ISSN 0236-235X (P)
ISSN 2311-2735 (E)


Next issue

Publication date:
16 December 2018

Latest issue articles

Order result by:
Public date | Title | Authors

1. A corpus manager system: corpus data architecture and models [№4 за год]
Authors: D.Sh. Sulejmanov, D.R. Mukhamedshin
Visitors: 534
Modern corpus data management systems allow solving a wide range of computer linguistics problems. Such systems are often based on ready-made solutions, which causes the problems with search query execution speed (data extraction), flexibility, scalability and extensibility of the system. This article considers the architecture of the corpus manager sys-tem for the Tatar Corpus. It also proposes a corpus data model that solves a number of problems of the existing corpus manager systems. The paper describes in detail the architecture of the software part of the system, which allows quick developing of additional modules and integrating third-party applications into the system. The corpus data model presented in the article is designed to solve quick search tasks using wordforms and lemmas (direct search), as well as morphological properties (reverse search). The developed model also allows phrasal and syntactic search, thus providing a solution for tasks related to more complex data extraction. The authors performed experiments in order to find the most optimal set of data storage technologies that allows solving the required problems and has the capability to expand the system functionality in the future. The paper presents the results of experiments, including the results of testing the performance of various CAD systems and data storages. The search module of the corpus manager system of the Tatar language corpus includes the described system architecture and the corpus data model. It shows a significant increase in the execution speed of search queries in comparison with similar modules in other systems.

2. Grammar for queries for heterogeneous data storage in proactive systems [№4 за год]
Authors: M.V. Shcherbakov, Tran Van Phu, Sai Van Cuong
Visitors: 608
The problem of storage and processing of heterogeneous data (both structured and non-structured) from various data sources is an important issue when implementing proactive support systems for a life cycle operation stage of complex multi-object distributed systems. The data are heterogenous, so it is required to store data descriptions (metadata) for subsequent automatic processing. In order to solve the problem of heterogeneous data efficient storage, an architecture called data lake is used. It implements mechanisms for data batch processing and real-time data processing. The task of improving methods for effective access to heterogeneous data includes the following subtasks: development of SQL query grammar for heterogeneous data; building a parser for recognizing queries according to new grammar; development of modules for processing heterogeneous data according to a query; development of recommendations (methods) for applying the developed modules in proactive decision support systems. The proposed grammar is based on the DML extension of the SQL language, in particular the SELECT statement extension. There is the generated parser using the ANTLR 3.0 library for processing the generated queries according to new grammar. Due to generation, there are some created classes in JAVA with their objects used for parsing queries. The generated parser and processing modules for heterogeneous data have become a basis for the new software. After testing the software that implements the proposed grammar in the proactive decision support system, the authors have analyzed the time of execution of unified queries with different volumes of heterogeneous data. The main result of the grammar application is the reduction in the heterogeneous data processing time within a single query.

3. Support tools for modeling logical rules in the RVML notation [№4 за год]
Authors: Yurin A.Yu., Dorodnykh N.O., Korshunov S.A.
Visitors: 457
Nowadays there is a wide range of different means and methods of knowledge representation and processing. At the same time, a rule-based model is still one of the most common and popular formalisms of knowledge representation for the development of expert systems for various purposes. In turn, the effective application of languages for rule-based knowledge base engineering requires the use and creation of specialized toolkits that provide the visualization of this process. The use of specialized notations, which extend existing ones (UML in particular) is promising in the field of visualization of rule base engineering. The Rule Visual Modeling Language (RVML) is an example of such UML extension. The paper describes the main elements of RVML and software that supports it. In particular, it considers the webcentric Web RVML Editor. This editor supports RVML and is designed for end-users who are non-programmers. It also provides integration with external specialized software that is: the Knowledge Base Development System to access converting modules and Personal Knowledge Base Designer to test knowledge bases. The paper describes the Web RVML Editor architecture, its main functions and GUI. The paper also presents the Tiny RVML Editor that is desktop software. It is a prototype of the Web RVML Editor and it duplicates it functionally. The described software provide multi-platform support for visual engineering of logical rules in RVML notation and use the common approach and principles of GUI organization.

4. Investigation of a combined algorithm for learning three-layer neural networks of different topologies [№4 за год]
Authors: Chernyshev Yu.O., Е.N. Ostroukh, L.N. Evich, S.D. Markin, P.А. Panasenko
Visitors: 508
When learning a neural network, the weighting factors are adjusted based on minimizing a calculation error. When the objective function has a complex character and a big number of local extremums, network learning using gradient optimization methods does not often guarantee the finding of a global extremum. Nowadays, the solution of this problem for a large class of problems includes using genetic algorithms as the main method for learning backpropagation networks. The development of these algorithms has continued in the study of bioinspired algorithms and their hybrid modifications. The use of bioinspired algorithms, which are based on random search methods, allows solving the problem of bypassing local extremums and has high convergence rate. The paper considers a combined bioinspired algorithm that solves the global optimization problem when there are problems associated with learning artificial neural networks. The network structure and the number of neurons in each hidden layer are important parameters affecting the effectiveness of artificial neural networks learning. Three-layer neural networks can solve many complex problems. However, the effect of the number of neurons in each hidden layer on the convergence rate is under-explored in the general case. The paper studies a combination of the firefly algorithm and gradient descent developed by the authors for the study of three-layer neural networks of various topologies. The conducted research made it possible to identify topology from artificial neural networks. This topology makes it possible to obtain the most optimal solution for fewer steps. The analysis of the learning algorithm performance is based on the exceptional-OR (Xor) function.

