The paper presents a study of the efficiency of loading and storing data in the three most common Data Lakehouse systems, including Apache Hudi, Apache Iceberg, and Delta Lake, using Apache Spark as a distributed data processing platform. The study analyzes the behavior of each system when processing structured (CSV) and semi-structured (JSON) data of different sizes, including loading files up to 7 GB in size. The purpose of the work is to determine the most optimal Data Lakehouse architecture based on the type and volume of data sources, data loading performance using Apache Spark, and disk size of data for forming analytical data systems. The research covers the development of four sequential ETL processes, which include reading, transforming, and loading data into tables in each of the Data Lakehouse systems. The efficiency of each Lakehouse was evaluated according to two key criteria: data loading time and the volume of tables formed in the file system. For the first time, a comparison of performance and data storage in Apache Iceberg, Apache Hudi, and Delta Lake Data Lakehouse systems was conducted to select the most relevant architecture for building analytical data systems. The practical value of the study consists in the fact that it assists data engineers and architects in choosing the most appropriate Lakehouse architecture, understanding the balance between loading performance and storage efficiency. Experimental results showed that Delta Lake is the most optimal architecture for systems where the priority is the speed of loading data of any volume, while Apache Iceberg is most appropriate for systems where stability and disk space savings are critical. Apache Hudi proved ineffective in data loading and storage evaluation tasks but could potentially be effective in incremental update and streaming processing scenarios.
The methods for modifying coniferous wood by impregnation with epoxy resins were improved in research. Implementation of such methods increased the efficiency of interaction between the polymer composition and the wood structure. A comprehensive methodology for experimental research has been developed and thoroughly detailed. It encompasses sample preparation, autoclave treatment procedures, curing conditions for the modifying material, and the mechanical test procedure. Full-scale experiments revealed the deformation characteristics and performance of modified and unmodified softwood species – pine, larch, and spruce – accounting for differences in failure mechanisms. Obtaining experimental data on the behaviour of epoxy-modified wood specimens under compression along the grain allowed a more complete assessment of changes in strength compared to untreated samples. Autoclave modification significantly increases wood strength: by 1.29 times for pine, 1.21 times for larch, and 1.36 times for spruce, confirming the effectiveness of this technology for improving mechanical properties.
This study presents the development of a physical model for the mechanical dewatering of wet compacts composed of a binary mixture of soft microporous particles of two different sizes. The layer of microporous particles is conceptualized as a double-porosity system, comprising extraparticle and two distinct intraparticle spaces, each characterized by unique permeability and compressibility properties. Analytical solutions for pressure distributions and consolidation ratios within each space were derived using Heaviside's operational method with Laplace and Fourier integral transformations. The results highlight the impact of particle size and structure on dewatering efficiency and elucidate the role of internal liquid flow within the compressible wet compact. The model also accounts for the consolidation behaviour of both intra- and extraparticle spaces, enabling more accurate predictions of the mechanical dewatering process.
This study investigates the dynamic corrosion behavior of 27SiMn hydraulic support steel in the mine water using electrochemical noise (EN) monitoring combined with potentiodynamic polarization curves (PPCs) and Mott-Schottky (M-S) analysis. To establish the relationship between electrochemical signals and corrosion mechanisms, the electrochemical current noise (ECN) data were processed by Hilbert-Huang transform, shot noise analysis, and recurrence quantification analysis (RQA). The results show that the corrosion process of 27SiMn steel exhibits three distinct stages during the spraying experiments. From the 1st to 4th cycle, the Hilbert spectrum displays dense, high-frequency, and aperiodic transient signals corresponding to metastable pitting. Concurrently, the steel exhibits increasing pitting susceptibility and p-type semiconductor behavior with high defect density. By the 5th cycle, the ECN transient signal significantly decreases, while RQA parameters increase markedly, indicating that the corrosion signal stabilizes and exhibits periodic characteristics. Meanwhile, the semiconductor behavior of 27SiM steel from p-type to n-p-type, due to the formation of complete corrosion product layers composed of Fe2O3, Fe3O4, and Mn3O4. During the 6th and 7th cycles, the Hilbert spectrum again shows clusters of high-density random transient signals, accompanied by a sharp decline in RQA parameters and a reversion to a p-type semiconductor state with high defect density.
The article addresses the urgent scientific and practical problem of overcoming the "cold start" effect in the development and deployment of Natural Language Processing (NLP) systems aimed at monitoring public opinion and sentiment analysis of Ukrainian-language content. The critical shortage of representative, balanced, and pre-labeled datasets that accurately reflect the specifics of social, economic, and political processes in modern Ukrainian society is identified as the main barrier to integrating advanced neural network solutions. To mitigate this deficit and eliminate resource-intensive manual labeling, the study proposes and programmatically implements a method of algorithmic synthesis of training corpora using the generative capabilities of Large Language Models (LLMs). To prevent semantic stagnation, algorithmic bias, and structural repetitions during mass generation, a controlled stochastic synthesis method was developed. Its core relies on utilizing a high entropy parameter (temperature) combined with the dynamic configuration of prompts using the "Subject-Predicate-Object" (S-P-O) matrix. The practical implementation of the high-performance generation pipeline was executed based on the asynchronous Node.js architecture. The application of parallel request dispatching, an adaptive backoff strategy for HTTP errors, and JSON Lines stream writing enabled the synthesis of a representative dataset comprising over 100,000 records. To prepare the generated corpus for subsequent machine learning (specifically for recurrent architectures like LSTM), an optimized preprocessing pipeline was engineered. It incorporates a deterministic hierarchical mapping method based on regular expressions, allowing the reduction of thematic diversity into 19 strategic domains and one integral layer with linear computational complexity O(n). Furthermore, to reduce the dimensionality of the feature space, a custom morphological reduction (stemming) algorithm for the Ukrainian language was developed, which effectively trims inflectional affixes while preserving word roots. As a result of frequency filtering and optimization, a structured vocabulary of 19,481 tokens was formed and initialized with alignment service vectors. The proposed architectural model and software solutions prove their efficiency for the rapid deployment of Ukrainian-language analytical NLP systems under the conditions of labeled data scarcity.
