Fulltext search in archive

Background: Benford’s law is a statistical phenomenon that predicts the probability of a particular digit at a particular position in a number. This law has been successfully applied in a number of areas, such as accounting. In the area of scientometrics, research has been devoted mostly to journal data.Objective: This paper investigates the conformity of Benford’s law with the citation counts of records retrieved from the Web of Science database. We evaluate the conformity levels with Benford’s law in the complete dataset. We determine the effect of document type (article, proceedings paper and review), year of publication (2014–2018) and Web of Science categories (254 categories) on the level of conformity of the citation counts with Benford’s law.Methods: The dataset of this research contains over 8.47 million records. All available records from the Web of Science were downloaded, so this set is the entire population of data available at the time of download. The distributions of the first significant digits in the citation counts of these records are compared with Benford’s law. Mean absolute deviation (MAD) recommended by Nigrini (2012) and sum of squared deviations (SSD) recommended by Kossovsky (2015) are used to categorize the similarity of the citation counts to Benford’s law.Results: The entire dataset of this study shows marginal conformity according to both MAD and SSD intervals (with a MAD value of 0.1257 and an SSD value of 29.9; a lower value indicates a better agreement). The review document type shows a high level of conformity, while proceedings paper shows a lower level. We found significant differences in conformity between Web of Science categories.Conclusion: This study mapped the level of conformity of the citation counts with Benford’s law in data from the Web of Science database. Further directions for possible research are suggested.

Background: SHapley Additive exPlanations (SHAP) methods are widely used to interpret machine learning models, yet most implementations assume feature independence. This assumption rarely holds in practice, especially when features are correlated, leading to biased and unstable attributions.Objective: We introduce Corr-SHAP, a correlation-aware SHAP approach that produces more faithful and stable feature attributions by explicitly modeling feature dependencies. Our aim is to enhance the accuracy, robustness, and scalability of SHAP explanations for models trained on correlated data.Methods: Corr-SHAP models feature correlations via a multivariate Gaussian approximation with a Ledoit–Wolf covariance estimator. We design a correlation-aware sampling distribution that penalizes redundant coalitions, improving computational efficiency in higher dimensions. To correct the induced bias, we employ a Self-Normalized Importance Sampling estimator, which re-weights samples by the ratio of the true Shapley kernel to the sampling probability. Our analysis establishes high probability error bounds in terms of Effective Sample Size, extending convergence guarantees to correlated feature spaces.Results: Across synthetic and real-world datasets, Corr-SHAP achieves Shapley value estimates that closely align with Kernel SHAP, while exhibiting substantially lower variance and more stable feature rankings. In correlated clusters, Corr-SHAP systematically down-weights redundant features, improving ranking fidelity without introducing bias. To further support scalability, we demonstrate that combining Corr-SHAP with Leverage-SHAP reduces variance in higher-dimensional settings.Conclusion: Corr-SHAP provides a statistically grounded and computationally efficient framework for SHAP value estimation under feature correlation. By integrating correlation modeling, bias correction, and variance reduction, it scales beyond small toy problems and delivers explanations that are both accurate and reliable, making it a valuable tool for practitioners analyzing complex real-world datasets.

The technological advancements in the field of E-healthcare have resulted in unprecedented generation of medical data which increases the risk of data security and privacy. Ensuring the privacy of Electronic Health Records (EHR) has become challenging due to outsourcing of healthcare information in the cloud. This increases the chance of data leakage to unauthorized users and affects the privacy and integrity of the user data. It requires a trustworthy central authority to protect the sensitive patient information from both internal and external attacks. This paper presents a blockchain based privacy preservation framework for securing EHR data. The proposed framework integrates the immutability and decentralized nature of blockchain with advanced cryptographic techniques to ensure the confidentiality, integrity and availability of EHR. The EHR data are stored in an InterPlanetary File System (IPFS) which is encrypted using a hybrid cryptographic algorithm. In addition, a novel smart contact based patient-centric access control is designed in this paper using a blockchain-based SHA-256 hashing algorithm to protect the privacy of patient data. The experimental results show that the proposed framework enables secure sharing of health information between network users with improved data privacy and security. Furthermore, the optimized search process reduces the time and space complexity compared to the traditional search process. Through the utilization of smart contracts, this framework enforces patient-centric access controls and allows patients to manage and authorize access to their medical data.

