Software & Systems

Articles of journal № 1 at 2013 year.

Order result by: Public date | Title | Authors |

1. Methode zur Lösung von Aufgaben beim Rufrouting auf der Grundlage einer neuen Bewertung der Termrelevanz [№1 за 2013 год]
Authors: Gasanova Т.О., Sergienko R.B., Semenkin E.S., Minker V.M.
Visitors: 5489
Das Call Routing, das auf der Verarbeitung von natürlicher Sprache basiert, stellt einen komplizierten und perspektivreichen Forschungsbereich auf dem Gebiet der intelligenten Verfahren zur maschinellen Verarbeitung und Interpretation von Sprache dar. Die Schwierigkeit ist bedingt durch die Komplexität bei der automatischen Interpretation von natürlicher Sprache. In diesem Artikel wird Schwerpunkt auf die Entwicklung von Algorithmen gelegt, die in der Lage sind, über die Effizienz der bestehenden Methoden in großen Datenbanken hinaus zu gehen, und die keine morphologische Analyse oder Filter der Art «Stoppwörter» erfordern. Bei der vorgeschlagenen Methode erfolgt eine Dekomposition der Klassifizierungsaufgabe, in der das Rufrouting auf zwei Stadien zurückgeführt wird: Feststellen der «residuellen» Klasse und Zuordnen der Objekte zu signifikanten Klassen. Zur «residuellen» Klasse gehören Objekte, die keiner signifikanten Klasse oder aber die nicht per se mehreren signifikanten Klassen zugeordnet werden können. Wir stellen eine neue Formel zur Bewertung der Termrelevanz bei der Bestimmung der signifikanten Klassen vor, die eine Modifizierung der Bewertung der Relevanz von fließenden Regeln in einem Fuzzy-Klassifikator ist. Wenn diese Formel nur für 300 der am meisten verbreiteten Worte für jede Klasse zur Anwendung kommt, kann eine Genauigkeit bei der Klassifizierung von 85,55 % erreicht werden.

2. Web service optimization using caching [№1 за 2013 год]
Authors: Melnikov I.I., Demidenkov K.A., Evseenko I.A.
Visitors: 6807
The paper describes a method for accelerating network data transfer through combination of caching, selective compression and encryption and a WAN-accelerator based on this method. The accelerator contains a client part and a server part. It encapsulates the data channel and caches data on the client and server side. Special hashes are used to synchronize cached data on the client and server side. If some data cached on the client side and some data cached on the server side are the same, a response will be extracted from client cache and won’t be waited from the server. Data can be quickly and effectively com-pressed by the methods of selective compression used in the accelerator if necessary. The WAN-accelerator encrypts data to improve data transfer security. WAN-accelerator server part cache asynchronous update helps to cut latency time of a response from a web-service caused by computational complexity and transaction time. Usage of the accelerator can be an effective solution for the problem of client application speed acceleration when a client application requires permanent communication with a web-service. The WAN-accelerator is an independent program module that doesn’t require integration with an existing web-service and its modification. It is transparent for a web-service and a client application. Especially it is useful for an organization that has a host system for data processing and management that contains one or more servers with data access through web-services and many clients on a great distance from them.

3. CUDA technology application for simulation of particles at parallel programming [№1 за 2013 год]
Authors: Астахова И.Ф., Коробкин Е.А.
Visitors: 9610
In article the method of simulation of the particles being in interaction, by means of technology of parallel programming of CUDA is considered. The definition оf interaction of particles is considered. The method of discrete elements is considered of the CUDA technology. In a method of discrete elements each particle l is represented as a body which has weight and center coordinates. Particles interact with each other and it is possible to observe system movement as a whole. The threadIdx variable is a three-component vector. Thus transparence is provided in work with such objects as the massif, a matrix or a volume grid is provided. Program realization is considered, the interface connecting all methods and algorithms together is considered.

4. The software implementation of a module for hydraulic calculations of two-phase gas-liquid flows [№1 за 2013 год]
Authors: Бабенко А.В., Юдовина Е.Ф., Korelshteyn L.B., Gartman T.N.
Visitors: 10120
The article is dedicated to software implementation of the basic approaches in analyzing two-phase gas-liquid flows in pipelines [1, 2]. The paper examines the internal structure of the developed software module (main classes, their hierarchy and calculation algorithm). It is shown how the software architecture of the module provides flexibility and openness of the product. The integration with well-known Russian software systems «Hydrosystem» and «Safety Valve» (developed by «NTP Truboprovod», Moscow) is also described. This integration has enriched the list of features of these programs by the following two-phase flow calculations: automatic prediction of flow pattern, the (automatical) selection of methods for calculating flow patterns, volumetric gas content, friction losses, local losses, calculating boiling and condensation parameters using a phase transitions search algorithm, calculating heat exchange with the environment (including pipes with thermal insulation), the calculation of flow distribution (for complex network pipelines).

