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


Next issue

Publication date:
16 June 2019

Articles of journal № 3 at 2014 year.

Order result by:
Public date | Title | Authors

1. An approach to case-based synthesis of functional programs [№3 за 2014 год]
Author: N.N. Fastovets
Visitors: 4631
This article deals with an automatic functional programs synthesis problem. We consider the way to apply case-based reasoning approach to program synthesis. In order to use such framework for our problem we have to consider several subproblems: definition of a "similarity" relation over program specifications space, search of the most suitable known program for a given specification and adaptation of a chosen program to new specification. We will consider adaptation method only in this article and leave other subproblems for further work. We can consider the problem of adaptation as a problem of program correction: we have to modify given (chosen in search procedure) program, so that it will satisfy a given specification. There is a method for functional logic program correction, which have a form of conditional term-rewriting systems. So, our idea is to translate a functional program to term-rewriting system and then apply the correction method to it. In this article we will consider procedures of such translation and adaptation of the correction method for our task.

2. Modular SOM for dynamic object identification [№3 за 2014 год]
Authors: A.N. Averkin, I.S. Povidalo
Visitors: 6012
Object identification is complicated if noises are present in the source data, some of the object parameters change according to unknown laws or the exact number of the object parameters is unknown. In such cases neural network can be applied for dynamic object identification. There are a lot of different types of neural networks that can be used for dy-namic object identification. Among different neural network architectures applicable for dynamic object identification a class of neural networks based on self-organizing maps (SOM) can be noted. In this article, a number of neural networks based on self-organizing maps that can be successfully used for dynamic object identification are described. Unique SOM-based mod-ular neural networks, inspired by mammal's brain cortex studies, with vector quantized associative memory and recurrent self-organizing maps as modules are presented. The structure and algorithms of learning and operation of such SOM-based neural networks are described in details. Some experimental results and comparison with some other neural networks are giv-en.

3. Motion logic in the system «Binary knowledge model» [№3 за 2014 год]
Authors: Plesniewicz G.S., Gorkina A.A.
Visitors: 5473
There are various applications where there is a critical need to reason about moving objects. V.S. Subramani-an offered and F. Yaman (with D. Nau's participation) developed so-called logic of motion (LOM) that provides the formal language including terms of classical logic and Newtonian mechanics. LOM combines propositional logic, constraint sati s-faction and geometric reasoning. LOM has declarative syntax and semantics based on models, and formaliz es reasoning about planned movement of objects. LOM can naturally describe schemes and plans for moving objects and there is a poss i-bility to predict their further locations. LOM consideres objects as material points in space; thus, they have no structure and characteristics (attributes). However, numerous real applications with modeling of moving objects also require a representa-tion of the structure and characteristics of these objects. Also, they require the description of complex objects in terms of simpler ones. G.S. Plesniewicz offered a project of the system "Binary Model of Knowledge" (BMK) for ontological (co n-ceptual) modelling. BMK contains formal languages for structural and logical specification of objects. This paper shows how to integrate LOM into BMK system. LOM and BMK nicely complement each other in applicability and expressive power. Therefore, we receive a system LOM+BMK for ontological (conceptual) modelling of applications considering moving structured objects behavior. A wide range of applications would benefit from using the system LOM+BMR. An example of air traffic control using LOM+BMK is considered.

4. Multidimensional data visualizing cognitive technologies for decision-making intelligent support [№3 за 2014 год]
Authors: Тsaplin V.V., Gorokhov V.L., Vitkovskiy V.V.
Visitors: 8436
The article describes principles and examples of cognitive machine graphics for developing Decision Support Systems (DSS). The cognitive machine graphics phenomenon is displaying graphic representations which create spectacular images in the human operator brain. These images stimulate its descriptive impressions, closely related to the intuitive mec h-anisms of thinking. The cognitive effect is in the fact that man perceives the moving projection as three-dimensional picture characterized by multidimensional data properties in the multidimensional space. After the multidimensional data visual as-pects study there appears the possibility for a user to paint interesting separate objects or groups of objects by standard ma-chine drawing. Next user can return to the image rotation procedure to check the intuitive user’s ideas about the clusters an d the relationship in multidimensional data. It is possible to develop the cognitive machine drawing methods in combination with other information technologies. They are the packets of digital images processing and multidimensional statistical analy-sis. The proposed method was based on the idea of possibility to assemble a cognitive image as object in hyperbolic space. In special sense it is possible to say that new kind of DSS – Cognitive Decision Support Systems (CDSS) appear.

5. Construction and research of e-learning subject ontology [№3 за 2014 год]
Author: Balashova I.Yu.
Visitors: 8080
The article considers the problem of integrating and analyzing knowledge in e-learning. This problem is caused by uncertain terminology, the lack of an unambiguous interpretation and reasonable domain knowledge classification. The author proposes a solution for the problem of constructing a unified domain-based knowledge model. Exploring different solutions in ontological investigation of e-learning has shown that considered ontologies involve conceptual locality and lo-cality in using. Their fragmentary nature does not allow forming the basis of a single model of knowledge. Thus, the author shows the need to develop a domain-specific ontology that fix common for e-learning domain knowledge. This will allow providing logical consistency between separate ontologies when combining. As a result of an ontological analysis, the key e-learning ontology classes are highlighted. They establish a common vocabulary of terms to represent domain knowledge. The paper constructs and proves a taxonomy of ontology classes which represents terms hierarchy on the investment relation. The structure of classes is investigated with links to standards. There is an analysis of classes’ instances. The article indicates perspective ways of development for key classes’ instances of ontology. It also determines the set of interclass relations (object properties) which define semantic domain structure. The author suggests the ways of domain ontology practical applica-tion. The directions for further research are formulated.