Artificial intelligence methods, especially deep learning, have seen increasing application in analysing personality and occupational data to identify individuals with psychological and neurological disorders. Currently, there is a great need for effectively processing mental healthcare with the integration of artificial intelligence such as machine learning and deep learning. The paper addresses the pressing need for accurate and efficient methods for diagnosing psychiatric disorders, which are often complex and multifaceted. By exploiting the power of convolutional neural networks (CNN), we propose a novel CNN-based natural language processing method without removing stop words for predicting psychiatric diagnoses capable of accurately classifying individuals based on their psychological data. Our proposal is based on keeping a richer linguistic and semantic context to accurately predict psychiatric diagnosis. The experiment involves two datasets: one gathered from a private clinic and the other from Kaggle, called the Human Stress Dataset. The outcomes from the first dataset demonstrate a remarkable accuracy rate of 98.51% when employing CNN, showcasing their superior performance compared to the standard machine learning techniques such as logistic regression, k-nearest neighbours and support vector machines. With the second dataset, our model achieved an impressive area under the receiver operating characteristic curve (AUROC) of 0.87. This result surpasses those achieved by existing state-of-the-art methods, further highlighting the efficacy of our CNN-based approach in discerning subtle nuances within the data and making accurate predictions. Moreover, we have compared our model with three other programs on the same dataset and the accuracy reached 78.52%. The results are promising to aid parents or clinicians in early and rapidly predicting the ill individual.

Background: The study investigates the emerging concept of Last Mile Logistics (LML), focusing on optimization routing for e-commerce deliveries. This research addresses the growing demand for efficient delivery systems driven by the rise of online shopping and technological advancements. Objective: This article aims to analyse existing literature, identify trends and highlight gaps in LML research. Methods: A systematic approach was used for the selection of studies and then bibliometric analysis was used to map and evaluate relevant studies. The Scopus citation database was used to gather peer-reviewed articles from 1999 to 2023. R-Studio and VOS Viewer were employed for data analysis and visualization. The initial search yielded 3,898 articles, which were filtered down to 1,606 articles. Key findings include the identification of the most productive journals, top-cited articles and influential keywords in LML research. Results: The study significantly emphasizes variables such as distance, costs, energy consumption and delivery time in logistics optimization. E-commerce emerged as a predominant topic, with substantial discussions on technology and transportation within LML. The research underscores the critical role of LML in the success of e-commerce. Conclusion: The findings highlight the need for ongoing innovation and optimization in delivery logistics to enhance efficiency and meet consumer demands. Future research should address the gaps identified, particularly in the context of emerging technologies such as autonomous vehicles and drones, to further improve last mile delivery systems.

Business intelligence (BI) is a crucial tool for organizations to gain a competitive advantage on the market. BI encompasses collection, analysis and utilization of data to enhance decision-making and drive organizational innovation. The system quality or management support most often influences the success of BI projects. The present article aims to verify the importance of management support in implementing BI solutions. The research was conducted in a global shipping company. The research is based on an illustrative case study in which four managerial tactics were tested. A different form of managerial support was applied to the four newly introduced reports provided by BI tools. The results confirm the importance of management support but also show the impact of other factors influencing user behaviour. Keeping the possibility of using the original data sources played a significant role. Thus, the habit effect manifested itself, as it strongly accompanies all efforts for process changes. So far, business intelligence tools are becoming a part of decision-making processes on an operational basis, but management support is still essential at all job levels.

Background: The use of artificial intelligence (AI) in education opens new possibilities for personalized learning. AI-driven systems allow students to progress at their own space, receive real-time feedback and have learning materials adapted to their individual needs. However, questions remain regarding studentsʹ perceptions of this approach and its effectiveness compared to traditional teaching methods.Objective: This study aimed to analyse university studentsʹ attitudes and preferences towards AI-driven personalized learning and identify key factors influencing its effectiveness and adoption.Methods: A mixed-method approach was employed, combining quantitative and qualitative data collection through a questionnaire survey conducted among students at the University of Ostrava. The data were collected in two phases during the winter semesters of 2023 and 2024, involving a total of 270 respondents.Results: The findings indicate that 64.1% of students perceived AI-generated and adapted chapters as more helpful and effective than traditional study materials. The most valued aspects were content adaptability, real-time feedback and increased motivation to learn. However, 18.1% of respondents viewed AI-driven instruction as less beneficial, citing limited interactivity, a lack of detailed feedback and insufficient customization for advanced learners as the main drawbacks.Conclusion: The research confirmed that AI-driven personalized learning can offer students a range of benefits, particularly in terms of adapting instructional content to individual needs, providing immediate feedback, and enabling self-paced study. However, certain challenges remain, especially regarding limited interactivity and insufficient depth of feedback, which may negatively affect students’ acceptance of such systems. To enhance the effectiveness and broader implementation of AI in educational practice, it is essential to focus on the development of interactive features, the improvement of analytical feedback, and the thoughtful integration of AI with traditional pedagogical approaches.

