Volume 7.
Issue 12.

The rapid growth of digital payment systems has significantly increased the sophistication and frequency of credit card fraud, making accurate and interpretable fraud detection models a critical requirement for financial institutions. While advanced machine learning algorithms have demonstrated high predictive performance, their limited interpretability often restricts adoption in highly regulated financial environments where transparent decision-making is essential. This paper proposes an Explainable Artificial Intelligence (XAI) framework that integrates hybrid machine learning models with explainability techniques to improve fraud detection while maintaining regulatory transparency. The proposed framework combines supervised learning algorithms, including Random Forest, Gradient Boosting, and Extreme Gradient Boosting (XGBoost), with feature attribution methods such as SHAP and LIME to provide interpretable explanations for fraud predictions. In addition, data preprocessing techniques addressing class imbalance and feature engineering are incorporated to enhance predictive performance. Experimental evaluation on publicly available credit card transaction datasets demonstrates that the hybrid framework achieves improved detection capability while enabling analysts to understand the key factors influencing model decisions. The proposed approach supports compliance with emerging regulatory expectations for explainable artificial intelligence in financial services and assists fraud investigators in making informed operational decisions. The framework contributes toward building trustworthy AI systems capable of balancing prediction accuracy with transparency in modern digital payment ecosystems.

Background
Chrononutrition, the strategic alignment of eating patterns with the body’s biological rhythms, has gained increasing attention as a potential means of improving metabolic health. Time-restricted eating (TRE), a key chrononutrition approach, may help regulate metabolism, yet its effectiveness in older adults remains uncertain. This systematic review and meta-analysis examined the effects of TRE and related chrononutrition interventions on metabolic health biomarkers among older adults aged 60 years and above. The primary outcomes assessed included body mass index (BMI), fasting glucose, systolic and diastolic blood pressure (SBP, DBP), total cholesterol, triglycerides, high-density lipoprotein (HDL), and low-density lipoprotein (LDL).

Methods
A comprehensive literature search was conducted across major databases (PubMed, MEDLINE, CINAHL, and Scopus) to identify randomised controlled trials and pre–post intervention studies published between 2015 and 2025. Study quality was appraised using Cochrane risk-of-bias tools. Random-effects meta-analyses were performed, and pooled effect sizes (Cohen’s d) with 95% confidence intervals were calculated.

Results
Seven studies, including both RCTs and pre–post designs, comprising a total of 1,086 participants, met the inclusion criteria. Pooled analyses indicated small, non-significant reductions in BMI (Cohen’s d = –0.10, 95% CI: –0.24 to 0.03, p = 0.14) and fasting glucose (Cohen’s d = –0.18, 95% CI: –0.61 to 0.25, p = 0.41). No statistically significant changes were observed for SBP, DBP, total cholesterol, triglycerides, HDL, or LDL (all p > 0.05). Heterogeneity across studies was minimal, and findings were consistent between RCTs and pre–post analyses.

Conclusion
Time-restricted eating appears to be a safe and feasible dietary approach for older adults; however, it does not produce statistically significant improvements in key metabolic biomarkers compared to control or baseline conditions. The modest effects observed may reflect the complex interplay of biological, behavioural, and sociocultural factors influencing chrononutrition in ageing populations. More diverse and methodologically robust studies are needed to clarify TRE’s role in promoting metabolic health and reducing chronic disease risks among older adults.

Hybrid cloud architectures have emerged as one of the most consequential paradigms in contemporary enterprise computing, driven by the dual imperatives of scalability and control. Organizations increasingly seek to integrate on-premises data infrastructures with elastic public cloud resources in order to balance regulatory compliance, cost efficiency, performance, and innovation. Within this evolving technological ecosystem, Amazon Redshift has assumed a critical role as a fully managed, cloud-native data warehouse that enables large-scale analytics, real-time data ingestion, and complex query processing across heterogeneous data environments. The present study develops a theoretically grounded and empirically informed examination of how Amazon Redshift functions within hybrid cloud architectures, focusing on architectural design, workload management, data integration, and performance optimization.

Drawing extensively on the comprehensive treatment of Redshift in the Amazon Redshift Cookbook by Worlikar, Patel, and Challa (2025), this research situates Redshift not merely as a standalone cloud service but as a foundational element in modern hybrid data platforms. The analysis integrates perspectives from cloud computing theory, column-oriented database design, and enterprise information systems to demonstrate how Redshift mediates between traditional data warehouse models and the distributed, elastic logic of cloud infrastructures. By synthesizing insights from academic literature, vendor documentation, and comparative studies of cloud platforms, this article articulates a nuanced understanding of how hybrid cloud strategies are operationalized through Redshift-based architectures.

