STREAM DATA CLASSIFICATION BASED ON BAYESIAN CRITERIA
L.S. Lomakina, A.N. Subbotin
The paper describes the issue of stream data classification. Stream data is described as a set of objects arriving from different sources at random moments of time. It might be a stream of data containing ocean coastal area sensors measure information and describing the parameters of the ecosystem condition, as well, it might be a stream of texts acquired from incoming emails attachments, etc. The Internet contains vast volumes of unstructured information. The lack of organization makes data inconvenient and resource-intensive to work with. Addressing to such an issue considered to be a relevant problem. Classification provides an opportunity to make it easier to work with unstructured information. The paper describes the algorithm for stream data classification based on Bayesian criteria. Text stream data model is proposed. This model allows applying natural language text classification algorithms to stream data. Naive Bayes classifier modification using tf-idf measure for evaluating the proximity of a classified document to a particular class that allows improving the classification quality is proposed. The classifier has been trained using the machine Fund of the Russian language. Software allowing text data stream extraction from the Internet and its classification using the proposed algorithm in real-time scale is proposed.
Keywords: classification, data stream, naive Bayesian classifier, Bayesian criteria.
СLASSIFICATION AND COMPARATIVE ANALYSIS OF TECHNOLOGIES OF MULTIFACTOR AUTHENTICATION IN WEB APPLICATIONS
D.S. Bogdanov, S.G. Klyuev
The relevance of the study is due to the growing pace of the introduction of multi-factor authentication mechanisms in web applications, the popularization of web technologies, as well as the lack of specific standards in the Russian Federation that describe the operation of multi-factor authentication technologies and establish requirements for web applications that use this technology. The work discusses existing authentication technologies and protocols both in a general form and in terms of authentication in web applications, considers the sequential operation of HTTP 1.0, 1.1 standard protocols, and their advantages and disadvantages are noted. Combinations of authentication factors used in the development of multi-factor authentication mechanisms were considered and correlated with existing authentication protocols. Based on the study, a classification of multifactor authentication technologies in web applications was proposed. The purpose of this study is a general analysis of authentication methods used in web applications, a comparative analysis of authentication protocols in web applications, classification of multifactor authentication technologies in order to highlight the most significant parameters for authentication systems and protocols, and then determine the rationality of using multifactor authentication in one or another web application. The materials of the work are of theoretical value for further research in this area.
Keywords: multifactor authentication, web application, two-factor authentication, classification, authentication protocols.
MODELING AND OPTIMIZATION OF RATING MANAGING THE OBJECTS OF ORGANIZATIONAL SOCIAL-AND-ECONOMICAL SYSTEMS
The article considers a class of social-and-economical systems, featuring the integration of objects into an organizational whole by the managing centre. The paper shows the possibility of the managing centre goals realisation based on the rating assessment results. The author also examines the directions of formalization of rating management mechanisms based on the construction of certain classes of mathematical and optimization models. The structure of three classes of models, characterizing the organizational system objects interaction with the control centre, is substantiated: namely, streamlining, resource support and rating status. The first class is based on the use of various modifications of the integrated assessment of indicators used for rating. The second one relies on the distribution of resource provision from the rating value. The third one builds on the formalization of the evolution in the rating status of reciprocity from changes in performance indicators of the objects functioning. The construction of mathematical models allows the researchers to move on to a formalized description of optimization problems for rating management. The characterization of the structures of optimization models corresponding to the main mechanisms of rating management such as block, and discrete programming is given. The article also shows the necessity of combining numerical methods and expert evaluation to solve these issues.
Keywords: rating management, social-and-economical organizational system, modeling, optimization.