Background: Cross-docking optimization plays a crucial role in supply chain management by enhancing efficiency, reducing costs and streamlining operations. However, challenges arise from inaccurate data and a lack of digital tools to support decision making. Objective: The objective of this study was to integrate business intelligence (BI) tools with cross-dock operation data to optimize warehouse layout and improve decision making processes. Methods: A combination of Microsoft Visio and Microsoft Power BI was used to visualize and optimize warehouse layout based on historical cross-dock operation data. The methodology focused on integrating real-time data with spatial layout visualization to minimize total travel distance within the warehouse. Results: The integration of BI tools led to a 10% reduction in total travel distance, enhancing operational efficiency and reducing costs. The study demonstrates that BI-based decision support tools offer significant advantages over traditional optimization methods. However, challenges remain in scalability, real-time adaptability and user adoption. Conclusion: The proposed BI-driven solution improved warehouse layout optimization and facilitated data-driven decision making. Future research should explore the integration of BI with other optimization techniques and investigate its scalability in different warehouse environments.

Background: Feature selection methods play a crucial role in handling challenges such as imbalanced classes, noisy data and high dimensionality. However, existing techniques, including swarm intelligence and set theory approaches, often struggle with high-dimensional datasets due to repeated reassessment of feature selection, leading to increased processing time and computational inefficiency. Objective: This study aims to develop an enhanced incremental feature selection method that minimizes dependency on the initial dataset while improving computational efficiency. Specifically, the approach focuses on dynamic sampling and adaptive optimization to address the challenges in high-dimensional data environments. Methods: We implement a dynamic sampling approach based on rough set theory, integrating the Long-Tail Position Grey Wolf Optimizer. This method incrementally adjusts to new data samples without relying on the original dataset for feature selection, reducing variance in partitioned datasets. The performance is evaluated on benchmark datasets, comparing the proposed method to existing techniques. Results: Experimental evaluations demonstrate that the proposed method outperforms existing techniques in terms of F1 score, precision, recall and computation time. The incremental adjustment and reduced dependence on the initial data improve the overall accuracy and efficiency of feature selection in high-dimensional contexts. Conclusion: This study offers a significant advancement in feature selection methods for high-dimensional datasets. By addressing computational demands and improving accuracy, the proposed approach contributes to data science and machine learning, paving the way for more efficient and reliable feature selection processes in complex data environments. Future work may focus on extending this method to new optimization frameworks and enhancing its adaptability.

E-hailing applications are becoming popular around the globe. However, the motivations and barriers to use these applications may differ in different countries. Therefore, the aim of current work is to explore noteworthy factors affecting the acceptance of the e-hailing application ‘Careem’ in the context of Saudi Arabia. Due to recent driving permission given to women in Saudi Arabia and the lesser acceptance of public transport in Saudi Arabia, this study is of fundamental importance. To achieve the purpose of this study, the technology acceptance model is extended by adding external variables: perceived convenience, perceived accessibility and confirmation. Careem application users participated in a survey that was created and circulated to acquire measurement of the factors. The data gathered from 428 users are analysed using structural equation modelling. The findings show that perceived usability, perceived ease of use, perceived usefulness and perceived convenience are crucial and direct predictors of intention to use e-hailing applications, while confirmation and perceived accessibility are discovered to be factors that have an impact on the intention to use e-hailing applications. The findings of the study highlight the significant contributing factors in the adoption and usage of e-hailing applications. Also, these factors can assist designers and developers of these applications to develop applications having high acceptance and usage.

