Data Structure Research Articles

Applying clustered artificial neural networks to enhance contaminant diffusion prediction in geotechnical engineering

This paper introduces a novel approach using Clustered Artificial Neural Networks (CLANN) to address the challenge of developing predictive models for multimodal dataset with extreme parameter values. The CLANN method strategically decomposes the dataset, derived from Finite Element Analysis (FEA), into clusters, each representing distinct diffusion behaviors, and applies specialized neural networks within these clusters. The CLANN model was rigorously evaluated and demonstrated superior accuracy and consistency compared to traditional methods such as the Adaptive Neuro-Fuzzy Inference System (ANFIS) and fuzzy expert systems. While these conventional models struggled to capture the full range of diffusion dynamics, particularly under extreme conditions, CLANN consistently provided predictions that closely aligned with the actual FEA data across all scenarios. The versatility of the CLANN approach extends beyond its application to soil contamination. Its ability to handle complex, multimodal datasets suggests that this methodology can be generalized to a wide range of scientific and engineering problems characterized by similar data structures. This makes CLANN not only a powerful tool for geotechnical engineers but also a promising framework for broader applications where traditional models fall short. The findings of this study pave the way for more accurate, reliable, and adaptable predictive modeling in diverse domains, enhancing our ability to manage and mitigate environmental and engineering challenges.

Perbandingan Uji Performa Impala dan Hive-Hadoop

The amount of data that is growing rapidly today also requires fast storage. This is because the need for data is also very important, and to access the data also requires fast time. Therefore, it is necessary to know the tools that support the processing of large amounts of data and fast time. The presence of Impala and Hive-Hadoop helps in making decisions about which tools to use to store data and quickly get the data needed. In this analysis and comparison, we want to know how the performance of these two tools is, namely whether Impala and Hive-Hadoop can access structured data. This research was conducted using experimental methods and manipulating the experimental environment using computer virtualization. The analysis results obtained are, Impala is faster than Hive-Hadoop because it reduces latency and Impala is not based on the MapReduce algorithm.

CATH v4.4: major expansion of CATH by experimental and predicted structural data.

CATH (https://www.cathdb.info) is a structural classification database that assigns domains to the structures in the Protein Data Bank (PDB) and AlphaFold Protein Structure Database (AFDB) and adds layers of biological information, including homology and functional annotation. This article covers developments in the CATH classification since 2021. We report the significant expansion of structural information (180-fold) for CATH superfamilies through classification of PDB domains and predicted domain structures from the Encyclopedia of Domains (TED) resource. TED provides information on predicted domains in AFDB. CATH v4.4 represents an expansion of ∼64 844 experimentally determined domain structures from PDB. We also present a mapping of ∼90 million predicted domains from TED to CATH superfamilies. New PDB and TED data increases the number of superfamilies from 5841 to 6573, folds from 1349 to 2078 and architectures from 41 to 77. TED data comprises predicted structures, so these new folds and architectures remain hypothetical until experimentally confirmed. CATH also classifies domains into functional families (FunFams) within a superfamily. We have updated sequences in FunFams by scanning FunFam-HMMs against UniProt release 2024_02, giving a 276% increase in FunFams coverage. The mapping of TED structural domains has resulted in a 4-fold increase in FunFams with structural information.

Developing an Ontology on Battery Production and Characterization with the Help of Key Use Cases from Battery Research

Materials science research faces challenges due to diverse and evolving measurements, materials, and methods. Managing research data in a way that is understandable, comparable, and reproducible is essential for high data quality, particularly for data science and machine learning applications. In Li‐ion batteries research data storage concepts and structures vary widely between institutions and researchers, leading to difficulties in data comparison and understanding. To address the issue of data structuring, battery production and characterization ontology (BPCO) is developed. The ontology builds on existing ontologies like the Platform MaterialDigital core ontology and quantities, units, dimensions, and types ontology to model standard battery production processes, characterization methods, and materials. The BPCO is based on a workflow structure to be accessible to nonexperts and, unlike highly specialized existing ontologies, models the whole production process removing the need for separate data structures and enabling the identification of dependencies between parameters. This work builds upon a previously published paper in which the taxonomy and fundamental strategies for ontology development are established. The article presents the developed ontology and its use for structuring research data in three key use cases, that is, different experiments performed to validate the ontology's capabilities, provide feedback, and ensure its applicability.

