Bulletin of National Technical University "KhPI". Series: System Analysis, Control and Information Technologies

TRUSTED RFID EVENT MODELING FOR AUDIT, SECURITY, AND PROVENANCE IN PATIENT APPOINTMENT WORKFLOWS

2026-05-14T17:26:34+03:00

This paper is devoted to the development and investigation of an approach to auditing, security, and provenance of RFID-based patient interactions within appointment-centered workflows in medical information systems. The paper analyzes the limitations of traditional RFID implementations in healthcare, which are typically focused on simple identifier retrieval and do not ensure contextual integrity, traceability, or verifiability of clinical actions, leading to risks related to inconsistent medical records and reduced reliability of appointment outcomes. The proposed solution treats RFID readings as trusted context-aware events embedded into the lifecycle of a patient appointment and forming a verifiable chain of clinical interactions. A formal model of a trusted RFID event is introduced, incorporating actor, temporal, spatial, and clinical context parameters, enabling its use as an atomic unit of audit and provenance. The approach establishes a relationship between RFID event chains and appointment results, where the outcome of a patient visit is derived from a sequence of validated and contextually consistent events rather than solely from declarative records. To ensure interoperability and standardized audit mechanisms, the proposed model is aligned with HL7 FHIR resources, including AuditEvent and Provenance, enabling representation of both event-level actions and their origins within a unified framework. A risk-based approach to RFID infrastructure security is incorporated, allowing differentiation of protection mechanisms depending on the criticality of clinical interactions. The client–server architecture of the medical information system is extended with event-driven server-side processing of RFID interactions, ensuring validation, authorization, and consistency of clinical workflows. The results demonstrate that the proposed approach improves auditability, traceability, and reliability of RFID-based patient appointment management compared to traditional identification-centric solutions, making it suitable for deployment in real healthcare environments requiring high levels of trust and accountability.

MODELING THE INFLUENCE OF CORPORATE KNOWLEDGE AND ORGANIZATIONAL CONTEXT OF AN IT COMPANY ON LOCAL QUALITY CRITERIA OF THE SOFTWARE DEVELOPMENT PROCESS

2026-05-15T16:42:06+03:00

The organizational context of an IT company has been parameterized by scale, business model, process maturity level, and resource constraints, and its corporate knowledge has been structurally decomposed into four classes: technological, process, empirical, and organizational, with defined inclusion criteria (formalizability, separation from the individual carrier, reusability). The local quality criteria – structuredness, manageability, predictability, measurability, adaptability, and optimizability – have been defined as evaluation criteria for an individual software development process, aligned with the nine process attributes of the SPICE model. A formalized three-level hierarchical model has been developed using the Analytic Hierarchy Process that establishes quantitative relationships among the quality of an individual software development process, its local criteria, and classes of corporate knowledge with regard to the organizational context of a specific IT company. A computational experiment has been conducted for the TEC.2 process «Stakeholder needs and requirements definition» in the context of a medium-sized outsourcing company, which revealed the dominance of empirical knowledge (43.4% of the global weight) and substantiated its prioritization. Sensitivity analysis confirmed the correct response of the ranking under contextual changes. The scientific novelty consists in establishing a quantitative dependence between classes of corporate knowledge and the quality of an individual software development process through local quality criteria with a linkage to the SPICE model attributes. The practical significance of the approach lies in replacing intuitive IT management decisions with a scientifically grounded mechanism for resource allocation toward the development of specific classes of corporate knowledge for the targeted improvement of the quality of specific software development processes. Further research will address resource limitations and the aggregation of individual process scores into the integral quality of the software development process.