Background: The relationship between creativity and generative artificial intelligence (AI) is often reduced to whether a technical system can be creative or how it will transform creative industries. The study is based on the concept of implicit religion, which applies to the relationship between generative artificial intelligence and humans.Objective: The study identifies two narratives that can be used to explain the relationship between humans and technology in the theoretical field of implicit religion: (Neo-)Romanticism and Enlightenment. (Neo-)Romanticism, emphasising the value of creativity in the process rather than the outcome, can be seen as a form of implicit religion. In contrast, the Enlightenment discourse focuses on traditional AI literacy themes such as understanding the principles of use, knowledge of tools, or the ability to evaluate individual applications in social and ethical practice critically.Methods: The study uniquely combines small-scale research, which involves in-depth qualitative analysis of a small group of participants, in this case, college students, who provide reflective data from their classroom experiences and creative journals. This is complemented by a theoretical analysis of the phenomenon of implicit religion as it relates to technology. This approach allows a comprehensive exploration of the topic.Results: The results of the qualitative research show that among the students there is a part of those inclined to the (Neo-)Romantic conception of the relationship between humans and artificial intelligence. This often manifests in rejecting technology, emphasising traditional art, creativity, and the educational process. The results show that reflecting this perspective, strongly connected to emotions and values, in the education process and in complementing conventional AI literacy concepts is necessary.Conclusion: The study shows that transformation of the economy and society is not just a technical or economic phenomenon but will require a more profound philosophical and cultural reflection that will allow us to leave the binary oppositional relationship between Enlightenment and (Neo-)Romanticism.

The extant literature highlights that mobile banking offers various benefits for consumers. However, there is only a limited number of studies that investigate mobile banking adoption by the elderly. This study investigates the factors influencing the adoption of mobile banking by the elderly in a developing country context. The authors explore the enablers, barriers and perceptions of the elderly towards mobile banking adoption. Data were collected through interviews and focus group sessions with respondents from KwaZulu Natal Province in South Africa. The study relies on the Actor-Network Theory as a lens through which to understand the interrelated factors that influence the elderly’s perception and adoption of mobile banking. The results reveal a low adoption of mobile banking by the elderly. In addition, it was evident that the barriers that influence the adoption of mobile banking by the elderly include a lack of information and understanding, security and trust issues, demographic factors, language, the complexity of mobile banking applications, and resistance to change. The identified important enablers towards the adoption of mobile banking include convenience, unlimited access, cost-effectiveness. The study proposes a mobile banking adoption model for the elderly and highlights the interrelated technical and non-technical factors influencing mobile banking adoption. Additionally, it offers design guiding principles aligned to the elderly’s needs and perceptions of mobile banking.

Background: Employee turnover is a persistent issue in human resource management, leading to significant costs for organizations. This study aims to identify the most effective machine learning model for predicting employee attrition, thereby providing organizations with a reliable tool to anticipate turnover and implement proactive retention strategies. Objective: This study aims to address the challenge of employee attrition by applying machine learning techniques to provide predictive insights that can improve retention strategies. Methods: Nine machine learning algorithms are applied to a dataset of 1,470 employee records. After data preprocessing and splitting into training and test sets, the models are evaluated on metrics including accuracy, precision, recall, F1 score and AUC. Model performance is optimized through hyperparameter tuning, using grid search with cross-validation. Results: Logistic regression achieves the highest accuracy and precision, making it the top-performing model overall. Random forest provides a balanced performance with strong AUC, offering a robust alternative. Conclusion: Human resources managers and directors should consider using logistic regression or random forest for predictive modelling of employee turnover, as these models have shown strong performance. Future research should employ causal analysis for deeper insights. Real-time monitoring and adaptive prediction could also enhance models, offering a dynamic approach to attrition management.

Home automation, supported by smart devices and the internet of things, works to enhance household control. However, the reliance on current systems with fixed rules poses challenges, which can be inflexible and anxiety-provoking for users who want control over their smart home devices, limit responsiveness to changing conditions and affect energy efficiency, comfort and security. To address this, the paper proposes a dynamic personalized recommender system that considers the user's current state and contextual preferences to suggest relevant automation services for smart home devices. The system uses an unsupervised algorithm to extract rules from past interactions and supervised algorithms to make recommendations based on those rules. The proposed context-aware recommender system for smart homes achieved a remarkable average accuracy of 86.99%, a recall of 76.06% and a precision of 82.67% on publicly available datasets, surpassing previous studies. It offers users an enhanced quality of life, energy efficiency and cost reduction, while providing service providers with increased engagement and valuable insights.

