Eastern-European Journal of Enterprise Technologies

Development of a TRIE-BERT pipeline for automatic spacing and low resource language classification in Batak Toba and Angkola scriptio continua texts

Muhammad Anggia Muchtar — 2026-04-30

Batak Toba and Batak Angkola texts written in scriptio continua form without spaces are the object of this study. The work solves a low-resource variety classification problem where the two varieties are similar and missing word boundaries introduce segmentation noise. A hybrid TRIE-BERT pipeline was developed in which trie automation performs deterministic spacing, the restored spacing is fixed, and the spaced text becomes a stable input interface for a Bidirectional Encoder Representations from Transformers (BERT) classifier. Experiments used a Batak lexicon of 19,070 word entries and 8,000 sentences, 4,000 per variety, evaluated under four data schemes from 1,000 to 8,000 sentences and five epoch settings from 5 to 50 with an 80:20 split. After lexicon recalibration of about 70 sentences, spacing reached 98 percent accuracy. The best setting at 8,000 sentences and 50 epochs achieved 0.85 test accuracy with 0.343 training loss, 0.85 ROC AUC, and 0.85 F1-score, exceeding a long short-term memory recurrent neural network baseline (LSTM-RNN) at 0.80 accuracy, 0.397 loss, 0.803 ROC AUC, and 0.80 F1-score. Class-wise evaluation yielded precision 0.81 and recall 0.92 for Toba and precision 0.90 and recall 0.79 for Angkola, explaining averaged precision 0.86 and recall 0.85. The improvement is associated with the combined use of deterministic trie-based boundary recovery and contextual BERT classification, where spacing is fixed before classification to reduce token ambiguity and stabilize the input structure. The results support Batak text processing pipelines that require automatic spacing and variety detection under limited labels, provided lexicon coverage is maintained and spelling variation is controlled

Development of multi-agent generative pipelines framework for learning plan generation with deterministic constraint verification

Mohammad Fadly Syahputra — 2026-04-30

Large language models (LLMs) are increasingly used to generate structured learning plans aligned with outcome-based education (OBE). The object of the study is a multi-agent workflow for generating a structured OBE learning-plan package with a final deterministic verification stage. The problem addressed is the low reliability of LLM-generated outputs, which frequently violate schema rules, numeric constraints, and cross-artifact consistency requirements. To solve this problem, a multi-agent generative pipeline is proposed, decomposing the task into six specialized agents followed by deterministic constraint verification applied to the final artifact bundle. Structural reliability is measured using completeness and compliance, while cross-artifact coherence is evaluated through redundancy, spacing, phase progression, and assessment fit. The evaluation involves 12 courses with 10 repeated runs per course (120 runs per variant) across four different LLMs to assess cross-model robustness. The results show that the multi-agent pipeline achieves completeness of 0.9682–1.00 and compliance of 0.9376–0.9698, significantly outperforming the single-agent configuration (completeness 0.5926–0.6580; compliance 0.4698–0.4853). These improvements are explained by task decomposition, which reduces structural failure propagation, and deterministic verification, which rejects invalid outputs and preserves referential integrity. Ablation analysis indicates that the Course Character agent exerts the highest impact on overall performance. The proposed framework can be applied in higher education curriculum planning under OBE conditions, using minimal course metadata and producing machine-verifiable structured artifacts

Development of a hybrid information retrieval–knowledge graph model for cross-framework competency alignment

Halim Maulana — 2026-04-30

The object of this study is the alignment of Indonesian computing graduate learning outcomes (CPL – Capaian Pembelajaran Lulusan) with three heterogeneous competency frameworks: the European skills, competences, qualifications and occupations (ESCO), the occupational information network (O*NET), and the Standar Kompetensi Kerja Nasional Indonesia (SKKNI). The problem addressed is the absence of a unified pipeline capable of simultaneously mapping CPL across these frameworks while accounting for national qualification hierarchies and cross-lingual constraints. An IR-KG (information retrieval–knowledge graph) model is proposed with a seven-stage pipeline using a hybrid scoring function S_final = α·S_sem + β·S_gr + γ·S_con, integrating TF-IDF (term frequency–inverse document frequency) semantic similarity, ESCO knowledge graph cohesion, and domain constraint scores from ISCED-F 2013 (International Standard Classification of Education: Fields of Education and Training) and APTIKOM 2022 (Asosiasi Pendidikan Tinggi Informatika dan Komputer) classifications. The balanced configuration (α = β = γ = 0.33) achieves a mean selection objective of 0.537, a 26.1% improvement over the semantic baseline. External consistency validation yields a relaxed consistency rate of 27.1% (8.7× above random baseline), confirming valid alignment signal capture. The CRI-KG (Career Readiness index based on knowledge graph) reveals a gradient R_SKKNI >> R_ONET > R_ESCO, exposing persistent gaps in international framework coverage. The pipeline is applicable for curriculum audit, qualification recognition policy, and national-to-international framework integration where labelled training data are unavailable

Devising an approach to analyze and automatically reconfigure the structure of websites

Ivan Dolotov — 2026-04-30

This study explores websites such as online stores, which are considered to be a set of interconnected web pages. The task addressed relates to the high computational complexity of manual analysis of the topology of modern websites, as well as the lack of formalized mechanisms that could make it possible to integrate the semantic features of web pages into the process of automated hyperlink reconstruction.