COUNTERFACTUAL MODEL OF A MEDICAL BUSINESS PROCESS

2026-05-15T16:49:00+03:00

The subject of the research is medical business processes. The aim of the work is to develop a counterfactual model of a medical business process in order to enable the selection of alternative decisions when providing primary medical care, subsequent outpatient treatment, as well as clinical pathways and rehabilitation. Achieving this aim is oriented towards ensuring the flexibility of medical business processes under resource, financial, and regulatory constraints. To achieve the aim, the following tasks are addressed: to develop the formal structure of a counterfactual model of a medical business process as an extension of a situational model of a generalized medical business process, including causal dependencies and a system of temporal rules that describe admissible event sequences and time intervals between them; to perform an experimental validation of a system of clinical constraints that defines the space of possible counterfactual scenarios. A counterfactual model of a medical business process is proposed which, in contrast to existing approaches, combines the situational structure of a medical business process, a causal graph of reasons for performing process actions, a system of constraints that determine the feasibility of process trajectories in accordance with clinical guidelines, as well as temporal rules that define the sequence of medical business process events in time. Experimental validation of the model, taking into account the system of constraints, the causal graph, and the temporal rules, has confirmed a reduction in the share of clinically unacceptable scenarios compared to the traditional approach to constructing counterfactual explanations. Further development of the proposed counterfactual model of a medical business process is associated with the design of an information technology for decision support in medical business processes, aimed at constructing individualized treatment pathways that take into account patient needs and the constraints of the healthcare system.

AUTOMATED OBJECT SEGMENTATION AND BACKGROUND REMOVAL IN MULTIMEDIA CONTENT PROCESSING SYSTEMS

2026-05-15T16:56:43+03:00

In today's rapidly developing digital technologies, image processing occupies a key place in the fields of computer vision and graphics. One of the most popular tasks is the automatic separation of foreground objects from the background, which is of critical importance for web design, e-commerce, the advertising industry, and augmented reality systems. Traditional manual editing methods are laborious and dependent on user experience, especially when working with complex textures or fuzzy object boundaries. The development of deep learning methods make it possible to automate these processes, providing high accuracy and speed of processing in real time. The aim of the work is to analyze existing algorithms for automatic background removal and develop an effective methodology for object segmentation using modern computer vision models to improve the quality of multimedia content. The work uses a comprehensive approach based on the use of the hybrid neural network architecture BEN2. The key feature of the method is the implementation of an innovative Confidence Guided Matting pipeline, which implements a two-stage Bayesian approach: first, coarse segmentation (BEN Base) is performed to create a "draft" mask, and then targeted refinement of boundaries in areas of low confidence of the model (BEN Refiner). As part of the research, a software application based on the Streamlit framework was developed, which provides automated inference of the BEN2 model. The system supports loading images in popular graphic formats, batch data processing, and visualization of results using an interactive slider. The high efficiency of the algorithm was experimentally confirmed: for complex objects, the architecture provides accurate matting of hair and small details, minimizing artifacts such as "ragged edges". The processing workflow was optimized by using local caching of models and supporting acceleration on GPU/CPU. The proposed approach based on the BEN2 architecture and boundary refinement mechanisms demonstrated higher accuracy compared to classical one-stage segmentation methods. The developed system is scalable and suitable for integration into real multimedia processing information systems, which allows significantly reducing the time for preparing graphic content and increasing its professional level.

DESIGN OF AN INFORMATION SYSTEM FOR RASTER IMAGE PROCESSING

2026-05-15T17:10:41+03:00

This paper addresses the problem of designing an information system for automated processing of raster images. The need to improve processing efficiency under conditions of increasing volumes of visual data and stricter quality requirements is substantiated. The main image processing operations are identified, including image import, color correction, noise and defect removal, scaling, and cropping. The limitations of existing software tools, which are mainly oriented toward manual or semi-automatic editing, are described. An approach to the development of the information system based on functional modeling is proposed. The structuring of the image processing workflow is performed using the IDEF0 notation, which enables the formalization of relationships between input data, control parameters, mechanisms, and outputs. A context diagram is constructed and decomposed, resulting in the identification of key subprocesses. Information flows between subprocesses and transition conditions between processing stages are defined. An algorithm describing the system operation is developed in the form of pseudocode, covering the full image processing cycle. The algorithm includes file format and integrity verification, metadata analysis, parameter normalization, iterative quality adjustment, defect detection and correction, as well as scaling and cropping. An early rejection mechanism for invalid data is implemented to reduce computational costs. An adaptive correction loop is proposed to achieve target quality metrics without a fixed number of iterations. Operation logging is incorporated to enable monitoring and traceability of processing results. The practical significance of the proposed system lies in its applicability for automated raster image processing across various domains. Its implementation improves processing efficiency, ensures consistency of quality parameters, and reduces the need for manual intervention. The limitations related to the iterative nature of the algorithm and restricted format support are identified. Directions for future research include the integration of machine learning methods and the extension of system functionality.