Research on Fine-Tuning Optimization Strategies for Large Language Models in Tabular Data Processing.

Recent advancements in natural language processing (NLP) have been significantly driven by the development of large language models (LLMs). Despite their impressive performance across various language tasks, these models still encounter challenges when processing tabular data. This study investigates the optimization of fine-tuning strategies for LLMs specifically in the context of tabular data processing. The focus is on the effects of decimal truncation, multi-dataset mixing, and the ordering of JSON key-value pairs on model performance. Experimental results indicate that decimal truncation reduces data noise, thereby enhancing the model's learning efficiency. Additionally, multi-dataset mixing improves the model's generalization and stability, while the random shuffling of key-value pair orders increases the model's adaptability to changes in data structure. These findings underscore the significant impact of these strategies on model performance and robustness. The research provides novel insights into improving the practical effectiveness of LLMs and offers effective data processing methods for researchers in related fields. By thoroughly analyzing these strategies, this study aims to establish theoretical foundations and practical guidance for the future optimization of LLMs across a broader range of application scenarios.

Data-Sharing System with Attribute-Based Encryption in Blockchain and Privacy Computing

With the development of the data-sharing system in recent years, financial management systems and their privacy have sparked great interest. Existing financial data-sharing systems store metadata, which include a hash value and database index on the blockchain, and store high-capacity actual data in the center database. However, current data-sharing systems largely depend on centralized systems, which are susceptible to distributed denial-of-service (DDoS) attacks and present a centralized attack vector. Furthermore, storing data in a local center database has a high risk of information disclosure and tampering. In this paper, we propose the ChainMaker Privacy Computing (CPC) system, a new decentralized data-sharing system for secure financial data, to solve this problem. It provides a series of financial data information and a data structure rather than actual data on the blockchain to protect the privacy of data. We utilize a smart contract to establish a trusted platform for the local database to obtain encrypted data. We design a resource catalog to provide a trusted environment of data usage in the privacy computing system that is visible for members on the blockchain. Based on cipher-policy attribute-based encryption (CP-ABE), We design a CPC-CP-ABE algorithm to enable fine-grained access control through attribute based encryption. Finally, We propose an efficient scheme that allows authenticated data-sharing systems to perform Boolean searches on encrypted data information. The results of experiment show that the CPC system can finish trusted data sharing to all organizations on the blockchain.

Joint Classification of Hyperspectral and LiDAR Data via Multiprobability Decision Fusion Method

With the development of sensor technology, the sources of remotely sensed image data for the same region are becoming increasingly diverse. Unlike single-source remote sensing image data, multisource remote sensing image data can provide complementary information for the same feature, promoting its recognition. The effective utilization of remote sensing image data from various sources can enhance the extraction of image features and improve the accuracy of feature recognition. Hyperspectral remote sensing (HSI) data and light detection and ranging (LiDAR) data can provide complementary information from different perspectives and are frequently combined in feature identification tasks. However, the process of joint use suffers from data redundancy, low classification accuracy and high time complexity. To address the aforementioned issues and improve feature recognition in classification tasks, this paper introduces a multiprobability decision fusion (PRDRMF) method for the combined classification of HSI and LiDAR data. First, the original HSI data and LiDAR data are downscaled via the principal component–relative total variation (PRTV) method to remove redundant information. In the multifeature extraction module, the local texture features and spatial features of the image are extracted to consider the local texture and spatial structure of the image data. This is achieved by utilizing the local binary pattern (LBP) and extended multiattribute profile (EMAP) for the two types of data after dimensionality reduction. The four extracted features are subsequently input into the corresponding kernel–extreme learning machine (KELM), which has a simple structure and good classification performance, to obtain four classification probability matrices (CPMs). Finally, the four CPMs are fused via a multiprobability decision fusion method to obtain the optimal classification results. Comparison experiments on four classical HSI and LiDAR datasets demonstrate that the method proposed in this paper achieves high classification performance while reducing the overall time complexity of the method.

