Software & Systems

Articles of journal № 4 at 2016 year.

Order result by: Public date | Title | Authors |

1. Models of heterogeneous agents’ behavior in a conflict and their implementation algorithms [№4 за 2016 год]
Authors: Mutovkina N.Yu., Klyushin A.Yu., Semenov N.A.
Visitors: 5163
The article discusses pair models of agents’ behavior in a conflict. Each agent has one of three possible behavioral types: deviating, compromise and coercing. A behavioral type determines an agent’s nature and specifies agent’s reaction that should be expected in response to specific actions. The authors select the need to share limited resources as the main cause of a conflict in simulation. The agents need resources to implement assigned tasks and to exist themselves. It is determined that in a conflict situation each agent tries to obtain more resources, than its opponent at the moment. The value of this commitment depends on an agent’s behavioral type. The article identifies the technique of selecting the best behavioral model. It also proposes the measures to resolve conflict between agents. The authors applied the basic concepts of fuzzy logic in simulation.

2. Methods of assessing the quality of university students’ education using a neuro-fuzzy approach [№4 за 2016 год]
Authors: A.R. Aydinyan, O.L. Tsvetkova
Visitors: 8382
Nowadays according to the Bologna process the system of education quality assessment are introduced for successful functioning of higher education institutions and enhancing their competitive in the international market. The accuracy of determining the university efficiency to a large extent depends on the correct choice of criteria to assess the factors that determine education quality. Education quality assessment is hampered by the fact that the value of this indicator depends on many factors, as well as the fact that during the pedagogical dimensions it is inevitable to work with non-numeric information. In order to achieve the objectives the authors have reviewed factors that have the greatest impact on student learning. Traditionally among these factors there are the following: the quality of interaction with a labor market during the formation of educational content, the quality of applicants and students, the quality of teaching technologies, teaching and logistical support of an educational process, the quality of teaching staff. The proposed methods of assessing the quality of students education is based on a two-level system. It is built on the basis of an adaptive neuro-fuzzy inference system (ANFIS) which is implemented in the Fuzzy Logic Toolbox package of MatLab and artificial neural networks. To evaluate each of the four groups of factors that influence the quality of a students’ learning process, there are four modules implemented using artificial neural networks. The number of inputs of each artificial neural network is caused by a number of factors included in an appropriate group. Factor group values go to the input of ANFIS, which is the second level of a quality assessment system for university students’ education. The output variable of ANFIS is a numerical evaluation of students’ learning quality. Two-level calculation simplifies data collection for artificial neural networks training and the expert assessments formation for training and formation of inputs of artificial neural networks and ANFIS. The proposed system of students education quality assessment provides a numerical estimate, which is an aggregate measure of an educational institution performance, showing learning process efficiency.

3. An information model of LibMeta semantic library [№4 за 2016 год]
Author: Ataeva O.M.
Visitors: 9217
The article considers libraries as information systems that provide the core functionality for working with data. The technology development determines the concept of both the library and its resources, which are not limited only by bibliographic records and their electronic submission now, but also bring the semantics of these resources to the front. Based on new opportunities offered by advances in technologies, a library user receives additional opportunities to work with digital library resources using descriptions of their range of interests in subject area terms based on standards with dictionaries, thesauri and ontologies. This allows organizing and describing his own collections, as well as his own resources, detailing a resource description and their area of interest by clarifying its terms. The paper considers basic requirements for such libraries and describes the developed system information model. A feature of the system is the ability to integrate data from sources integrated in the LOD cloud.

4. Semantic analysis and methods of text meaning representation in computer linguistics [№4 за 2016 год]
Author: Batura T.V.
Visitors: 24122
The paper is devoted to the problems of semantic analysis of texts. The article discusses different methods, such as dependency diagrams, semantic network, approaches based on lexical functions and thematic classes, frame, ontological and logical models of knowledge representation. At the moment, there are many methods of representing sentence meaning. Creating new methods of semantic analysis is significant in solving many problems of computational linguistics such as machine translation, automatic summarization, text classification and others. Development of new tools for semantic analysis is equally important. Despite the fact that some of the scientific and technical ideas in natural language processing are evolved, many problems of semantic analysis remain unsolved. Most of researchers came to the conclusion that the dictionary for semantic analysis has to operate meanings and, therefore, describe the properties and relationships between concepts, rather than individual words. But there is a question: how to organize and represent information in these dictionaries to search it fast and conveniently, and in addition, take into account the changes in the natural language (the disappearance of old and the emergence of new concepts). This paper attempts to answer this and other questions. The article undertakes an attempt to systematize known achievements in the field of a semantic analysis, and in any measure to find the answer to this and other questions.