The discussion advances the argument that hybrid cloud data warehousing represents a transitional but durable configuration in the evolution of enterprise analytics. While full cloud migration remains a strategic goal for many organizations, regulatory constraints, legacy investments, and performance considerations ensure that hybrid models will persist. Amazon Redshift, when embedded within a hybrid cloud framework, becomes a socio-technical mediator that aligns technical capabilities with organizational strategy. This study concludes that understanding Redshift in hybrid contexts requires moving beyond product-centric evaluation toward a broader theory of infrastructural integration in the cloud era.

The article examines how modern AI pipelines automate first-pass résumé screening for technology roles while preserving recruiter control and legal defensibility. Reproducible evidence from recent peer-reviewed and preprint studies shows that structure-aware parsing stabilizes downstream extraction, domain-adapted encoders raise retrieval quality over keyword ATS baselines, and multilingual, taxonomy-anchored skill extraction reduces manual curation for heterogeneous applicant pools. Governance syntheses specify safeguards—audits, calibrated thresholds, span provenance, and contestability—suitable for large employers. An overlay architecture is derived that integrates with existing ATS workflows without occluding the résumé, uses embedding-first retrieval with lightweight re-ranking, and logs interpretable artifacts for audits. Industry benchmarks on time-to-fill, cost-per-hire, and screening effort motivate the intervention and calibrate expected cycle-time reductions and quality-of-hire gains for U.S. tech hiring. The contribution is a consolidated, deployment-oriented blueprint that connects empirical gains in ranking and extraction with organizational guardrails and recruiter-facing design choices.

This study analyzes the Titanic escape room project as a paradigmatic instance demonstrating a viable resolution to this challenge. Amid the rapid expansion of the experience economy and the market for location-based entertainment (LBE), the creation of high-tech immersive spaces, such as escape rooms, presents a multifaceted challenge in integrating narrative, technology, and safety. Using a single case-study methodology, the work deconstructs the project’s architecture through the prism of three theoretical constructs: the disappearing computer (calm technology) from human–computer interaction (HCI), human-in-the-loop (HITL) systems, and multilevel safety models adopted from human–robot interaction (HRI). Results indicate that seamless immersion is achieved via technological invisibility, wherein a sophisticated show-control network functions as a dynamic narrative engine. Interactivity is ensured via HITL models that actively treat the player as a system component. Narrative-directed perceived safety, separated from absolute physical safety, ensures the safety of physical effects using a multilayered architecture. The article proposes an integrated conceptual model synthesizing these approaches, discussing how it is practically meaningful when someone designs and evaluates complex interactive entertainment systems. Researchers and practitioners in engaging design, control, and human-computer interaction who design or analyze complex interactive spaces will find this paper helpful.

The article examines professional congruence as an integrative quality of future teachers and proposes a project-based pedagogical model for its purposeful formation. Unlike an approach that treats congruence only as correspondence between inner experience and external behavior, the study considers it through interconnected communicative, emotional-regulatory, behavioral, and reflective components. The methodological framework combines project tasks, pedagogical cases, role-playing, reflective journals, psychological training, observation, questionnaires, and video analysis. The study argues that project activity is particularly valuable because it places pre-service teachers in situations that require them to coordinate personal position, verbal expression, emotional response, and professional action. The reported experimental results indicate positive dynamics in communicative congruence, emotional regulation, behavioral coherence, communicative confidence, and reflective analysis. On this basis, a staged model is proposed for diagnosing, developing, practicing, reflecting on, and assessing professional congruence. The model may be used to strengthen practice-oriented teacher education and to support the formation of an authentic professional identity.

The article presents a comprehensive analysis of the role of artificial intelligence in optimizing operational processes and managing port logistics. The study is conducted within a theoretical and analytical framework that integrates the concepts of digital transformation, intelligent transport systems, and sustainable supply chain management. The analysis is based on recent research focusing on the application of machine learning, neural networks, and digital twins to predict vessel dwell times, container availability, and enhance the efficiency of port operations. Particular attention is given to the socio-economic effects of AI adoption in the maritime sector, including the reduction of manual labor, transformation of professional roles, and the growing need for investment in human capital. The paper summarizes key areas of AI implementation in logistics, including predictive maintenance, intelligent navigation, crane automation, and digital safety systems. The novelty of the study lies in viewing artificial intelligence not only as a technological tool but also as a strategic mechanism for building a sustainable, adaptive, and socially responsible model of port management. The findings of the study may be useful for researchers in transport analytics, digital logistics professionals, port terminal managers, and developers of intelligent management systems.