EVALUATION OF STORAGE AND FUNCTIONING CHARACTERISTICS OF ARTIFICIAL NEURAL NETWORKS ON THE BASIS OF A NEUROCOMPUTER DEVICE
A solution is proposed for assessing the temporal and quantitative characteristics of the storage and processing of artificial neural networks based on a neurocomputer device. The most popular and used topologies of artificial neural networks are considered (single-layer and multi-layer perceptron, Hopfield networks, Hamming networks, BAM networks, Jordan networks, Elman networks, ART networks of various modifications, Grossberg star, Kohonen networks, radial basis neural networks, backward propagation networks, convolutional networks) for which analytical relationships are given to evaluate the training cycle of an artificial neural network, the amount of necessary memory and the amount of data transmitted. The difference between the proposed results is that to assess the functioning of artificial neural networks, approaches and characteristics inherent in the class of neuroprocessor devices are offered and only when implementing the calculations presented in the neural network logical basis, which allows to increase the efficiency of the task solution based on neurocomputer devices. An artificial neural network was considered using a set-theoretic approach, which allowed us to obtain analytical relationships based on the number of neuron emulation operations and connections between neurons in accordance with the topology of the neural network.
Keywords: artificial neural network, neuroprocessor, storage, evaluation.
MATHEMATICAL MODEL TO DETECT ANOMALIES USING SENSITIVITY ANALYSIS APPLYING TO NEURAL NETWORK
UDC 519.25: 004.891.3
R.V. Scheglevatych, A.S. Sysoev
The transition to the digitalization in various spheres of economic and social activity is accompanied by the emergence of large amounts of data, processing which it is necessary to identify certain dependencies and build models of processes or systems. The task to identify anomaly values in dig data sets is relevant. Existing algorithms to detect anomalies are based on different approaches and have their own advantages and disadvantages. However basic schemes of all methods are similar and use at the initial stage the separation of data in a typical for system or process and those that are not, then follow structural and parametric identification of the model, and at the final stage the trained model is used to separate the data. To improve the accuracy of algorithms, they can be modified to take into account the data structure or to combine heterogeneous mathematical models. The paper describes a combined approach to build the system for detecting anomalies based on the Isolation Forest algorithm and sequential application of a neural network classifier. To reduce the dimension of neural network input vector, the approach to Sensitivity Analysis based on applying Analysis of Finite Fluctuations to the neural network model is synthesized and described. It is presented the numerical example that shows the adequacy of the proposed approach to data analysis.
Keywords: mathematical model, anomalies, sensitivity analysis, neural-network models.
DEVELOPMENT OF A PROGRAM TO ESTIMATE THE TIME OF POSTING A MESSAGE IN THE ONLINE SOCIAL NETWORK VKONTAKTE
E.B. Gribanova, A.S. Savitsky
The relevance of the study is due to the high popularity of social networks for the transmission of information, as well as the influence of the time of posting a message on the number of views and, accordingly, the degree of its distribution. The article presents a description of the program for estimating the time of posting messages in groups of the social network Vkontakte. Algorithms of data collection and processing are presented. The description of used requests to VK API by means of standard methods is given. The developed program takes into account the activity indicators of participants and their individual characteristics (the number of friends, groups, posts on the wall, etc.), as well as the number of messages published by other members of the social network. Linear convolution was applied to obtain the integral characteristic . The program is implemented using the C# language. Microsoft Excel spreadsheet processor was used to store data about subscribers and results. An example of estimating the time of posting a message in the selected group of the social network Vkontakte is considered. The developed program can be used by administrators of social network communities to estimate the time of posting in groups of the online social network and to choose the best moment of publication.
Keywords: social networks, time of post, news feed, user activity.
CONFIGURATION OF COMMERCIAL BANK PROJECT PORTFOLIO BASED ON SIMULATION MODEL
The task of building a simulation model of configuration of a portfolio of commercial bank projects is solved. The main processes of simulation model of commercial bank project activity are considered. The logic scheme of the modeling algorithm is considered, which is the structure of the model of the system functioning process. The analysis of stochastic factors identified on the basis of consultations with specialists from various structural divisions of the commercial bank, study of specialized literature was carried out. The factors are classified according to the types of banking risks they cause. Probability distributions are selected for each of the selected accident factors. The task of calculating the total risk of a portfolio of commercial bank projects, based on a simulation model, has been formalized. The structure and relationships of submodels in the simulation model of commercial bank project activity are defined. Elements of simulation model of configuration of portfolio of commercial bank projects forming submodels using different presentation methods are described. Relationships between risk types of bank projects (input parameters) and the aggregate risk of the business bank portfolio (result indicator) are defined. Experiments were carried out with the model, based on the factor plan. The possibility of clarifying and evaluating the configuration of the portfolio of commercial bank projects based on the theory of planning experiments with the simulation model is shown.