RESPONSIBILITY-WEIGHTED AGGREGATION OF QUALITY CRITERIA IN MULTI-LAYER IOT SOFTWARE ARCHITECTURES

2026-05-15T17:39:16+03:00

Additive quality models are widely used in architectural evaluation because they are transparent, computationally simple, and suitable for integration into ranking and optimization procedures. However, in multi-layer Internet-of-Things systems, directly summing the Edge, Fog, and Cloud contributions introduces a structural bias: the resulting score depends not only on criterion fulfillment but also on the number of layers in which the criterion is realized. This leads to inter-layer double-counting, destroys the unified interpretation scale of criteria, and complicates cross-criterion and cross-scenario comparison. To address this problem, the paper introduces the layer-responsibility matrix W, which distributes the total responsibility for each quality criterion across architectural layers. A corrected aggregation formula is derived as a weighted sum of normalized layer-level contributions under row normalization of W. The paper also provides a lightweight elicitation procedure that allows architects to instantiate W from scenario characteristics, control logic, dominant risks, and computational placement. The basic properties of the proposed formalism are established, including non-negativity, boundedness, invariance with respect to the number of layers, scenario adaptivity, and interpretability. A numerical example demonstrates how the proposed mechanism eliminates inflated criterion values while preserving the linearity of aggregation. A decision-level numerical example further shows that responsibility-weighted aggregation can reverse the ranking of candidate portfolios and thereby change the architectural decision outcome. The approach is further illustrated through two case studies, infrastructure monitoring and control, and bionic prosthesis software, showing that the same aggregation rule remains valid across domains, whereas responsibility distributions vary according to domain logic. The results justify treating W as an independent component of formal decision support for IoT software architecture.

INFORMATION TECHNOLOGY FOR CONSTRUCTING EXPLANATIONS USING TEMPORALLY ORDERED INPUT DATA OF AN INTELLIGENT SYSTEM

2026-05-16T18:21:04+03:00

The subject of the article is the process of generating explanations for the decisions of intelligent systems whose input data are temporally ordered sequences of events with time delays. The aim of the work is to develop an approach to constructing explanations for the decisions of intelligent systems with temporally ordered input data that takes into account time delays in the input event sequences. To achieve this aim, the following tasks are addressed: to develop a method for processing temporal delays that includes a delay estimation component based on the cross‑correlation function and a time‑encoding component; to develop an information technology for constructing explanations for the decisions of an intelligent system based on the temporal order of input data, which integrates the developed method into a single pipeline for explanation construction and verification; to carry out experimental evaluation of the method and the information technology. A method for processing temporal delays in causal dependencies between nodes of a dynamic graph in a temporal graph neural network is proposed, which differs from known approaches by combining components of correlation‑based estimation of optimal temporal shift, phase‑shifted time encoding, and adaptive fusion of the obtained representations, thereby enabling the incorporation of causal dependencies into explanations through the estimation of time delays. An information technology for constructing explanations for the decisions of an intelligent system based on the temporal order of input data is proposed, which includes the stages of adaptive construction of temporal event graphs, building a temporal graph neural network with temporal delay processing, generation and subsequent verification of explanations based on temporal algebra, thus providing the formation of explanations that take into account changes in the order of the intelligent system’s input events. The experimental evaluation has confirmed that the temporal delay processing method adapts to deterministic, stochastic, and cyclic delays.

INFORMATION TECHNOLOGY FOR OPTIMAL SERVICE PLACEMENT PREDICTION IN A MULTICLOUD ENVIRONMENT USING MACHINE LEARNING

2026-05-16T18:30:16+03:00

The relevance of the work is due to the need to improve the efficiency of service distribution management in multi-cloud infrastructures, where optimal service placement directly affects latency, performance, reliability, and rational use of resources. The object of the study is the process of placing cloud services in a multi-provider environment. The subject of the study includes machine learning methods and algorithms that are used to predict optimal decisions for placing cloud services in a multi-provider environment based on measured performance indicators. The purpose of the study is to develop and evaluate models for predicting optimal placement of cloud services in a multi-provider environment using historical data on latency, response time, and load balancing efficiency. The work uses an open dataset, the Multi-Cloud Service Composition Dataset, which contains characteristics of services from AWS, Azure, Google Cloud, and IBM providers. Six machine learning algorithms implemented using the Python programming language and the Scikit-learn library were used for prediction. The obtained results showed that models based on Gradient Boosting and Naive Bayes provide the highest consistency of the metrics Accuracy, Precision, Recall and F1-score, reaching values of about 0.97–0.98, which confirms their suitability for the tasks of optimizing the placement of cloud services in a multi-cloud environment. Other developed models demonstrated lower stability of results, which limits their application in real conditions. The conclusions substantiate the possibility of using machine learning methods and algorithms to build adaptive load management systems in multi-cloud environments, and also identify prospects for expanding the proposed information technology by including additional parameters, such as energy consumption, computing cost and fault tolerance.