Distinguishing coupled dark energy models with neural networks

We investigate whether neural networks (NNs) can accurately differentiate between growth-rate data of the large-scale structure (LSS) of the Universe simulated via two models: a cosmological constant and Lambda cold dark matter (CDM) model and a tomographic coupled dark energy (CDE) model. We built an NN classifier and tested its accuracy in distinguishing between cosmological models. For our dataset, we generated $f growth-rate observables that simulate a realistic Stage IV galaxy survey-like setup for both Lambda CDM and a tomographic CDE model for various values of the model parameters. We then optimised and trained our NN with Optuna aiming to avoid overfitting and to maximise the accuracy of the trained model. We conducted our analysis for both a binary classification, comparing between Lambda CDM and a CDE model where only one tomographic coupling bin is activated, and a multi-class classification scenario where all the models are combined. For the case of binary classification, we find that our NN can confidently (with $>86<!PCT!>$ accuracy) detect non-zero values of the tomographic coupling regardless of the redshift range at which coupling is activated and, at a $100<!PCT!>$ confidence level, detect the Lambda CDM model. For the multi-class classification task, we find that the NN performs adequately well at distinguishing Lambda CDM, a CDE model with low-redshift coupling, and a model with high-redshift coupling, with 99<!PCT!>, 79<!PCT!>, and 84<!PCT!> accuracy, respectively. By leveraging the power of machine learning, our pipeline can be a useful tool for analysing growth-rate data and maximising the potential of current surveys to probe for deviations from general relativity.

UTILIZING RETRIEVAL-AUGMENTED GENERATION IN LARGE LANGUAGE MODELS TO ENHANCE INDONESIAN LANGUAGE NLP

The improvement of Large Language Models (LLM) such as ChatGPT through Retrieval-Augmented Generation (RAG) techniques has urgency in the development of natural language translation technology and dialogue systems. LLMs often experience obstacles in addressing special requests that require information outside the training data. This study aims to discuss the use of Retrieval-Augmented Generation (RAG) on large-scale language models to improve the performance of Natural Language Processing (NLP) in Indonesian, which has so far been poorly supported by high-quality data and to overcome the limitations of traditional language models in understanding the context of Indonesian better. The method used is a combination of retrieval capabilities (external information search) with generation (text generation), where the model utilizes broader and more structured basic data through the retrieval process to produce more accurate and relevant text. The data used includes the Indonesian corpus of the 30 Juz Quran translation into Indonesian. The results of the trial show that the RAG approach significantly improves the performance of the model in various NLP tasks, including token usage optimization, text classification, and context understanding, by increasing the accuracy and relevance of the results

Methodology of a Comprehensive Approach to the Analysis, Structuring, and Aggregation of Data Based on the Requirements of Resource-Intensive Applications

The work presents a methodology for in-depth data description for resource-intensive applied tasks, particularly in the field of materials science. It consists of three basic stages. In the first stage, quantitative and semantic requirements are formalized, and working datasets are defined, which can be classified as reference, structured, instrumental, or experimental-computational, depending on the source. These datasets are stored in a specialized research area for further investigation. The second stage involves the analysis and hierarchical structuring of the accumulated information content. This allows for the refinement and comparison of data, storage of different sets, and the application of various processing tools. In the third stage, the data is aggregated and represented as objects in the domain area according to the database schema of the application support system. The steps outlined in the proposed methodology enable full utilization of the specialized research area of informational content for building object models of application scenarios, structuring the domain data representation schema, and forming event-information forms. The practical application of this methodology in the examined subject area clarified the requirements, criteria, and filters for information search, data description, parameter cataloging, and their verification when solving a number of practical tasks. The software implementation of this approach could be oriented towards a web platform for local and cloud technologies, allowing remote collaborative access to content requirements and solution catalogs with role-based access for different categories of researchers and users.