5. Ontologies in the problems of building a concept domain model [№4 за год]
Authors: Vinogradov G.P., A.A. Prokhorov
Visitors: 342
The purpose of the research is the problem of communication between information systems and between an information system and an operator. The problem is relevant due to the IT intellectualization that leads to the fact that an information system acquires the features of an intellectual agent. So, there are new prospects for the digital economy, where information interaction at the network level plays an important role. The methodological basis of the research are the results from theoretical semiotics, from the use of ontology as a form of real world knowledge representation, as well as the publication of domestic and foreign scientists on the problem of communication between information systems. The research includes the communication process analysis. It is revealed that the intelligent agent considers a message as a decision result either to inform or to change the status of another agent. It is shown that communication should be considered as one of the types of control, such as behavior. The authors prove that, in relation to the theory of behavior, this approach requires including psychological aspects into the subject area of the theory of choice. First of all, these are the problems of perception, awareness, understanding the properties of the situation of choice and building assessments of the results of choice, communication, conflict, the introduction of new concepts and the relations between them of a number of others. Communication process support systems should contain a domain model that is understandable and interpreted by a user. It is shown that conceptual modeling of the subject domain based on ontologies is the most appropriate way to build it. The conceptual model consists of a conceptual structure and a concept content description. A conceptual structure is a set of concepts on which four abstraction maps are defined: generalization, type designs, association, and aggregation. A concept content description is carried out using database tables. The essential difference of the used conceptual model from others is describing an association as a usual concept. It is shown that, due to the semantic invariance of conceptual interpretation, it is possible to improve technological and operational characteristics of an information system. The proposed approach creates the basis for the development and organization of the interface of artificial entities both among themselves and with an operator based on a certain “professional language”, which is relevant to a particular subject domain. This involves considering an interface as a sign system and using methods of the theory of artificial intelligence to build it.

6. Uniform distribution of computing processes and network load by the physical and logical cluster cores in UNIX-like systems [№4 за год]
Authors: E.V. Palchevsky, A.R. Khalikov
Visitors: 409
The article is devoted to the uniform distribution of computing processes in clusters. The designed hardware-software module Distribution to distribute computing processes (including network load) by physical and logical cores of the CPU through modification of Markov chains in a computing cluster. The paper shows load dependence of physical resources on the launched tasks on a computing cluster. The use the hardware-software Distribution module is justified. The authors also calculate the distribution of the circuit to transfer computing processes between physical servers of a cluster. The load distribution system consists of three stages: the first one is the algorithm development, the second one is technical implementation, and the third one is testing of a hardware-software module. The first stage presents the features of the developed hardware-software module: start, stop, restart, network stack load testing, kernel load setting. The second stage gives the scheme of the developed hardware-software module. The third stage includes testing a hardware-software module. During the testing, it was confirmed that the load on the computing resources of the physical server decreases. The developed solution reduces the load by 11.15 units. It also allows simultaneous launching of complex and resource-intensive computing processes without disrupting computer performance.

7. Development of a service industry software infrastructure in digital economy [№4 за год]
Authors: Ivaschenko A.V., S.A. Korchivoy
Visitors: 360
This paper considers the problem of software development and performance evaluation for virtual intermediary operators in service industry at the stages of investment, as well as evaluation of the progress of the digital economy project implementation. The authors note the specific features of such solutions related to virtualization of services in digital social systems. In order to solve this problem, it is proposed to implement a model of infrastructural return of digital economy. The model is based on correlation of virtual benefits of digital economical system and its software infrastructure development expenses. Infrastructure implementing costs can consider investor marketing expenses, as well as the costs of R&D projects initiation and development considering investment risks. The model is based on a formal representation of products and services as interrelated services that require single or multiple actions of certain costs. It is proposed to present implementation of services based within the considered digital platform as a graph that connects IT infrastructure objects, service providers and providing services using the relations of infrastructural supply and implementation. In addition, there is a new effect of service emission. It arises when a subset of existing services and service accesses leads to generation of new services based on combination and intersection without using infrastructure. Analysis of the software platform efficiency is based on calculation of it’s implementation costs correlated with the income generated by the provided services. The proposed model has been used in practice as a part of the software for service industry virtual intermediary operators. It provides digital features by coordination of interaction between services providers and customers in integrated information space.

