Vol 31, No 11 (2025)

The methods of constructing the resulting homogeneous scales for the correct application of the additive aggregation mechanism in the process of solving problems of multi-criteria evaluation and selection of objects presented in different types of measurement scales are considered. An assessment of the degree of uniformity is proposed when the maximum and minimum values of the criteria coincide. The method of transition to the resulting scale according to the criterion of equal distance of the points of division of the continuous scale from the gradations of the discrete scale is considered, and the method of constructing the resulting scale by dividing the initial continuous scales of criteria into segments, the numbers of which correspond to the gradations of the ordinal scale, is also considered. With this transition, the estimates of objects from the division segment correspond to the gradation of the ordinal scale, which leads to a loss of information distinctness of objects in the resulting scale. In this regard, this article presents an effective method of transition to the resulting scale, which ensures a one-to-one correspondence between the estimates of objects in the initial quantitative and in the resulting quantitative-ordinal scale. The presented methods are mathematically justified and demonstrated in solving the problem of multi-criteria evaluation of the effectiveness of the largest companies in the US aerospace and defense industries.

The paper considers the possibility of applying decision theory methods to solve the problem of load balancing in distributed information systems. A review and analysis of methods for evaluating and comparing multi-criteria alternatives is carried out, their advantages and disadvantages are shown in the context of the problem being solved. The most effective methods have been selected. Authors proposed a method for solving the balancing problem based on the integration of decision-making and machine learning methods. An example of using the proposed methodology is considered.

The paper is devoted to the issues of modernization of individual service solutions of the National Research Computer Network of Russia (NIKS) aimed at the tasks of effective continuous monitoring and management of the network and the services provided on its basis. The key areas and general principles of monitoring global digital infrastructures are presented and discussed, taking into account the pronounced features of national research and education networks. The main functions and software tools used for monitoring industry telecommunications networks in the field of science and education are listed, with an emphasis on freely distributed open source products. The architectural features and specific examples of original software solutions adapted for monitoring and advanced analytics of network traffic data and put into operation in the process of a step-by-step modernization of the NIKS monitoring and control system are considered.

Estimating cardinality (the number of records) plays a key role in creating effective query execution plans in a DBMS. In previous works, the authors have developed a selective Evaluation Cardinality (EVACAR) method, which has advantages over existing methods for evaluating the cardinality of query plans. The article presents the results of modification of the ENVACAR method due to double caching of database table entries. The experimental results confirming the effectiveness of the developed optimization method and its advantage over existing modern BayesCard, DeepDB and FLAT methods are presented.