Within the framework of this study, a website is crawled in order to obtain complete HTML documents, from which the structural features of pages are extracted (the number of headings, depth of embedding, presence of <article>, number of incoming links, etc.). The resulting vectors make it possible to construct cosine similarity matrices to assess the mutual proximity of pages. An approach has been proposed to rebuilding the link structure of the website taking into account this similarity; a comparison of the initial and transformed website was carried out using the metric characteristics of modularity, clustering, diameter, and similarity distribution.

The results demonstrate that taking into account the DOM structure allows for the formation of a logical, reasonable distribution of pages between clusters. And the subsequent automatic procedure for setting hyperlinks makes it possible to improve structural integrity by establishing effective relationships between thematically close pages.

The practical significance of this work involves the possibility of using the proposed approach for automated optimization of internal links of static websites. As a result, the architecture of the web resource is improved, website navigation becomes transparent, and website indexing by search engines is increased

Evaluation of a machine learning-assisted interactive evolutionary non-dominated sorting genetic algorithm -II framework for hyperparameter optimization of U-Net in agricultural land segmentation

Artughrul Gayibov — 2026-04-30

The object of the study is the hyperparameter configuration space of the U-Net architecture for agricultural land segmentation from Sentinel-2 satellite imagery.

The problem being solved is the excessive cost of multi-objective hyperparameter optimization, because non-dominated sorting in the non-dominated sorting genetic algorithm II (NSGA-II), with complexity O(MN²), becomes a bottleneck for deep segmentation models. To address this problem, an interactive evolutionary non-dominated sorting genetic algorithm II (IENSGA-II) framework is evaluated, in which a logistic regression classifier is trained on hyperparameter vectors and Pareto ranks from initial NSGA-II generations, then used to predict ranks in subsequent generations instead of full sorting. Unlike existing surrogate-assisted approaches, this work predicts Pareto ranks without additional model evaluations. On the panoptic agricultural satellite time series (PASTIS) benchmark, the framework reduced execution time by 20.07%, 16.39%, and 38.80% for 5, 10, and 15 generations, and in the 10-generation setting improved validation criteria, reaching an area under the receiver operating characteristic curve (AUC) of 0.9072 versus 0.9004 and validation loss of 0.6057 versus 0.6212. These results were achieved because the method accelerates selection rather than replacing model evaluation, while AUC-based tie-breaking preserves preference for more accurate solutions among candidates with same predicted rank. Effectiveness stems from a regular relationship between hyperparameters and Pareto ranks in early evolutionary data. In practice, the method is used in resource-constrained multi-objective learning when initial generations provide representative data for rank prediction

Development of system-type-centric method for architectural design of computer-software automation systems

Ihor Polataiko — 2026-04-30

The object of the study is the architectural design process and architectural documentation model of computer-software systems in the automation domain.

Existing architectural design methodologies are predominantly agnostic about the nature of computer-software systems, lacking the specific guidance required to address the fundamentally divergent architectural natures of heterogeneous computer-software systems in modern automation.

Within the scope of this study, a system-type-centric method of architectural design of computer-software automation systems is proposed. The proposed method introduces a comprehensive viewpoint-oriented architectural documentation model and a reproducible iterative architectural design process.

The method hierarchically structures architectural concerns at different levels, ensuring a systematic transition from abstract to concrete design. By defining a viewpoint applicability matrix across the different system types, the method ensures the creation of relevant, standardized, structured, consistent, traceable and interoperable architectural artifacts.

The method is based on the system-type-centric paradigm and the concepts of standards and approaches focused on architectural viewpoints. This combination is the main distinguishing feature of this method, which ensures the applicability of the method to a wide range of computer software systems, addressing the differences in the architectural nature of different types of systems.

The proposed method increases the interoperability of architectural design artifacts, defines a prescriptive process that structures architectural design and forms the basis for tools for automated AI-assisted architecture generation

Determining weight coefficients in the control system for a symmetry-compensating device using a fuzzy logic apparatus

Petro Plieshkov — 2026-04-30

This study investigates processes of automatic control over parameters of the distribution network mode at a voltage of 6–10 kV with a sharply changing asymmetric load.

The task addressed relates to the compromised efficiency of existing systems for automatic control over symmetry-compensating devices, which are based on multi-criteria optimization methods with unchanged weight coefficients. Such systems continue to give maximum priority to reducing the steady-state voltage deviation even if it meets regulatory requirements. This leads to incomplete utilization of regulating reserve of the symmetry-compensating device.

The system of automatic control over the symmetry-compensating device has been improved by integrating a fuzzy controller into its structure. This controller changes the values of weight coefficients in the objective function depending on the current values of the network mode parameters. The computer simulation has confirmed the effectiveness of the designed system, which provided a reduction in the mathematical expectations of the reactive power factor M[tanφ] by 13.9% and the voltage unbalance factor M[k_2U] by 8.5% compared to the baseline.