The automated classification of historical document scripts holds profound implications for historians, providing unprecedented insights into the contexts of ancient manuscripts. This study introduces a robust deep learning system integrating an intelligent feature selection method, elevating the script classification process. Our methodology, applied to the CLaMM dataset, involves preprocessing steps such as advanced denoising through non-local means and binarization using Canny edge detection. These steps, pivotal for image cleaning and segmentation, set the stage for subsequent in-depth analysis. To enhance feature detection, we employ the Harris corner detector, followed by a k-means clustering process to eliminate redundancy and outliers. This process facilitates the extraction of consistently sized patches, capturing distinctive features of various scripts in historical manuscripts. The dataset undergoes rigorous training using precise convolutional neural network (CNN) models, empowering our system to discern intricate patterns and features for informed decision-making during the classification process. Ultimately, for the definitive script classification of an entire document, we employ a majority voting mechanism on the patches. The results highlight the effectiveness of this comprehensive approach, with the system achieving an impressive accuracy rate of 89.2%. This underscores the system proficiency in accurately classifying historical document scripts, offering a reliable and efficient solution for historians and researchers. The robustness of our methodology positions it as a compelling tool for meticulous analysis of historical manuscripts, contributing significantly to the field of historical document research and preservation.

With the development of modern data processing, mining and collection technologies, various companies and institutions will have more opportunities to make these data operations faster and more efficiently. From the economic perspective, processing personal data is evidently lucrative and companies would therefore like to obtain as much data as possible. This paper analyses and summarizes existing and emerging social and economic trends, implications and issues caused by clashes between European legislation on personal data protection (the GDPR) and current data processing practices. Utilizing both quantitative and qualitative data, the article attempts to scrutinize the implications of the conflict between the rising demand for privacy and personal data protection on the one hand and the ever-growing need to process and store personal data, especially by commercial organizations, on the other. Analysing databases, legislation, reports, statistical data and surveys, the paper attempts to provide an estimate of the value of personal data and the consequences of poor handling of personal data.

The integration of digital technologies into political processes represents a pivotal shift in the operational dynamics of political parties, highlighting an evolution from traditional to digital platforms. This transformation offers greater member engagement and decision-making transparency while also introducing new challenges in terms of security and control. This study presents an analytical overview of the Helios voting system as implemented by the Czech Pirate Party, used not only for primary elections, offering a critical examination of its operational framework, security measures, verification processes, and implications for intra-party democracy and member engagement. The Czech Pirate Party's engagement with Helios exemplifies a broader trend toward digitalization within political organizations, reflecting both the potential benefits and inherent challenges of such technological adoption. Drawing on theoretical insights from the literature on political party transformation in the digital age, this article assesses how i-voting systems like Helios can potentially enhance or detract from internal democracy. The discourse around digital platforms is nuanced, revealing opportunities for increased democratic engagement alongside risks of enhanced centralization and control. Empirical data from a survey of the Czech Pirate Party's membership base provides insights into perceptions of the Helios system's usability, security, and its impact on democratic processes within the party. The study further investigates Helios's technical aspects, focusing on its security protocols, verification mechanisms, and governance structures related to data and platform ownership, to understand its implications for electoral integrity and party governance. The research aims to contribute to the dialogue on e-voting platforms, highlighting the need for a careful balance between embracing technological advancements and upholding democratic principles.

Background: To objectively evaluate the capabilities of large language models (LLMs), we need to develop tools that enable such assessment. While numerous benchmarks exist, the vast majority are in English and focus on general knowledge, often overlooking the cultural and factual specifics of smaller countries.Objective: Currently, there is no benchmark that tests LLMs΄ knowledge of Slovak realia. At the same time, LLM performance in this domain remains inadequate. To objectively measure and compare these capabilities, our goal is to develop and validate a specialized benchmark for assessing LLMs΄ knowledge of Slovak cultural and factual context.Methods: We created a set of 35 questions on Slovak culture, geography, history and language. We designed them to provide unambiguous answers suitable for automated evaluation. Subsequently, we presented the questions to three major language models—DeepSeek V3, OpenAI GPT-4o and Llama 3.Results: DeepSeek scored 54% of correct answers, OpenAI GPT scored 51% and Llama scored 40%. The models scored best in geography questions. Overall scores show that models are not very good in recognising Slovak realia.Conclusion: We present the benchmark for evaluating large language models on Slovak-related knowledge. Even the most advanced current models, including OpenAI GPT and DeepSeek, answered only around half of the questions correctly. This highlights a significant gap in international LLMs΄ understanding of culturally specific facts, underscoring the need for specialized, nationally tailored language models.