5. Software determination of the degree of transient processes attenuation with simulink-models [№1 за 2013 год]
Author: Бильфельд Н.В.
Visitors: 15158
When comparing and analyzing transients in control systems the quality of transients is mostly used. Main indicators are a dynamic error, an integral quality, a degree of attenuation and other. Only first two of all are calculated exactly, but a degree of attenuation depends of the attenuation process’s shape. When we analyze and compare many shapes the automation of quality parameters calculating is needs. Some methods to automate the process of quality parameters calculating with simulink-models are described in the article. The first is useful when the transition process is faded or aperiodic (then the attenuation will be equal to one). Second method is useful in any case including not-faded transition processes. Third method can be used in any case too. Images shows the calculated values illustrates all the methods. Fragments of M-files with graphics of transition process are dumped.

6. A method of constructing an aggregate adaptive caching algorithm [№1 за 2013 год]
Authors: Бурилин А.В., Шефова Н.А., Гордеев Р.Н.
Visitors: 8371
The paper is focused on the problem of data caching in the client-server architecture. Today the problem of constructing caching algorithms is very relevant and a lot of implementations are proposed, but there is a field for further research. The construction of adaptive caching algorithms is especially important. In this paper we generalize adaptive algorithms by adding a control parameter in an aggregate model of the caching algorithm that is responsible for the degree of influence of aggregation algorithms.

7. Call routing problem solving method based on a new term relevance estimation [№1 за 2013 год]
Authors: Гасанова Т.О., Сергиенко Р.Б., Semenkin E.S., Минкер В.М.
Visitors: 7150
Call routing based on Natural Language Understanding remains a complex and challenging research area in machine intelligence and language understanding. This challenge is due to the difficulty in automated natural language understanding. This paper focuses on the design of algorithms which are able to outperform existing methods on large dataset and do not require morphological and stop-word filtering. The proposed approach decomposes the classification problem into two steps: detection the residual class and utterance categorization to meaningful classes. Class residual includes utterances which cannot be assigned to any useful class or which can be assigned to more than one class. We present the new formula for term relevance estimation which is a modification of fuzzy rules relevance estimation for fuzzy classifier. Using these formulae for only 300 frequent words for each class we achieve an accuracy rate of 85,55 %.

8. Designing process representation in functionally adapted form for store of the project solutions class [№1 за 2013 год]
Authors: Gorbachev I.V., Pokhilko A.F.
Visitors: 10927
The interoperability of distributed and concurrent designing is defined by the representation of CAD systems project solutions. Nowadays solution exchange is performed through standard ISO 10303 STEP format, created in the frames of CALS technology and doesn’t allow modifying solutions. This article describes project solutions extraction capability from design activity, designing object class models formation, as well as store, map and further use of information in the context of the functionally adapted representation. The functionally adapted representation concept allows solution exchange by the other method. In this case, the main thing is the fixation of solution construction procedure (consists of operation chain) in the procedure chain. On the base of procedure chain mini CAD is generated by author’s instrumental environment. Mini CAD is small software application for designing. Mini CAD functionality is determined by required operation set for designed object creation. In this case, the logic and the modify ability of project process is saved inside of selected project operation class.

9. A method for classification of objects based on a modified analytic hierarchy [№1 за 2013 год]
Authors: Гордеев Р.Н., Бурилин А.В.
Visitors: 10102
We are going to investigate the problem of the multi-criteria decision-making under conditions of undetermined information. A method of classifying and ranking objects using a modified method of analytic hierarchy process, which is based on the method of comparing objects of alternatives based on the theory of possibility and Dempster's rule, which allows to select alternatives under incomplete, inaccurate and uncertain expert's information was proposed. The proposed approach avoids the construction of a complete matrix of pairwise comparisons of alternatives required for the application of the classical method of analytic hierarchy process and most of its modifications, by adding the frame of differences as a basis for comparison of alternatives. In addition we proposed a method of ranking alternatives, based on the theory of possibilities. The proposed methods are illustrated on the example of estimation models alliances between banks and insurance companies.

10. Principles and software tools for visualization and analysis public contracts data in science and technology [№1 за 2013 год]
Authors: Горюнов И.Г., Пономарев С.А., Петров А.Н., Стрелец А.В.
Visitors: 10099
This article tells us about chosen principle for analysis accumulated information about implementation of the activities of the federal programs in science and technology in order to identify and statistically processes «coherence» logical facts of public contracts information objects. The article describes the basic functionality of the developed information system for analysis coherence of public contracts complexity-structured data. A description of bidirectional graphs using is shown for visualization and visual analysis state orders data. Used algorithms of graph theory and auxiliary software (libraries) for the construction, visualization and analysis of graph lists. Examples of the system settings screen forms, examples of visualization and analysis test data of government contracts on the graph are shown.

| 1 | 2 | 3 | 4 | Next → ►