Bridging theory and data: A computational workflow for cultural evolution

Cultural evolution applies evolutionary concepts and tools to explain the change of culture over time. Despite advances in both theoretical and empirical methods, the connections between cultural evolutionary theory and evidence are often vague, limiting progress. Theoretical models influence empirical research but rarely guide data collection and analysis in logical and transparent ways. Theoretical models themselves are often too abstract to apply to specific empirical contexts and guide statistical inference. To help bridge this gap, we outline a quality-assurance computational workflow that starts from generative models of empirical phenomena and logically connects statistical estimates to both theory and real-world explanatory goals. We emphasize and demonstrate validation of the workflow using synthetic data. Using the interplay between conformity, migration, and cultural diversity as a case study, we present coded and repeatable examples of directed acyclic graphs, tailored agent-based simulations, a probabilistic transmission model for longitudinal data, and an approximate Bayesian computation model for cross-sectional data. We discuss the assumptions, opportunities, and pitfalls of different approaches to generative modeling and show how each can be used to improve data analysis depending on the structure of available data and the depth of theoretical understanding. Throughout, we highlight the significance of ethnography and of collecting basic cultural and demographic information about study populations and call for more emphasis on logical and theory-driven workflows as part of science reform.

Anemia status and its determinants among reproductive-age women in Tanzania: A multi-level analysis of Tanzanian demographic and health survey data.

Anemia is significantly associated with morbidity and mortality in women of reproductive age. Determining the prevalence and identifying associated risk factors remains an important topic in public health. Therefore, this study aimed to estimate and identify the risk factors for anemia in women of reproductive age in Tanzania. A cross-sectional study utilizing secondary data from the Tanzania Demographic and Health Survey of 2022 was conducted. Weighted total samples of 8,921 reproductive-age women were included in the analysis. A multilevel Poisson regression model was employed to account for the hierarchical structure of the TDHS data. Deviance was used to compare the models. In the multivariable multilevel mixed-effects Poisson regression model with robust variance, adjusted prevalence ratios with corresponding 95% confidence intervals were used to identify the determinants of anemia. The prevalence of anemia among women of reproductive age in Tanzania was 42.02% [95% CI: (40.79%-43.25%)]. Of these, 19.82%, 19.35%, and 2.84% had mild, moderate, and severe anemia, respectively. The regression results revealed that women from households with the poorest wealth quantiles, underweight, pregnant status, high community poverty level, and women from the eastern and Zanzibar administrative zones had a higher prevalence of anemia. While currently employed, moderate alcohol consumption and use of hormonal contraceptive methods were associated with a lower prevalence of anemia. The findings of the study showed that anemia is a severe public health issue among women of reproductive age in Tanzania, affecting more than four in ten women. Interventions targeting the improvement of nutrition, access to healthcare services, and education on anemia prevention and management should be prioritized to reduce the burden of anemia effectively.

Digitalization and Dynamic Criticality Analysis for Railway Asset Management

The primary aim of this paper is to support the optimization of asset management in railway infrastructure through digitalization and criticality analysis. It addresses the current challenges in railway infrastructure management, where data-driven decision making and automation are key for effective resource allocation. The paper presents a methodology that emphasizes the development of a robust data model for criticality analysis, along with the advantages of integrating advanced digital tools. A master table is designed to rank assets and automatically calculate criticality through a novel asset attribute characterization (AAC) process. Digitalization facilitates dynamic, on-demand criticality assessments, which are essential in managing complex networks. The study also underscores the importance of combining digital technology adoption with organizational change management. The data process and structure proposed can be viewed as an ontological framework adaptable to various contexts, enabling more informed and efficient asset ranking decisions. This methodology is derived from its application to a metropolitan railway network, where thousands of assets were evaluated, providing a practical approach for conducting criticality assessments in a digitized environment.