Integration of the internet of things (IoT) and robotics into the internet of robotic things (IoRT) presents inherent security and trust challenges. This article introduces an innovative blockchain-centred framework designed to address these challenges. By harnessing the features of oneM2M and blockchain technology, the architecture enhances accessibility, security and data management within IoRT. Encompassing key aspects such as data flow, user authentication, health data management and security, the proposed framework ensures compliance with the oneM2M standard. A notable contribution is the integration of a private blockchain network on ZedBoard development boards, focusing on access control and data integrity verification. External IoT data are securely transmitted to a cloud platform hosted on AWS, ensuring streamlined data management with minimal impact on the resources of ZedBoards. In a stride towards bolstering IoRT security, the study implements elliptic curve digital signature algorithm (ECDSA) with Radix-2^w optimizations on ZedBoards, demonstrating a significant 58.63% reduction in addition operation costs compared to standard ECDSA. The practical implementation on ZedBoards validates the effectiveness of this approach, offering a comprehensive solution for securing IoRT networks. This research caters to industries relying on the seamless integration of IoT and robotics, concluding with noteworthy findings and suggesting potential avenues for future research, including optimizations in hardware/software co-design for ECDSA on Zynq boards.

The main goal of the paper is to identify the causes of non-realization of project prediction and to propose a new framework for project prediction. A secondary goal is to explain why the approaches to project prediction used currently do not provide satisfactory results. The research was realised in the form of qualitative research using semi-structured interviews. The findings reveal that the main causes of non-realization of project prediction are follows: there is no methodology that could be practically used; simplified approaches to project prediction usually have low reliability for which reason they are generally unusable; suitable input data and information for project prediction are not available. The main contribution made by the paper is the identification of causes of non-realization of project prediction and the proposal of a new framework for project prediction that respects changing conditions during the lifecycle of the project and changes in the way of thinking in project prediction. A prerequisite for its application is a functioning system of knowledge management in projects, including the realization of post-project analysis.

Background: Large language models (LLMs) excel at various tasks but often encounter difficulties when extended reasoning requires maintaining a consistent internal state. Identifying the threshold at which these systems fail under increasing task complexity is essential for reliable deployment. Objective: The primary objective was to examine whether four LLMs (GPT 3.5, GPT 4, GPT 4o-mini and GPT 4o) could preserve a hidden number and its arithmetic transformation across multiple yes/no queries and to determine whether a specific point of reasoning breakdown exists. Methods: A modified “Think a Number” game was employed, with complexity defined by the number of sequential yes/no queries (ranging from 1 to 9 or 11). Seven prompting strategies, including chain-of-thought variants, counterfactual prompts and few-shot examples, were evaluated. Each outcome was considered correct if the revealed number and transformation of the model remained consistent with prior answers. Results: Analysis of tens of thousands of trials showed no distinct performance cliff up to 9–11 queries, indicating that modern LLMs are more capable of consecutive reasoning than previously assumed. Counterfactual and certain chain-of-thought prompts outperformed simpler baselines. GPT 4o and GPT 4o-mini attained higher overall correctness, whereas GPT 3.5 and GPT 4 more often displayed contradictory or premature disclosures. Conclusion: In a controlled, scalable reasoning scenario, these LLMs demonstrated notable resilience to multi-step prompts. Both prompt design and model selection significantly influenced performance. Further research involving more intricate tasks and higher query counts is recommended to delineate the upper boundaries of LLM internal consistency.

The World Wide Web (the Web) has become part of people’s daily lives. Although the Web, like the Internet itself, was designed as a decentralized network, hand in hand with the increase in its interactivity Web users gradually concentrated on a limited number of platforms. As a result, providers of these large international platforms have become centres of power that can easily influence users’ behaviour and what information they can access. This paper is based on an integrative literature review and its aim is to describe the development of the Web from its beginnings to the present. This development is viewed from the perspective of centralization of the Web and the reactions that this centralization has provoked, especially the current trends towards the so-called re-decentralized Web. More specifically, the paper focuses on the privacy imperative that might act as a driving force for the re-decentralized Web, and on the technological innovations enabling development of truly decentralized platforms and applications. This paper contributes to the discussion of implications that a wider adoption of the re-decentralized Web could bring in the near future.