8. Building a specialist’s knowledge model in the digital economy [№4 за год]
Authors: Vinogradov G.P., N.G. Vinogradova, D.A. Shapel
Visitors: 382
The purpose of the study is to investigate the problem of forming a training standard and the construction of a higher education specialist knowledge model. These problems are considered important when managing the process of professional personnel training for the digital economy. The authors used the methods of system analysis, the theory of active systems, the theory of information management, coordinated management in organizational systems, methods and approaches of the artificial intelligence theory. The research shows that Russia's entry into the Bologna Process caused some downturns. Specialists’ training effectiveness and quality has become lower. It has become difficult to integrate higher education institutions into a new social and economic system of the country. Specialists’ training goals, including the goals at all training stages, do not conform to the goals of economic entities. At the same time, universities are the ones who are mainly responsible for training specialists who meet the requirements of the digital economy. One of the ways to overcome these downturns might be the transition of universities to the design and technological type of work organization. Then the most promising form of educational project management is the implementation of the model of information interaction between parties concerned (organizations and universities) within the framework of active self-developing network expert environments. The paper presents the structure of an active self-developing network expert environment and its functioning principles when forming specialist’s model standards and knowledge. This approach allows considering the construction of training standards and development of a specialist knowledge model in the context of corporate intellectual potential management strategies. The implementation of this approach is a multi-stage process of collecting, analyzing, evaluating and processing large amounts of information. The authors propose a system that provides searching, filtering, summarizing information, determining its value and prospects. It is established that the method of improving the quality of educational project decisions is the synthesis of network expertise technologies and the methodology of the theory of active systems. The paper shows that the effective formation of training standards and a specialists’ knowledge model is possible within the framework of self-organizing expert environments that are presented as active systems with heterogeneous agents with a counter method of communicating information with the active influence of the center. The effective implementation of this approach implies changing the goals of the higher education system from meeting the needs for training higher education personnel шт order to develop an innovative economy. This will ensure the interaction of universities and organizations in the management of their intellectual capital in the digital economy on a new legislative basis.

9. Optimal feedback control for one class of nonlinear systems via quadratic criterion [№4 за год]
Authors: A.P. Afanasiev, I.I. Emelyanova, S.M. Dzyuba, A.N. Pchelintsev
Visitors: 408
The paper considers the problem of synthesizing optimal control for non-linear systems with a separated linear part upon quadratic criterion. This problem is rather significant, since such models are used to described many models of technical, physical, biological and economical systems. The paper presents a method of synthesizing optimal feedback control. This method is based on a special method of successive approximations. It is developed from the results described in earlier papers of the authors. These papers have proved that, if the system operation time is small enough, then the suggested method converges uniformly. This establishes the existence of a solution for the initial problem in the form of a feedback control law, and provides a procedure for approximate development of this law. If the system operation time is long, as it usually happens, then it is impossible to prove the uniform convergence of the successive approximations method due to the specific nature of the method. However, numerous computational experiments have demonstrated that simple convergence of the method can be in many cases. Moreover, it can be demonstrated that, if the method converges within a random time interval, its convergence implies the existence of a solution for the initial problem. Thus, it additionally provides an approximate solution of the problem in the form of a feedback control law. Therefore, the suggested method is justified for solving problems in many real world situations. The paper presents an example of applying this method for solving the problem of stabilization of Watt governor operation modes. This example has practical significance, as stabilization of Watt governor is a classical problem of the control theory. The numerical implementation of the example applies the methods of object-oriented programming. The numerical experiment has shown that the method in question allows stabilizing Watt governor operation within a rather wide range of parameter extraction.

10. A modified algorithm of random multiple access with successive interference cancellation [№4 за год]
Author: S.A. Zhezhera
Visitors: 382
The subject of the study is the analysis of the throughput and reliability of wireless networks in multiple random access systems. The article investigates the possibility of increasing the efficiency of random multiple access with interference mitigation. In a multiple-access system, data transmission is based on the competition of stations for the overall transmission medium. Each node that does not have a debt passes a packet while risking to fall into random conflicts. The aim of the article is a new solution for the joint operation of an algorithm with an interference mitigation procedure when the base station is able to subtract a subscriber signal from a previously received signal in order to recover a residual signal to improve the reliability of wireless networks with pulsed traffic. The modified algorithm is based on the idea of searching in the set of received signals of an orthogonal subset that allows simultaneously extracting all the set of conflicting signals. The algorithm uses a full search and therefore has considerable computational complexity. Research results revealed that the process of conflict resolution is accelerating when using the procedure for canceling interference. In order to do this, subscriber signals should be stored in special additional memory cells. When the multiple access algorithm is modified into an outgoing signal from each subscriber, the service information of the previous attempts to transmit the signal is added. The exclusive right to use the procedure to cancel interference can reduce the probability of conflicts and thus increase the quality of data transmission. The paper proposes to use a modified algorithm from the ALOHA family with the possibility of canceling interference in networks using the Poisson input stream.

| 1 | 2 | 3 | Next →