The increase in the efficiency of the control system is explained by its ability to adaptively change the weight coefficients. Thus, if there is a margin for the steady-state voltage deviation, the fuzzy controller reduces the weight coefficient that meets this criterion, directing the regulating resource of the symmetrical-compensating device to increase the level of reactive power compensation and reduce the level of asymmetry.

A feature of the designed control system is the use of the levels of approximation of the current values of the steady-state deviation and voltage asymmetry to their regulatory permissible limits as input variables in the fuzzy controller.

The designed automatic control system could be used in distribution electrical networks with a voltage of 6–10 kV at a sharply changing asymmetrical load

Construction of a model for optimal scheduling of energy storage systems in the day-ahead market considering market strategies and technical constraints

Oleksandr Havva — 2026-04-30

This study investigates the process of optimal planning of energy storage system operating schedules in the day-ahead market. The task addressed is to prevent unprofitable operation of energy storage systems.

A comprehensive mathematical model of daily planning of energy storage operation modes has been built, which takes into account technological efficiency, hardware wear processes, as well as asymmetry of market payments. This forms an adaptive filter with cutting off cycles of economically inefficient arbitrage. The approach is based on separation of the vector of control variables into two independent charging and discharging flows. This has made it possible to take into account the asymmetry of tariffs and the structure of losses in a differentiated manner.

The study on the sensitivity of the model revealed the existence of areas of adaptive regulation and parametric insensitivity to fluctuations in price signals. In the area of parametric insensitivity, the model forms a stable cost-effective solution regardless of price fluctuations. In the area of adaptive regulation, the model changes the volumes of a separate operating cycle depending on a change in the market price spread. It has been established that ignoring the accompanying payments for the area of adaptive regulation leads to an overestimation of income by 35.6 percent and the formation of unprofitable operations. Under conditions of low market price spread, the deviation in the financial result reaches 91.6 percent; however, the parametric insensitivity of the model guarantees profitable operation.

The scope of model's practical implementation is decision support systems for operators of energy storage systems. The conditions of use assume the presence of deterministic forecasts of market price signals for the calculated operating day. Implementing the model contributes to the extension of the life cycle and increase in the operational profitability of energy storage systems

Improving the approach to flexibly control formulations for a clean-in-place station

Roman Mirkevych — 2026-04-30

A clean-in-place (CIP) station has been investigated in this study. The task addressed relates to the lack of flexible methods for controlling formulations for a CIP station.

In most industrial systems, the cleaning algorithms are rigidly embedded in the control program, which complicates modification of technological procedures without intervention in the program code. This paper reports the development of a formalized approach to controlling the formulation for a CIP station based on the IEC 61512 standard principles, which ensures the separation of technological procedure descriptions from the implementation of equipment control functions.

The study includes an analysis of requirements within the IEC 61512 standard for formulation description, as well as the construction of a formalized model of procedures. An approach has been proposed to represent cleaning formulations as a structured list of technological operations with process parameters, which makes it possible to formalize the sequence of cleaning stages. It has been shown that such a technique of formulation description could be directly implemented in programmable logic controllers (PLCs) and systems of supervisory control and data acquisition (SCADA) without using additional specialized tools.

The results are attributed to the application of principles of equipment and procedure decomposition in accordance with the IEC 61512 standard, which makes it possible to separate process control from equipment control. A distinctive feature of the proposed approach is the possibility of modifying CIP cleaning recipes without changing the PLC program code, which reduces the time required for system reconfiguration.

The results could be practically implemented in the design and modernization of automated control systems for CIP stations in the food, pharmaceutical, and biotechnology industries, as well as in the development of SCADA interfaces for controlling formulation-related processes

Development of battery charging-discharging current regulation strategy by using fuzzy logic controllers

Rahim Mammadzada — 2026-04-30

The object of this study is to regulate charge and discharge currents in battery systems and its affect on thermal behavior under dynamic battery operating conditions. The problem to be solved is the development of battery management framework which accounts for charging current regulation and a load shedding mechanism for discharge current regulation for temperature and voltage control of battery bank.

Simulation results demonstrate that the proposed hybrid framework for the battery control system maintains stable temperature levels during both charging and discharging cycles, where during charging the current is regulated based on terminal voltage, temperature and rate of change of temperature, while during discharging control is based on temperature and rate of change of both temperature rise and voltage drop.

The distinctive feature of the proposed framework is the hybrid control structure that applies different regulation strategies for charging and discharging processes by using fuzzy logic controllers. In this approach, fuzzy logic adjusts the charging current according to the state of charge, voltage, and temperature of the battery cells, at the same time load shedding and load transfer regulate the discharge current to maintain safe operating conditions.

The proposed system was simulated in MATLAB with 3 charge–discharge cycles. It maintained temperature within 45°C setpoint with minimal overshoot and voltage drop amount not more than 25% nominal battery bank voltage.

The results can be applied to the development of battery management systems for battery banks in energy storage, electrical vehicles, essential power supplies in industrial applications, and solar/wind power generation system