The contemporary evolution of data-intensive applications is increasingly shaped by the convergence of microservice architectures, event-driven computing paradigms, and serverless execution models. Within this transformation, data warehousing platforms are no longer passive repositories of historical data but have become active, real-time analytical backbones that mediate operational decision-making, organizational learning, and algorithmic automation. This article develops a comprehensive theoretical and empirical synthesis of how event-driven microservices and serverless infrastructures reshape modern cloud data warehousing, with particular attention to performance, scalability, architectural governance, and technical debt. Drawing upon established architectural theory, empirical studies of microservice event management, and platform-specific design patterns, the analysis demonstrates that the shift toward asynchronous, event-centric data pipelines introduces both unprecedented analytical agility and new forms of architectural fragility.

The study is grounded in the architectural patterns and operational recipes articulated for Amazon Redshift-based warehousing environments, which illustrate how distributed compute, decoupled ingestion, and materialized analytical views can be orchestrated into a coherent analytical ecosystem (Worlikar et al., 2025). These platform-specific insights are integrated with broader research on event-driven microservices, including performance trade-offs, failure propagation, and operational complexity (Cabane & Farias, 2024; Laigner et al., 2024; Chavan, 2021). Serverless computing literature further contextualizes these transformations by highlighting the socio-technical consequences of ephemeral execution, cost elasticity, and infrastructural abstraction (Baldini et al., 2017; Castro et al., 2019; Hellerstein et al., 2018).

By synthesizing these insights, the article contributes a theoretically grounded and practically relevant framework for understanding how modern data warehousing can be designed to support both real-time analytics and long-term organizational learning. The implications extend beyond technical optimization, suggesting that the future of data warehousing is inseparable from broader questions of organizational governance, software sustainability, and the political economy of cloud computing.

Under the conditions of rapidly accelerating digital transformation and the intensification of migration processes in 2024–2025, foreign language education is undergoing not merely an evolution but a qualitative paradigm shift: from the dominance of a narrowly understood communicative competence to the priority of comprehensive intercultural and existential literacy. In this study, the method of creative interpretation of stage-by-stage learning (МТИПО) is theoretically substantiated and empirically tested; it represents an integrative model for mastering language, speech patterns, cultural codes, and folk traditions in their mutual interdependence. In contrast to classical linguodidactic approaches, which often artificially separate linguistic and cultural components, the proposed model interprets folklore and tradition not as a set of fixed knowledge to be mechanically assimilated, but as a dynamic, open semantic field that serves as material for creative transformation and personal appropriation by learners. Within the framework of the study, a three-component stage-by-stage learning model was implemented: a perceptive-cognitive stage, at which culturally significant meanings of folklore texts and practices are decoded; a reproductive-reconstructive stage, which presupposes the internalization of typical speech and behavioural patterns; a stage of creative interpretation, associated with the construction of new cultural artefacts through digital storytelling and the use of generative AI technologies. The data obtained on the basis of standardized language competence tests, the Cultural Intelligence Scale (CQS), and qualitative analysis of digital projects demonstrate a statistically significant advantage of the experimental group. This confirms the hypothesis that the creative interpretation of traditions, in their conjunction with the contemporary technological context, contributes to a deep integration of the learner’s linguistic and cultural identity, making the process of foreign language education existentially meaningful and sustainable.

The gasoline blending station is a key unit at an oil refinery, ensuring the production of finished products compliant with modern technical specifications. The purpose of this study is a comprehensive analysis of the technological process and layout of a gasoline blending station operating at an oil refinery in Uzbekistan. The investigation covers the feedstock base, equipment, sequence of operations, automated process control system, and features of enhanced fuel production. The process is based on the compounding of three components-light naphtha, reformate, and butane fraction-followed by the addition of specialized detergent additives to produce QuWatt-branded fuel. A critical factor ensuring consistent product quality is the adjustment of the blending recipe based on real-time laboratory data on the properties of the feedstock components. The analysis confirms a high level of reliability and safety of the unit, achieved through a multi-level system of alarms, interlocks, and automatic regulation, enabling effective control of the complex technological process and the production of gasoline grades strictly in accordance with national standards.

The study presents a multilevel neurocognitive analysis of eyelash extension practice, treating it as a high-level psychomotor skill comparable in organizational and sensorimotor complexity to microsurgical manipulations and performance disciplines. The research introduces an original interdisciplinary framework integrating knowledge from cognitive neuroscience, the psychology of motor learning, materials science, and industry safety regulations. The author’s hypothesis is formulated as the claim that progression to an expert trajectory is determined by the synergy of two complementary components: cognitive precision, formed through sequential stages of skill acquisition and supported by neural networks of sustained attention and executive control, and sensory stability, ensured by reducing external (hygienic and environmental) and internal (chemical effects of the formulations used) sources of sensory noise. To empirically close the theoretical model, an analysis of innovative techniques and materials was conducted using the H-Lashes case, including the Clean Reputation protocol, the H-Lashes Powder formulation, and the Dual Lash Balance method. The findings have applied significance for reconfiguring educational trajectories, designing new product solutions, and strengthening the professional status of specialists in the beauty industry.