CONTROL FLOW GRAPH RECOVERY FOR DYNAMICALLY LOADED CODE VIA SYMBOLIC LIBRARY RESOLUTION

2026-05-16T18:42:05+03:00

Control Flow Graphs are one of the main data sources for software analysis that use dynamic and static software analysis methods. Protected software and modern malware increasingly depend on dynamic code loading techniques to evade static analysis. Usage of runtime dynamic linking mechanisms introduces unresolved indirect calls that stop static Control Flow Graph recovery. This serves to hide dynamic library that can be used for prevention of security analysis. To address this limitation, an analysis technique is proposed that combines symbolic execution with speculative library preloading to recover Control Flow Graphs from binaries by using dynamic loading. The methodology uses custom software hooks that intercept dynamic loading operations during symbolic execution and perform actual library loading into the analysis state. The module is based on a two-level architecture that stores interception functions and instruction tracking at the same time, all within a symbolic execution environment. To avoid executing potentially malicious code that dynamic instrumentation tools require, the analysis was conducted entirely through symbolic execution, making it safe for malware analysis. For evaluation a batch of 16 synthetic benchmarks was used, employing various obfuscation techniques including encrypted library names, network-triggered loading, environment-derived paths, multi-stage decryption chains, fileless execution and manual executable and linkable format parsing. The experiments results show that module recovers on average 29.8 % additional Control Flow Graph nodes and 26.5 % additional edges compared to static analysis alone, achieves 100 % precision and 100 % recall in library detection, with all discoveries validated through Frida-based dynamic instrumentation.

METHOD FOR VERIFYING THE BALANCE OF MENTAL MODELS OF AN INTELLIGENT SYSTEM’S DECISION

2026-05-16T18:49:58+03:00

The subject of the study is the process of verifying mental models of an intelligent system’s decision. The aim of the work is to develop an approach for assessing the balance of explanations of intelligent systems’ decisions with respect to their negative and positive aspects. In accordance with this aim, the following main tasks are addressed: to develop an approach to assessing the balance of explanations of intelligent systems’ decisions based on the proportional representation of negative and positive characteristics in an explanation; to develop a general method for verifying the balance of mental models of an intelligent system’s decision that takes into account the structural and weighted coverage of the set of negative aspects by the decision’s mental models; to carry out an experimental evaluation of the proposed method using a set of user reviews of the operation of a recommender system that contain information about mental models of the proposed decisions. An approach to assessing the balance of explanations of intelligent systems’ decisions that accounts for negative aspects is proposed. The approach involves constructing a reference weighted set of essential negative aspects of a decision, extracting negative elements of the explanation, computing indicators of structural and weighted coverage, and assessing the proportionality of the presentation of negative information. This approach provides a quantitative estimate of the extent to which an explanation reflects the limitations and potential negative consequences of applying a decision in relation to its positive properties. A method for verifying the balance of mental models of an intelligent system’s decision is proposed. The method includes the stages of constructing a reference set of negative aspects of a decision based on the analysis of user reviews, extracting the negative component of the decision’s mental models, computing indicators of structural and weighted coverage, assessing the proportionality and relevance of the presentation of negative aspects, and forming an integral measure of the balance of mental models. The method enables refinement of a mental model taking into account shortcomings of the practical application of a decision by the user that arise due to incompleteness of the models. Experimental evaluation of the method based on a set of user reviews has shown that using reviews as a source of information about users’ mental models makes it possible to construct a reference set of negative aspects of an intelligent system’s decision that reflects usage problems and risks important to users.