Lateritic Ni–Co Prospectivity Modeling in Eastern Australia Using an Enhanced Generative Adversarial Network and Positive-Unlabeled Bagging

AbstractThe surging demand for Ni and Co, driven by the acceleration of clean energy transitions, has sparked interest in the Lachlan Orogen of New South Wales for its potential lateritic Ni–Co resources. Despite recent discoveries, a substantial knowledge gap exists in understanding the full scope of these critical metals in this geological province. This study employed a machine learning-based framework, integrating multidimensional datasets to create prospectivity maps for lateritic Ni–Co deposits within a specific Lachlan Orogen segment. The framework generated a variety of data-driven models incorporating geological (rock units, metamorphic facies), structural, and geophysical (magnetics, gravity, radiometrics, and remote sensing spectroscopy) data layers. These models ranged from comprehensive models that use all available data layers to fine-tuned models restricted to high-ranking features. Additionally, two hybrid (knowledge-data-driven) models distinguished between hypogene and supergene components of the lateritic Ni–Co mineral systems. The study implemented data augmentation methods and tackled imbalances in training samples using the SMOTE–GAN method, addressing common machine learning challenges with sparse training data. The study overcame difficulties in defining negative training samples by translating geological and geophysical data into training proxy layers and employing a positive and unlabeled bagging technique. The prospectivity maps revealed a robust spatial correlation between high probabilities and known mineral occurrences, projecting extensions from these sites and identifying potential greenfield areas for future exploration in the Lachlan Orogen. The high-accuracy models developed in this study utilizing the Random Forest classifier enhanced the understanding of mineralization processes and exploration potential in this promising region.

A multi-feature fusion graph attention network for decoding motor imagery intention in spinal cord injury patients.

Electroencephalogram (EEG) signals exhibit multi-domain features, and electrode distributions follow non-Euclidean topology. To fully resolve the EEG signals, this study proposes a Temporal-Frequency-Spatial multi-domain feature fusion Graph Attention Network (TFSGAT) for motor imagery (MI) intention recognition in spinal cord injury (SCI) patients. The proposed model uses phase-locked value (PLV) to extract spatial phase connectivity information between EEG channels and continuous wavelet transform to extract valid EEG information in the time-frequency domain. It then models a graph data structure containing multi-domain information. The gated recurrent unit and GAT learn EEG's dynamic temporal-spatial information. Finally, the fully connected layer outputs the MI intention recognition results. After 10 times 10-fold cross-validation, the proposed model can achieve an average accuracy of 95.82%. Furthermore, this study analyzes the Event-Related Desynchronization/Event-Related Synchronization and PLV brain network to explore the brain activity of SCI patients during MI. This study confirms the potential of the proposed model in terms of EEG decoding performance and provides a reference for the mechanism of neural activity in SCI patients.

A novel controllability method on temporal networks based on tree model

Temporal networks have become instrumental in modeling dynamic systems across various disciplines, presenting unique challenges and opportunities in understanding and influencing their behavior. Controllability, a fundamental aspect of network dynamics, plays a pivotal role in manipulating these systems towards desired states. In this paper, we embark on a comprehensive exploration of controllability within the realm of temporal networks. A new method for controlling temporal networks is proposed, in which the intervention of all the dynamics of temporal networks can provide the possibility to speed up the network controllability processes. In the proposed method, the network dynamics are stored in the tree data structure to reduce the computational complexity of the algorithm for finding control nodes while maintaining essential information in controllable processes. Results show that the proposed algorithm with linear complexity of O(N2logNΔt4)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\\varvec{O}}({{\\varvec{N}}}^{2}{\\varvec{l}}{\\varvec{o}}{\\varvec{g}}{\\varvec{N}}{\\Delta {\\varvec{t}}}^{4})$$\\end{document}. Evaluation against conventional methods on experimental datasets reveals notable improvements: a 41.8% reduction in the minimum number of control nodes, a 36.37% decrease in time of receiving fully control network, and a 38.5% reduction in control algorithm execution time compared to layered model-based control methods.

Automated Structural Bolt Micro Looseness Monitoring Method Using Deep Learning.