Background: Photovoltaic (PV)-based energy harvesting systems are crucial for ensuring the sustainability and long-term operation of wireless sensor networks (WSNs), especially in remote or infrastructure-less environments. Given the critical role of battery performance in WSN reliability, efficient energy management through Maximum Power Point Tracking (MPPT) algorithms is essential to adapt to variable environmental conditions such as solar irradiance and ambient temperature.Objective: This study aims to comparatively assess the performance of four widely adopted MPPT algorithms—Perturb and Observe (P&O), Incremental Conductance (IC), Fuzzy Logic (FL), and Particle Swarm Optimization (PSO)—in enhancing battery charging efficiency in PV-powered WSNs under dynamic environmental conditions.Methods: A simulation-based evaluation framework was developed using MATLAB/Simulink to model a PV-powered WSN system. Each MPPT algorithm was implemented and tested using the same simulation conditions, with key performance metrics including voltage and current overshoot, response time, energy transfer efficiency, and adaptability to fluctuating irradiance and temperature profiles. A Proportional-Integral (PI) controller was also used to manage the battery charging process, and environmental profiles were varied across simulation periods to assess algorithm robustness.Results: The PSO algorithm achieved superior performance across all metrics, demonstrating the fastest response time (0.1 s), lowest overshoot (14.8 V, 25 mA), and highest energy transfer efficiency. IC and FL methods showed balanced adaptability and performance, while P&O lagged in both responsiveness and efficiency. The simulation results also confirmed that environmental conditions significantly affect PV panel output and battery State of Charge (SoC), highlighting the necessity for adaptive MPPT solutions.Conclusion: This study provides a unified and realistic comparative analysis of major MPPT algorithms for PV-powered WSNs. The PSO algorithm emerges as the most effective, though its computational complexity may limit its application in low-power systems. IC and FL serve as promising alternatives for scenarios with resource constraints. The findings contribute to the design of environmentally adaptive and energy-efficient WSNs, paving the way for their robust deployment in real-world settings.

The editorial summarises the development of the Acta Informatica Pragensia journal over the last three years and complements the journal statistics for the years 2019–2025. Thanks to the indexing of the journal in Web of Science and Scopus citation databases, the world's most prestigious scientific citation databases, the journal began to profile itself as international with regional roots and a core community of Editorial Board members from Central Europe. The paper also presents the journal metrics and statistics of submitted and accepted articles for the observed period. Against the background of the current development of tools based on generative artificial intelligence, the perspectives presented in selected articles previously published in Acta Informatica Pragensia are discussed in the context of current and future directions of academic publishing. Finally, unfair practices of authors that I have encountered in our journal as Editor-in-Chief are presented and some others that are currently resonating in academic communities are also listed.

In recent years, as technology has advanced and more and more activities have become digitized, cybersecurity has become a top priority for governments around the world. Cybersecurity is essential for protecting computer systems, networks and data from cyberattacks that can have a negative impact on individuals, businesses and governments. Indeed, biometrics is a key means of cybersecurity that can help prevent unauthorized access, identity theft and unauthorized changes to data. This paper presents an innovative contactless palmprint recognition system, integrating two types of features to enhance accuracy and efficiency. Our approach employs two distinct feature sets: handcrafted features, based on the Pyramid Histogram of Oriented Gradients (PHOG) and Local Phase Quantization (LPQ) techniques and deep features extracted through deep learning-based image analysis methods such as DCTNet, DSTNet, PCANet and ICANet. Furthermore, we used a sophisticated Deep Rule-based (DRB) classifier for classification tasks. Experimental results obtained using a typical database demonstrated excellent identification rates, surpassing significantly those reported in similar studies.

Background: Text steganography preserves the privacy of secret messages by hiding them in cover text. However, existing text steganography techniques embed messages by introducing distortions in text, reducing the similarity between the cover and stegotext. Objective: The objective of this study was to design a method that increases the number of embedding choices and locations to hide more secret bits per distortion in the cover text. The goal is to enhance both embedding capacity and imperceptibility.Methods: A text steganography method is proposed that uses eight zero-width characters (ZWCs) to embed secret messages in the cover text. The proposed method also treats every character in the cover text as a potential embedding location. With eight embedding choices and bit encoding based on embedding locations, more bits can be hidden with fewer insertions in cover text.Results: Experimental results confirm that the proposed method embeds a greater number of bits per insertion of ZWC in the cover text. It also requires a smaller number of insertions to embed secret messages of comparable length. Consequently, the proposed method achieves higher embedding capacity and better imperceptibility compared to existing text steganography methods.Conclusion: The proposed method presents a substantial improvement in text steganography by increasing embedding capacity per distortion and preserving high similarity between cover and stegotext, thus enabling more secure covert communication.

