No 1 (2023)

The issues of automatic text documents classification of the university in the electronic document management system are considered. A two-stage classification method based on machine learning and a numerical representation of documents is presented. It is proposed at the first stage of the method to reduce the collection size by screening out documents that do not belong to accepted classes (according to the probability of novelty of documents). At the second stage, the selection of documents with the highest occurrence frequencies of words characteristic of accepted classes documents is carried out (the formation of support vectors). The document is assigned a class to which most of the closest documents belong in accordance with the accepted distance metric. A set of programs for the text documents classification has been implemented, which is the basis for the information support of the university electronic document management system, and studies have been carried out confirming the effectiveness of the proposed method.

The paper considers the problem of psycho-emotional text processing, aimed at identifying the psychological characteristics of the author of the text and identifying the emotional characteristics of the text based on methods of psycholinguistics and artificial intelligence. A tool for psychoemotional analysis of texts in Russian is described as well as application of the tool to analysis of the VKontakte users’ reaction to fake messages is presented.

Algorithms for automatic comparison of the traces of the rifling lands (secondary toolmarks) are currently under active development. The calculation of a quantitative estimate based on the results of the comparison of the similarity of the traces of the rifling lands is also being investigated. Determining the optimal criterion for categorizing striae as "matching" in coincident secondary toolmarks is an actual problem. Our studies have shown that assigning traces to the "matching" class can be performed using Shannon information entropy. In this case, the Shannon entropy is used to estimate the criterion of optimal striae intersection. To carry out a model experiment, a sample of 344 pairs of obviously matching and 344 pairs of obviously non-matching secondary toolmarks was formed. On its basis, 200 validation subsamples were formed. The realizations characterized by the lowest entropy were calculated for each criterion of striae overlap in width (10%-100%, 20%-100%, ..., 100%) for all subsamples. It was shown that for secondary toolmarks on bullets fired from a Makarov pistol, this criterion is the interval of striae overlaps in width from 60% to 100%.

This paper evaluates the performance of a promising system for recording violations of traffic rules using RFID tags embedded in state license plates of vehicles. This system was tested on the roads of the Republic of Tatarstan and the Central Ring Road in the Moscow Region, and at present it is planned to introduce the system in pilot mode in several regions of the Russian Federation. To assess the probability of identifying a vehicle under various conditions, both a probabilistic model and simulation methods have been developed. It is shown that, in comparison with the "classical" identification methods currently used, this approach allows achieving higher work efficiency and minimizing possible errors in the identification of the offending vehicle.

The problem of manufacturing a certain number of high-quality products in flexible discrete industries is considered. Since a certain number of products always goes to waste, it is necessary to provide for the manufacture of an excess number of products, which will require an additional number of blanks, resources, etc. The problem is studied using a modeling methodology that allows analyzing the efficiency of automation of a given set of technological operations at the initial stage of technological preparation of flexible production.

The quality of requirements is critical to the success of complex technical systems projects. The paper presents the main procedures for quality control of requirements and the main directions of instrumental support for quality control of requirements. The shortcomings of the existing tools of instrumental support are listed. To overcome these shortcomings, it is proposed to apply machine learning algorithms. The main directions of research in the field of application of machine learning algorithms in the problems of requirements quality control are proposed. The experimental results obtained by the author, demonstrating the feasibility of the proposed approach, are presented. In some tasks, it was possible to achieve a quality of assessment comparable to that of an expert.

Effective human resource management prevents excessive costs, improves the quality of products and services, and promotes better workforce planning. The use of expert networks is aimed at improving the company's business processes based on know-how and innovative technological solutions. Improving the efficiency of management decisions in corporate expert networks is a significant task. This improving can be based on the automation of processes associated with the extraction, structuring and use of information and knowledge about innovations, projects and experts. A method that provides information support for such processes can be used to solve this problem by processing and providing only the most necessary information and possible solutions. This article discusses in detail such a method, which includes algorithms aimed at extracting innovations, their analysis and selection, organizing the process of creating projects for the implementation of selected innovations, forming teams for the joint implementation of such projects, updating the competencies of project executors. The method also includes a competence ontology used by these algorithms. The conducted experimental studies have shown an improving the efficiency of management decisions in corporate expert networks in terms of qualitative and quantitative criteria through the use of the developed method.

The paper presents the new algorithm for solving one problem from the scheduling theory. The method is based on the principle of graph coloring and allows simultaneous processing of several details in one workplace. The problems of scheduling theory are briefly analyzed and the place of the given problem is determined within the general classification of problems. The scheduling algorithm and the program on the basis of it have been developed to solve this problem for various optimality criteria. Two versions of the program have been implemented. The first one follows directly the data structures and the sequence of actions of the graph coloring method. In the second version, the structures of the linear representation of the graph are used, as well as multi-step operations are introduced, which made it possible to increase the efficiency of the scheduling algorithm. The time characteristics of the program execution on a different number of details for two versions of the program are given. The prospects for the development of the program and the scope of its application are discussed and could be rather wide, from agribusiness, such as optimizing the production of meat products, to manufacturing enterprises with a significant range of product line.