The detection of bolt loosening in key components of aircraft engines faces problems such as complex and difficult-to-establish bolt loosening mechanism models, difficulty in identifying early loosening, and difficulty in extracting signal features with nonlinear and non-stationary characteristics. Therefore, the automated structural bolt micro looseness monitoring method using deep learning was proposed. Specifically, the addition of batch normalization methods enables the established Batch Normalized Stacked Autoencoders (BNSAEs) model to converge quickly and effectively, making the model easy to build and effective. Additionally, using characterization functions preprocess the original response signal not only simplifies the data structure but also ensures the integrity of features, which is beneficial for network training and reduces time costs. Finally, the effectiveness of the proposed method was verified by taking the bolted connection structures of two key components of aircraft engines, namely bolt connection structures and flange connection structures, as examples.

Structural Complexity Significantly Impacts Canopy Reflectance Simulations as Revealed from Reconstructed and Sentinel-2-Monitored Scenes in a Temperate Deciduous Forest

Detailed three-dimensional (3D) radiative transfer models (RTMs) enable a clear understanding of the interactions between light, biochemistry, and canopy structure, but they are rarely explicitly evaluated due to the availability of 3D canopy structure data, leading to a lack of knowledge on how canopy structure/leaf characteristics affect radiative transfer processes within forest ecosystems. In this study, the newly released 3D RTM Eradiate was extensively evaluated based on both virtual scenes reconstructed using the quantitative structure model (QSM) by adding leaves to point clouds generated from terrestrial laser scanning (TLS) data, and real scenes monitored by Sentinel-2 in a typical temperate deciduous forest. The effects of structural parameters on reflectance were investigated through sensitivity analysis, and the performance of the 3D model was compared with the 5-Scale and PROSAIL radiative transfer models. The results showed that the Eradiate-simulated reflectance achieved good agreement with the Sentinel-2 reflectance, especially in the visible and near-infrared spectral regions. Furthermore, the simulated reflectance, particularly in the blue and shortwave infrared spectral bands, was clearly shown to be influenced by canopy structure using the Eradiate model. This study demonstrated that the Eradiate RTM, based on the 3D explicit representation, is capable of providing accurate radiative transfer simulations in the temperate deciduous forest and hence provides a basis for understanding tree interactions and their effects on ecosystem structure and functions.

A Research Approach to Port Information Security Link Prediction Based on HWA Algorithm

For the protection of information security, link prediction, as a basic problem of network science, has important application significance. However, most of the existing link prediction algorithms rely on the node information of the graph structure, which is not applicable in some graph structure data involving privacy. At the same time, most of the algorithms only consider the general graph structure and do not fully consider the high-order information in the graph. Because of this, this paper proposes an algorithm called hypergraph-based link prediction with self-attention (HWA) to solve the above problems. The algorithm can obtain hypergraphs without knowing the attribute information of hypergraph nodes and combines the graph convolutional network (GCN) framework to capture node feature information for link prediction. Experiments show that the HWA algorithm proposed in this paper, combined with the GCN framework, shows better link prediction performance than other graph-based neural network benchmark algorithms on eight real networks. This further verifies the validity and reliability of the model in this paper and provides new protection ideas and technical means for information security.

Entropy corrected geometric Brownian motion

The geometric Brownian motion (GBM) is widely used for modeling stochastic processes, particularly in finance. However, its solutions are constrained by the assumption that the underlying distribution of returns follows a log-normal distribution. This assumption limits the predictive power of GBM, especially in capturing the complexities of real-world data, where deviations from log-normality are common. In this work, we introduce entropy corrections to the GBM framework to relax the log-normality constraint and better account for the inherent structures in real data. We demonstrate that as the deterministic components within the data increase, entropy decreases, leading to refinements in GBM’s predictive accuracy. Our approach shows significant improvements over conventional GBM in handling distributions that deviate from log-normal behavior, as demonstrated through both a simple dice roll experiment and real-world financial data. Beyond just financial modeling, this research also opens up new avenues for generating synthetic data that better captures real-world complexity, enhancing applications in fields like machine learning and statistical modeling.