Cloud computing technologies have come a long way and are available to virtually any company today. However, which factors will cause the company to decide to implement these services? Based on existing research abroad, we compiled a Technology-Organization-Environment (TOE) framework and proposed questions that support individual factors in our model to address this problem. Small and medium-sized enterprises (SMEs) in the Czech Republic actively participated in the research, from which we received 99 valid responses. Our results show a significant influence of four factors. The first factor is relative advantage, and the second is competitive pressure. In our case, companies are convinced that thanks to cloud computing, they will gain a more advantageous position over competitors, especially in the area of costs, increased productivity and entry into new industries. At the same time, they are convinced that competing cloud computing companies are implementing and taking advantage of it. The third factor is compatibility. This factor may be the cause of the temporary expansion of only simple implementations. The fourth factor is industry. So, companies perceive pressure to implement cloud computing in their business area. To support the further expansion of cloud computing, it is necessary to continue highlighting the cost benefits of cloud computing. At the same time, it certainly makes sense to bring new applications with a simple billing model and simple integration between the most used applications.

The objectives of this study are to analyse the content of publications focused on the area of information ethics and discover patterns, knowledge and thematic trends. The main research question is: What is the intellectual and topical structure of the field of information ethics? We apply bibliometric analytical methods, including co-citation analysis (41 most cited authors out of 9947), co-word analysis (127 keywords), visualizations (maps) and analysis of time periods in strategic diagrams. These methods are interpreted with the use of previous content analyses and results of a Delphi study. The dataset covers publications between 1988 and 2023 collected from Web of Science using the search term “information ethics” in titles, keywords and abstracts (469 records). The study presents the research background and objectives, related research review, research methods and findings. Results are visualized in maps of topics and trends. We investigate the intellectual and thematic structure of information ethics, including numbers of publications, main disciplines, the intellectual structure (authors, topics, trends) and identify four time periods (1988-2005, 2006-2012, 2013-2019, 2020-2023) visualized by strategic diagrams. The study reveals the multidimensionality and multidisciplinary dynamic evolution of information ethics. The main trends are the topics of ethics of artificial intelligence and algorithms, data ethics, ethics of information literacy, informational privacy and dis/misinformation. We find that information ethics studies are embedded in wider contexts of the information crisis and design of public digital services. We propose education and information literacy courses related to ethical sensitivity, data ethics and the use of AI tools. The study contributes to bridging the gap between information ethics studies and human information interactions. Our results confirm the increasing interest in ethics of artificial intelligence.

Over the past few decades, the enormous expansion of medical data has led to searching for ways of data analysis in smart healthcare systems. Acquisition of data from pictures, archives, communication systems, electronic health records, online documents, radiology reports and clinical records of different styles with specific numerical information has given rise to the concept of multimodality and the need for machine learning and deep learning techniques in the analysis of the healthcare system. Medical data play a vital role in medical education and diagnosis; determining dependency between distinct modalities is essential. This paper gives a gist of current radiology medical data analysis techniques and their various approaches and frameworks for representation and classification. A brief outline of the existing medical multimodal data processing work is presented. The main objective of this study is to spot gaps in the surveyed area and list future tasks and challenges in radiology. The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (or PRISMA) guidelines were incorporated in this study for effective article search and to investigate several relevant scientific publications. The systematic review was carried out on multimodal medical data analysis and highlighted advantages, limitations and strategies. The inherent benefit of multimodality in the medical domain powered with artificial intelligence has a significant impact on the performance of the disease diagnosis frameworks.

Breast cancer, a pervasive global health concern, necessitates early detection for an improved prognosis. Mammography, a pivotal screening tool, faces challenges in interpretation, motivating the integration of advanced computational models. This paper offers a comprehensive examination of breast cancer classification through mammography, focusing on machine learning (ML) and deep learning (DL) approaches. The discussion encompasses widely used mammography datasets, preprocessing techniques, data augmentation and diverse classification algorithms. Noteworthy datasets include LAMIS-DMDB, EMBED and INbreast. Preprocessing involves denoising and contrast enhancement, employing techniques such as Wiener filtering and histogram equalization. Data augmentation, a critical factor in handling small datasets, is explored using basic and advanced techniques, including generative adversarial networks. ML algorithms analyse entire mammograms, while DL techniques, notably convolutional neural networks, focus on localized regions of interest. Despite promising strides, challenges persist in obtaining high-quality datasets and ensuring model interpretability, as well as the strong similarities between cancer and non-cancer regions and irrelevant feature extraction. The paper concludes by outlining potential research directions to further transform breast cancer prognosis and treatment.