COMPARATIVE STUDY OF TRANSFORMER-BASED AND INTELLIGENT DOCUMENT ANALYSIS METHODS FOR AUTOMATED EXTRACTION OF MEDICAL DATA FROM PDF DOCUMENTS

2026-05-20T17:10:03+03:00

This paper presents a study on automated processing of medical laboratory reports in PDF format, with a focus on text recognition and structured information extraction. The research investigates the effectiveness of different approaches to optical character recognition (OCR), including classical methods and transformer-based models, as well as techniques for extracting key medical data from unstructured and semi-structured text. A comparative experimental analysis was conducted using medical documents with different structural characteristics, including tabular and text-based formats. The study evaluates the performance of OCR methods and extraction pipelines using a set of quantitative metrics, including Character Error Rate (CER), Word Error Rate (WER), Exact Match (EM), Precision, Recall, and F1-score. The obtained results demonstrate that OCR accuracy alone does not guarantee high-quality structured data extraction, as recognition errors significantly affect downstream processing and reduce the reliability of extracted information. Special attention is given to layout-aware approaches that utilize the structural properties of PDF documents. The proposed method based on direct text extraction using pdfplumber shows superior performance by preserving spatial relationships between document elements and eliminating the need for OCR in documents with an embedded text layer. This approach ensures higher stability and accuracy when processing structured medical data. The findings highlight that the main challenge in processing medical documents lies in the extraction stage rather than in text recognition. The study demonstrates the importance of integrating layout-aware and intelligent extraction methods for improving the reliability, robustness, and scalability of automated data processing systems. The proposed approach can be used as a foundation for developing medical information systems and decision support tools aimed at efficient and accurate clinical data management.

RESEARCH ON METHODS FOR ASSESSING THE QUALITY OF WEBSITE BACKGROUND DESIGN FOR USERS WITH COLOR VISION IMPAIRMENTS

2026-05-20T17:21:49+03:00

The purpose of the work is to study methods for assessing the quality of background design of websites for users with color blindness. Information about types of color blindness is provided, existing methods for assessing the quality of colors for web design are analyzed. Examples of existing sites that have functions for assessing the quality of web resources are provided. Existing methods for assessing the quality of sites are identified and existing web resources are analyzed. A developed comprehensive model for assessing the quality of background design of websites is presented, based on the creation of algorithms for finding colors on images and sites, assessing their quality, and building algorithms for simulating types of color blindness. Examples of the work of the implemented system for assessing the quality of background design of websites for users with color blindness, its capabilities, and structure are described. The object of the study is the process of determining the effectiveness of background design of websites in terms of the perception of content by users with color blindness. The subject of the study is methods for assessing the quality of website background design, including the selection of color schemes, contrast of text and background, and adaptation of information display for users with color blindness. The novelty of the work is the creation of a model that combines the assessment of the quality of website design, determining their colors and providing recommendations for improving the color scheme of the site. The practical value lies in the fact that an information system has been developed - a site that provides recommendations for the optimal selection of color schemes and website design for users with color vision impairments, which allows to increase the accessibility and convenience of information perception.

EVALUATING THE GENERALIZATION ABILITY OF AI-GENERATED TEXT DETECTORS TO UNSEEN GENERATORS

2026-05-20T17:34:06+03:00

Many AI-generated text detectors demonstrate high performance on datasets constructed within typical evaluation protocols. In particular, classical models based on stylometric features, such as text length, punctuation patterns, and aggregated formality indicators, can effectively capture statistical regularities of machine generation. However, their performance decreases substantially when texts produced by previously unseen generators are encountered. Under such conditions, feature distributions shift, which leads to a decline in classification quality, primarily due to an increase in false negative errors. This paper investigates the generalization ability of detection models under conditions involving an unseen generator. The study compares classical stylometric models and transformer-based approaches using the LOGO (Leave-One-Generator-Out) evaluation protocol. The task is formulated as binary text classification across two domains, Reddit and Wikipedia, and involves three generators, namely ChatGPT, Davinci, and Dolly. The classical models include Random Forest and Gradient Boosting, whereas the transformer-based approaches are represented by DistilBERT and RoBERTa. Model performance is evaluated using Accuracy, Precision, Recall, F1, and Macro-F1, with the final results averaged across multiple random initializations. The results show that transformer-based models demonstrate a higher ability to generalize to texts produced by unseen generators. In contrast, stylometric approaches exhibit a substantial degradation in performance, particularly depending on the domain and text length. Error analysis indicates that the main factor behind this decline is the increase in false negative errors. An additional analysis of feature importance shows that classical models rely heavily on surface-level textual characteristics, which do not ensure stable generalization across different generators. Therefore, the findings highlight the importance of evaluating AI-generated text detectors under the LOGO protocol to ensure robust performance in the presence of new language models.