5. The monitoring of information trends system’s architecture based on the free software [№4 за 2016 год]
Authors: S.A. Belyaev, А.V. Vasilev, S.A. Kudryakov
Visitors: 8460
The article describes a software system designed to identify sources of information trends in the analysis of publications on news sites, social networks and blogs. The main feature of the system is the construction of a graph of information dissemination in the Internet. The authors prove the relevance of this problem, despite the presence of ready-made solutions, which scan data in the Internet. The paper also focuses on a problem of exponential increase in the volume of information requiring processing. The article describes a model to formalize the process of analyzing information trends and marked differences from published approaches to solving this problem. The authors propose basic steps for automation solutions based on this model. Special attention is paid to the possibility and validity of using software products with open source code for solving individual subtasks. To build the system the authors offer a layered architecture that demonstrates the possibility of rational use of free software and give a sequence of operation of the system. On the basis of the architecture and the proposed model there is developed software that provides the solution to the problem of monitoring information trends. The results of testing based on several news sites. The paper proposes some approaches for further development of the solution.

6. Modeling of placing service-oriented applications in a software-defined infrastructure of the virtual data center [№4 за 2016 год]
Authors: Bolodurina I.P., D.I. Parfenov, Reshetnikov V.N.
Visitors: 9465
The information technology development in recent years leads to using distributed service-oriented business applications. Thus, research on mechanisms of managing hosted applications in a cloud environment is currently important, as well as efficient allocation of appropriate resources for their work within the existing constraints. The aim of the study is to simulate the structure and interaction mechanisms of typical cloud service-oriented applications, as well as definition of approaches to optimization of their placement in a storage system. The authors developed a generalized model of a service-oriented cloud application that includes three basic layers: an application model, a service model, and a resource model. As a resource in the model they consider some kind of object that describes the behavior and characteristics of individual elements of an infrastructure depending on the current status and settings of the virtual data center of a software-defined infrastructure. A distinctive feature of the model is that it is a universal resource that allows us to consider it from user’s point of view (for whom it is a closed system) and from the point of view of the software-defined infrastructure of a virtual data center (for which it is an open system). The novelty of the model is simultaneous describing data placement, associated applications, as well as the state of the virtual environment, taking into account the network topology. To work out the resource model in the virtual data center of software-defined infrastructure in detail the authors developed a model of a software-defined storage as a separate element. It allows defining data parameters and applications’ connections with virtual and physical nodes of a data storage.

7. Parallel calculations as a way to increase the efficiency of computing aerodynamic problem solutions [№4 за 2016 год]
Authors: S.A. Burenkov, O.Yu. Shamayeva
Visitors: 7367
The article proposes software to increase the efficiency of solving aerodynamic problems in high-performance cluster systems. The software is implemented on the basis of the proposed parallel-sequential algorithm for solving sparse systems of linear algebraic equations of high dimensionality using the method of generalized minimal residual. The article provides a theoretical estimate of the developed algorithm complexity, as well as a comparison of the results of theoretical estimates and computational experiment results. The software is implemented for a distributed memory computation model and is used to solve a model problem of an aerodynamic flow of a profile with an air flow. There are the results of a computational experiment on calculating the density and pressure of the blowing profile air flow in the high-performance cluster, which illustrate the effect of using a parallel program. The article analyzes the characteristics of model problem parallel solution acceleration, which allow determining a threshold of dimension of systems equations. Below the threshold the organization of parallel computing adversely effects on the efficiency of computing resources. The paper also studies the effect of a sparse coefficient matrix degree on parallel implementation acceleration. The given results can be applied not only on computational hydrodynamics and gas dynamics, but also energetics, biology, chemistry, pharmacology, astrophysics, and others.

