION-ORIENTED DATABASES: THE FORMATION OF PERSONALIZED GRAPHIC DOCUMENTS FOR EDUCATIONAL DESIGN SUPPORT
V.V. Mironov, A.S. Gusarenko, G.A. Tuguzbaev
There is considered the task of creating personalized blanks for project documents in office graphics formats. Personalized blanks are understood as documents filled with specific design data in order to free the developer from routine actions during subsequent design. Two levels of graphic personalization complexity are noted: parametric and structural. The formation of personalized blanks is carried out in two stages: template development; template personalization. At the first stage, a blank template with preliminary marking of personalization points is manually developed in the environment of a graphical editor. At the second stage, software processing of the template is performed, in which personalization points are found in the template and personal data from the database is placed in them. Personalization based on situationally-oriented databases is discussed — an integrator of heterogeneous data based on an information processor controlled by a built-in highly abstract hierarchical situational model. Access to heterogeneous data is specified in the situational model in the form of virtual documents that are mapped onto heterogeneous real data. The features of mapping a virtual document to documents in VDX and FODG, as well as VSDX and ODG are considered. If in the first case, mapping to an XML file is required, then in the second to a ZIP archive, in the folders of which XML files are located. Fragments of situational models are considered that provide personalization based on: searching nodes containing identification tags in the XML document tree and replacing them with database data. Compared to the traditional approach, this gives a simpler task definition. The practical use of the results for information support of educational design in the discipline of “database” is discussed. There is a decrease in the complexity of the routine part of the project, an increase in the possibilities of creative activity in the process of educational design.
Keywords:personalized documents, situation-oriented database, hierarchical situational model, virtual document, VDX, VSDX, ODG, FODG.
SOFTWARE IMPLEMENTATION OF DIJKSTRA’S ALGORITHM IN GRAPH-STRUCTURAL MODELING OF SOCIO-ECONOMIC SYSTEMS USING METAGRAPHS
N.Y. Zhbanova, A.I. Miroshnikov
The developing of social networks, various data collection and storage systems and the accumulation of large amounts of information graph structures received another incentive for development. Since one of their main advantages is the ability to present compact and understandable models of complex systems. The graph-structural approach is especially productive in solving problems related to the analysis of various nature networks such as social networks, web graphs, traffic road networks and others. These objects are characterized by a large volume and complex structure so the question of using generalized graph structures to describe them is very relevant. A number of advantages can be distinguished as a justification: reducing the dimension of the source data without losing information, reducing the complexity of interpreting intermediate results, the ability to remodel tasks at several stages and others. The article is devoted to issues related to the development of the mathematical apparatus for constructing graphostructural models of complex systems. Generalizations of the concept of a graph such as undirected and oriented hypergraphs and metagraphs are considered. An example of a software implementation of a hierarchical system based on a metagraph is given. This approach allows to vary the degree of data detail without its modification.
Keywords:graph structures, metagraphs, socio-economic systems, graph-structural modeling, Dijkstra’s algorithm, graph condensation.
METHOD OF DETECTION OF HIDDEN DATA USING STEGANOGRAPHY METHOD KOCH-ZHAO
The article proposes an image analysis algorithm with a built-in message based on the Koch-Zhao steganographic method. The object of research is the image into which the Koch-Zhao embedding was carried out. The key idea is a comparative analysis of pairs of coefficients of a discrete cosine transform. For this, the dependence of the difference of the coefficients on the block number is constructed. Numerical differentiation of this dependence makes it possible to determine the boundaries of the embedded message. After that, the analysis of the initial dependence on the selected interval makes it possible to establish the applicable parameters of the Koch-Zhao method. Scientific novelty lies in the development of the steganographic analysis algorithm of the Koch-Zhao method, based on the analysis of discrete cosine transform coefficients, characterized in that it contains an automatic step search module that allows you to determine the embedding parameters and extract the message. Computer experiment completed. It is established that the proposed algorithm makes it possible to establish with absolute accuracy the size, content and location of the hidden message, in the case when it is embedded in a continuous sequence of blocks. The practical significance of the results lies in the fact that the developed software package allows stego analysis of images with data using the Koch-Zhao method with low filling of the stegocontainer (less than 40% of bits of the zero bit layer).
Keywords:analysis of DCT coefficients, identification of stego inserts, analysis of a stegocontainer, analysis of images with inserts, Koch-Zhao method.