Keywords: project portfolio, commercial bank, configuration, financial plan, simulation model, banking project.
METHOD OF DYNAMIC CONTENT AGREEMENT OF LEARNING AND GAME SCENARIOS IN ADAPTIVE LEARNING GAMES
A.V. Khairov, O.A. Shabalina, A.V. Kataev
This article discusses the models and technologies of adaptation in adaptive learning games. All known approaches to the development of adaptive learning games are ad-hoc solutions, and transferring them to other games is practically impossible. To solve this problem, a model was developed to adapt to the user’s knowledge level based on the knowledge space. The proposed method supports a non-linear learning process and allows you to build personal learning strategies, which can be used to adapt the learning process to each student. The main problem of the proposed method is the immersion of a non-linear learning scenario in a game context. The process of developing adaptive learning games that immerse the learning script in the game context based on the 3I approach requires the development of design solutions to ensure the compatibility of non-linear learning and game scripts. In the general case, more than one game action, which can be selected based on the game situation, can be associated with each learning action. This allows you to expand the game scenario and increase game attractiveness without losing the logical connections of the training scenario. These studies are related to the development of an adaptation method based on the proposed model, and its implementation in an adaptive educational game.
Keywords: learning system, learning game, adaptive learning game, adaptation model, adaptation method, immersion method, nonlinear scenario, knowledge space.
MODELING AND OPTIMIZATION OF ADAPTIVE MULTICOMPONENT SYSTEMS BASED ON ALGEBRAIC STRUCTURES
D.V. Zhevnerchuk, P.S. Kulyasov
The paper describes the issues of modeling and optimization of multicomponent systems. The classification of interfaces by types is presented and the corresponding notation is introduced for each of them. A three-dimensional structure is proposed that describes and systematizes operations on the interfaces of the components of the simulated system (integration, conjugation, and filtering), which is an unweighted directed graph that has the properties of an algebraic lattice in each of the three dimensions. For each type of operation, a partial order relation is substantiated on a set of interfaces and an algebraic lattice representation is presented with justification. The proposed structure can be used as an index. With its help, a quick search for the desired interface can be carried out, for optimization of which a depth search algorithm is proposed, which is modified taking into account the design features of the graph. The proposed algorithm has less computational complexity than the classical one, and does not have its main drawbacks – incompleteness and inoptimality when searching with depth restriction. It also shows the application of the proposed approach to the modeling of multicomponent systems based on algebraic structures using the example of a higher education educational program presented as a set of components interacting via standardized interfaces.
Keywords: multicomponent system, algebraic model, standardized interface, lattice.
COLLECTION AND PROCESSING OF EXPERT DATA IN THE METHOD OF EVALUATING THE QUALITY OF INFORMATION SYSTEMS “INSPECTOR”
I.M. Azhmuhamedov, O.M. Romanova
His article discusses the process of assessing the consistency of expert data and calculating the final value of the desired (resulting) indicators for the quality assurance methodology for the “Auditor” information systems. The purpose of the study is to solve the problem of selecting and adapting survey methods (online questioning) of experts; assessment of the consistency of the data; calculating the final value of the resulting indicators according to experts. To conduct an online survey of experts, the Delphi method was adapted. Its modification consists in the following: after conducting a round of questioning, it is necessary to determine the consistency of expert opinions on each indicator to be evaluated, and if the consistency of experts on any of the sets of assessments is not achieved, it is necessary: 1) to receive comments from experts on assessments of “problematic” ones ( inconsistent) indicators; 2) to organize a second round of the survey, while before the start of the second round, each participant receives expert comments on each of the uncoordinated issues. This stage will be repeated until the opinions of experts are agreed upon. It is proposed to evaluate the consistency of expert opinions by calculating the value of the confidence interval. To evaluate the total values of indicators, it is recommended to use the median. The materials of the article are of practical value for organizations of various fields, conducting an assessment of the quality of their own information systems independently, and for IT companies providing services for this assessment.
Keywords: quality assessment of information systems, expert judgment, Delphi method, confidence interval, median, arithmetic mean, online questioning.