MODELS AND METHODS FOR INTELLIGENT MANAGEMENT OF PERSONALIZED LEARNING TRAJECTORIES BASED ON AN INTEGRATED IRT-FORGETTING FRAMEWORK

2026-05-20T17:41:36+03:00

Despite the widespread adoption of modern learning management systems, most of them primarily focus on providing access to content and assessing outcomes, while overlooking the formalized modeling of individual learning trajectories and the dynamics of knowledge forgetting. This paper presents a comprehensive set of models and methods for adaptive learning, developed within the FAHRAI platform (Framework for Adaptive Hierarchical Review and Instruction). The following key components are proposed: (1) an integrated model that multiplicatively combines the two-parameter Item Response Theory model (IRT 2PL) with a memory retention function, enabling improved calibration compared to classical IRT; (2) a dynamic memory stability model incorporating a composite response quality indicator and a context-dependent multiplier; (3) a hierarchical task selection method based on a multi-level system of strategies and context-dependent weights of a composite prioritization score; (4) an approach based on the Wilson confidence interval for statistically reliable estimation of mastery level, significantly reducing the rate of false-positive decisions compared to naive accuracy; (5) a composite readiness metric integrating IRT parameters, memory retention, and the statistical reliability of the estimation. For each model, a formal description, theoretical justification, and numerical examples are provided. The proposed set of models and methods constitutes a unified theoretical framework for the development of adaptive learning systems, enabling improved accuracy in predicting learning outcomes, greater efficiency in repetition scheduling, and enhanced reliability in assessing individual learner progress.

DOMAIN-SPECIFIC LANGUAGE FOR INTELLIGENT TESTING OF MICROSERVICE SOFTWARE SYSTEMS BASED ON MOCK-OBJECT TECHNOLOGY

2026-05-20T18:03:30+03:00

Testing service-oriented and microservice-based systems is challenging due to strong dependencies on external services and distributed environments. Mock-based testing is widely used to address this issue; however, existing solutions rely on low-level configuration mechanisms that increase complexity and reduce maintainability. This paper proposes a domain-specific language (DSL) for intelligent specification and generation of mock services. The approach enables high-level, domain-oriented description of mock behavior and supports transformation into executable configurations for existing mocking platforms. The proposed solution aims to improve readability, reduce configuration effort, and enhance testing efficiency in distributed software systems. The paper reviews existing tools for mock-based testing of service-oriented applications and provides a comparative analysis based on ease of use, configuration flexibility, and the complexity of supported test scenarios. The analysis shows that existing solutions consistently trade off between expressiveness and accessibility, a limitation that the proposed DSL aims to address. To evaluate the proposed approach, a comparative experiment was conducted across four test scenarios of varying complexity. DSL-based configurations were compared against equivalent configurations defined without the DSL, using three metrics: specification size (SLOC), maximum nested block depth, and maintainability index. The results show that the DSL reduces cyclomatic complexity. The composite quality score of DSL-based configurations exceeds the baseline by 38 % on average. These findings confirm that the proposed DSL simplifies the creation of mock services and makes distributed system testing more accessible to a wider range of project participants.

FUZZY LOGIC IN DORMITORY ACCOMMODATION RECOMMENDATION SYSTEMS: ACCOUNTING FOR UNCERTAINTY AND LINGUISTIC STUDENT PREFERENCES

2026-05-14T17:07:25+03:00

The problem of student dormitory room allocation is considered as a multi-criteria decision-making task under uncertainty. The relevance of the study is driven by the need to improve the quality of accommodation processes through intelligent recommendation approaches that account for individual student preferences, rather than relying on simple filtering mechanisms. The aim of the paper is to develop a recommendation model that evaluates the suitability of dormitory rooms based on both quantitative characteristics and subjective user preferences. To achieve this goal, a fuzzy logic approach is applied to model the linguistic nature of user requirements and to handle uncertainty in decision-making. The proposed model calculates partial compatibility scores for multiple criteria, including price, distance, comfort, noise level, and social compatibility, and aggregates them into an overall suitability measure for each student–room pair. The effectiveness of the approach is evaluated by comparing it with the Simple Additive Weighting method, used as a baseline. The results show that the fuzzy logic-based approach provides more differentiated and flexible recommendations, better representing diverse student preferences and improving the quality of ranking alternatives. The analysis of score distributions shows that the fuzzy logic approach yields a smoother, more continuous range of compatibility values, whereas the additive weighting method tends to produce clustered scores, reducing the ability to distinguish between alternatives. The proposed method can be used as a decision-support tool in dormitory management systems to enhance allocation efficiency and user satisfaction.