8. 3D objects rendering using clustered shading [№4 за 2016 год]
Authors: Giatsintov A.M., Mamrosenko K.A.
Visitors: 8164
The article describes methods of lighting and shading of 3D objects that significantly enhance the realism of three-dimensional virtual scenes, their advantages and disadvantages. It also presents approaches for organization of light sources data in graphics card memory. Usually a large number of light sources are required in order to shade the 3D scene correctly. With forward rendering lighting is computed in the following way: influence of each light source on the resulting image is computed for each vertex in 3D scene and each pixel of the framebuffer. Therefore, visualization subsystem performance may be significantly reduced when there are a lot of light sources in a scene. In order to solve this problem we can use the methods of deferred rendering. The main idea of deferred rendering is to separate the geometry processing phase from lighting phase. An image is rendered in several phases. Scene geometry is rendered only once and information about color, normal and depth of each pixel is stored in temporary G-Buffer that is used in the following phases of rendering. But deferred rendering also has a number of significant downsides. This article provides information about some methods that resolve most of drawbacks of deferred rendering while retaining its advantages. They include tiled visualization and clustered visualization. The paper provides performance measurements of visualization subsystem with various rendering methods.

9. The method of obtaining a scan of clothing details considering deformational aptitude of a material [№4 за 2016 год]
Authors: Gorelova A.E., Kornilova N.L., Z.R. Grigoreva, Vasilev D.A., L.A. Kornilova
Visitors: 9399
The research presents an algorithm for obtaining patterns of garment (3DK) using a phased conversion of triangulation part network of three-dimensional product construction using methods of geometric surface transformation and physical modeling of material behavior. On the first stage a 3D cover transforms into a flat view by projecting network cell edges on a deployment plane. Next, we use physical modeling, which consists in giving to each node of triangulation network the properties of a material point with a mass equal to 1, and giving to each edge an elastic connection, which seeks to result its length into an original length. The process involves two time steps: a spurt to unbalance the system of material points and transition of a detail to a sustainable equilibrium state (iteration). Then the process continues until reaching the minimum of total force on a detail. It means the maximum possible approximation of geometrical characteristics 3DK and a pattern. Quality evaluation of the algorithm is implemented by constructing scans of orbicular hemisphere segments. The paper shows that the patterns obtained in the first stage of conversion are practically unsuitable for cutting details of fabric, because it has an insufficient area. As a result of physical modeling, the details area reduces to its original value, herewith the patterns perimeter is always at least 3DK. In addition, there can be seen discrepancy between the main overall dimensions of details, which indicates that there is a need to use forced molding or deformation of materials properties to obtain the garment shape. The magnitude of the deformation determines the correct choice of materials for a model. The method is implemented in CAD of tight-fitting products BustCAD (Ivanovo, Russia) and showed good practical results.

10. Assessment of influence of anti-virus software on quality of information-computing system functioning [№4 за 2016 год]
Authors: Drobotun E.B., D.V. Kozlov
Visitors: 6065
One of the most important indicators of the antivirus software effectiveness, along with the quality of information systems protection against computer viruses and malware, is resource intensity of these means. Ideally, antivirus software should not have any impact on the quality of tasks and functions assigned to a protected computing system. However, it should be noted that to build any kind of system (including anti-virus software as a protection system against computer viruses and malware) that would not have any adverse effect on a protected object is virtually impossible. Based on this fact, one of the criteria of choosing antivirus software to protect any information system should be a minimum degree of influence of selected antivirus software tools on the quality of the tasks and functions assigned to the protected system. As an index of the degree of antivirus software influence on a computing system, the article proposes to use a generalized indicator of protected computing system performance degradation. This indicator is determined based on partial indicators of performance degradation when dealing with individual tasks assigned to the protected system, taking into account the importance of each task. The article also presents the results of evaluation of the degree of seven common antivirus software influence on a general purpose computing system.

| 1 | 2 | 3 | Next → ►