6. Ellipsis recovery as a problem of automatic text processing [№3 за 2014 год]
Authors: Malkovskiy M.G., Minyaylov V.S., Starostin A.S.
Visitors: 5608
Nowadays the importance of automatic syntactic analysis of natural language texts is obvious. Phenomenon of homonymy is a feature of any natural language. It is the main reason which explains complexity of the task of automatic syn-tactic analysis: the same language signs placed into different contexts can be interpreted differently. Note that in a natural languages there is one more phenomenon besides homonymy. Its potential ambiguity is shown also well. This phenomenon is ellipsis. Despite common nature of both phenomena and the fact that homonymy is studied by computer linguistics during decades, the ellipsis is mentioned in some papers only in passing. Nowadays even the simplest ellipsis cases are a hard obsta-cle for syntactic analysis algorithms. An approach to solve the problem of automatic syntactic analysis is proposed in the paper. Correct processing of elliptic constructions is a natural element of a general scheme of syntactic analysis including three parallel functioning mechanisms (mechanisms of predictions, mechanism of mergers and estimation mechanism) is the main feature of the approach.

7. Linear solution for quadratic programming problems [№3 за 2014 год]
Author: Tatarenko S.I.
Visitors: 8210
The work describes a quadric maximum search method on a polyhedral restrictions set. It is based on solving linear equations systems when their dimension no higher than a number of objective function variables. In the beginning, a quadric matrix is worked reduced to a normal view using orthogonal transformations and scaling. Then, using variables re-placement, the initial problem is reduced to a problem of searching the minimum distance between an unconditional extremum point and a polyhedral restrictions set. Perpendiculars to different dimension polyhedron sides are constructed to measure the distance. The special order of sides search is developed to reduce the number of studied sides. The subject to re-search includes only the sides containing the top which is the closest to an unconditional extremum point and seen from this point. If there is not only one closest top, then the side containing all these tops and the sides of lower dimension having not less than two shared closest tops with the first side are investigated.

8. Anomaly detection based on Markov chain model with production rules [№3 за 2014 год]
Authors: S.M. Kovalev, A.V. Sukhanov
Visitors: 5039
The paper presents a new technique for anomaly detection in temporal data. This technique uses Markov chain model to represent a temporal profile of process behavior. The model includes the production rules that specify transition probabilities between process states considering its historical data. Such model has been trained and tested in target process. Therefore, we try to find scarce observations that we consider anomalous. These anomalous observations had the low level of our model support. The model has been tested on Coffee model that is used for benchmarking of anomaly detection tech-niques. The computational experiments showed that this technique can distinguish normal behavior from normal activities in observed process. The value of our technique is the fact that it combines advantages of stochastic and one-class techniques. In this connection such technique can be used in wide range of application areas connected to monitoring and diagnosis of emergency technological situations.

9. Features of creating a custom qualifier for situation display on the electronic map [№3 за 2014 год]
Authors: Voytsekhovsky S.V., Devyatkin A.M., Kotenok A.A., Obukhov A.V.
Visitors: 7971
Nowadays the solution of target distribution problem in the staffs of military districts is performed "manually". It does not allow react quickly to sudden situation changes and making adjustments in the target allocation process. In the context of sharp increasing of troops mobility, decrease of time spending on planning their actions is not only relevant, but also one of the most important tasks of military science. An automated complex model for target allocation of enemy destruc-tion objective was designed at the academy as a part of the research work. The model was tested during a joint strategic exe r-cises "West-2013". This software package development allowed improving the fire damage planning group work and a com-mander’s decision making efficiency. To form spatial data in a geographic information system, classifiers developed using GOST R 52439-2005 are a manda-tory requirement. Thus, the design of a software system model with an electronic map visualization system, its application technologies and testing showed the need in making a decision to use several well -known classifiers or to create a new digital classifier. It was not possible to solve all the problems for software complex model for target allocation of enemy d estruction objec-tive using just known classifiers. Furthermore, the use of not one but several classifiers of symbols can complicate a visual i-zation system operation algorithm significantly. Therefore, in order to reflect the situation and the results of solving target al-location computational problem in the program using symbols, a Rigel classifier have been created. The article presents the results of creating digital classifier to show the situation and the results of solving electronic map computational problem.

10. Simulation systems comparative analysis for information processing research in a globally distributed automated information systems [№3 за 2014 год]
Author: Bondarenko A.A.
Visitors: 12400
This article examines the problem of simulation system choice for globally distributed automated information systems (AIS) research. A special type of AIS, the so called globally distributed ones, has emerged in recent years. Such sys-tems differ from their predecessors thanks to having a central database and a complicated structure of hundreds local databas e management systems (DBMS) distributed over a significant area. The author describes the key features of the discussed sys-tems and insists on the necessity of studying information processing technologies progressing in them. The choice of the simulation method for such cases of analysis is well-grounded. The article also analyzes popular and reasonable software packages related to simulation: Simulink, GPSS World, Extend and Arena; and gives generalized descriptions of every soft-ware product. The author considers the thesis of non-obviousness while discussing the choice of software products and offers to evaluate their quality presented as a set of system properties essential for solving the occurred problem. To construct th e hierarchy of features the author uses the classification of software products possibilities given in "Simulation" by David Kelton. Indicator value summation for local or integral tree groups is used as an aggregating mechanism. At the same time quality weights of various systems are compared. It helps to choose the most optimum (with higher quality weight) system. As a result of the study, the author suggests using Simulink simulation system for analyzing globally distributed automated information systems.

| 1 | 2 | 3 | 4 | Next →