AN INTELLIGENT SYSTEM FOR DISH-LEVEL DIET PLANNING BASED ON AN OPTIMIZATION MODEL

2026-05-14T16:38:35+03:00

The paper addresses the problem of developing intelligent systems for personalized nutrition planning. Modern research in this field demonstrates a transition from classical formal diet models to hybrid architectures that combine two methodological paradigms: knowledge-driven and data-driven approaches. However, there is some methodological gap between them. Knowledge-driven models provide mathematical rigor and guarantee the satisfaction of nutritional and resource constraints, but they are usually limited in adaptability and personalization. In contrast, data-driven approaches, including modern generative models, demonstrate high flexibility and the ability to incorporate behavioral data, yet they do not provide formal guarantees of optimality and constraint satisfaction. This contradiction motivates the development of an integrated intelligent nutrition planning system that combines the advantages of both approaches. The objective of this study is to develop an intelligent dish-level nutrition planning system whose core is a formalized multicriteria diet optimization model. Unlike the classical diet problem, where optimization is performed over individual food products, the proposed approach models nutrition at the level of complete dishes, which improves the practical feasibility, interpretability, and usability of the resulting dietary plans. The mathematical model is formulated as a multicriteria optimization problem in which the decision variables represent the number of dish portions, while constraints reflect nutritional, energetic, logical, and temporal requirements. The proposed model is implemented within a multilayer system architecture consisting of a data layer, an optimization core, an intelligent decision-support layer, and a user interaction layer. The optimization core ensures mathematical correctness and computes optimal solutions, whereas the intelligent layer provides adaptation, personalization, and interpretation of results. The model is further extended to a dynamic form using a rolling planning horizon, allowing the diet plan to be updated as new data and user preferences become available. Computational experiments have demonstrated that changes in criterion weights lead to transitions between several stable optimal meal structures, reflecting the discrete nature of the considered multicriteria optimization problem.

VERIFICATION OF EMPIRICAL MODELS OF SIMPLEX METHOD COMPLEXITY USING THE MODIFIED GMDH

2026-05-14T16:54:01+03:00

This paper presents the results of constructing an empirical dependence of the number of arithmetic operations of the standard simplex method as a function of the problem dimensionality defined in canonical form. To address this problem, the principle of model self-organization based on a modified combinatorial algorithm of the Group Method of Data Handling is applied. The modification of Group Method of Data Handling consists in forming a residual representation that is assumed to contain the sought empirical complexity function, as well as in applying a preliminary partitioning algorithm for the basis functions, which are additively included in the residual representation, into two non-overlapping classes: the class of main dominant components and the class of refining residual components. This approach significantly reduces the combinatorial search procedure. Based on theoretical complexity estimates available in the literature, a redundant set of basis functions was constructed. Five fundamental models were considered: Borgwardt’s polynomial estimate, the Adler – Megiddo quadratic bound, a basic polynomial form, the smoothed analysis model of Spielman – Teng, and a general mixed model previously proposed by the authors. To expand the search space, a logarithmic component was added to the basis functions. Thus, the residual representation includes six basis components, along with all their squares and pairwise products. The optimal structure was selected using a symmetric regularity criterion by evaluating more than 134 million alternative models, the number of which is uniquely determined by the cardinality of the set of refining residual components. To ensure the correct application of the least squares method for strongly nonlinear regressors of different scales (whose absolute values differ by several orders of magnitude), the necessity of scaling input data by their standard deviation, with optional centering, was experimentally justified. The efficiency of this approach was confirmed through a specially designed experiment with a known ideal solution, including a constant term (bias) at the order of 100000 and normally distributed noise with an amplitude of 1 % of the mean value of the ideal regression on the values of the input variables of the experiment. The results of simulation modeling on a dataset of 13690 randomly generated linear programming problems demonstrate the clear superiority of the mixed polynomial-logarithmic structure within the extended class of candidate functions.