Search Results (4,969)

The pronounced heterogeneity of high-Andean soils constitutes a critical constraint to the sustainable productivity of quinoa in southern Peru, where current yields (1.6 t ha⁻¹) remain well below potential (>5 t ha⁻¹). This study aimed to develop a spatially predictive model of a weighted soil quality index (SQIw), the edaphic supply of nitrogen (N), phosphorus (P) and potassium (K), and the agricultural gypsum requirement by integrating edaphoclimatic covariates through regression–kriging. A total of 198 quinoa-cultivated soil samples were analysed; a minimum data set (MDS) was defined using correlation and principal component analyses, and regression–kriging was applied to map SQIw and the variables of interest. The MDS comprised electrical conductivity (EC), organic matter (OM), available P, exchangeable Na, sand, clay, and effective cation exchange capacity (ECEC); exchangeable Na (Wi = 0.160) and available P (Wi = 0.158) received the largest weights in the SQIw. SQIw values ranged from 0.22 to 0.84 and supported a five-class soil quality taxonomy; spatial modelling revealed a dominance of moderate-quality soils across the territory (85.21% of the agricultural area, 13,461.19 ha). The model achieved R² = 0.56, RMSE = 0.05, and MAE = 0.04 for SQIw. Most of the area (12,175.65 ha; 77%) exhibited an intermediate gypsum requirement (9.73–14.33 t ha⁻¹). Nitrogen and phosphorus showed the greatest territorial limitations, whereas potassium was largely non-limiting (84.82–570.17 kg ha⁻¹). These results indicate that sodicity and N–P deficiencies are the primary functional constraints; the generated maps enable prioritisation of gypsum amendments and targeted variable-rate fertilisation strategies to optimise the sustainability of quinoa production in the Altiplano. Full article

Between 1993 and 2024, a total of 274 hard ticks (Ixodidae) were collected from 138 subterranean localities in Croatia. This study represents the most extensive survey of hard tick fauna in subterranean habitats in Croatia to date. The collected specimens were classified into three genera and seven taxa, including two taxa that could not be identified to the species level (one from the genus Ixodes and one from Haemaphysalis). The genus Ixodes was the most abundant, comprising five taxa, whereas Haemaphysalis and Hyalomma were each represented by a single taxon. The highest diversity of hard ticks was recorded in subterranean habitats in Dalmatia, followed by north-western Croatia and Slavonia. Ixodes vespertilionis Koch, 1844 was the dominant species in the collected sample, representing 81.0% of all specimens, and was recorded in all studied regions. This species was present throughout the entire year, whereas I. hexagonus Leach, 1815 was recorded during nine months, I. frontalis (Panzer, 1798) during four months, and the remaining taxa during shorter periods. The largest number of I. vespertilionis specimens was collected in spring (33.2%), while the lowest number was recorded in winter (16.6%). The record of I. frontalis represents the first documented occurrence of this species in subterranean habitats in Croatia. Full article

Semi-arid territories in Latin America face chronic water stress; limited observability and fragmented institutions constrain effective water resources management (WRM). This narrative review synthesizes peer-reviewed evidence (2020–2026) on smart technologies that strengthen basin- and utility-level WRM, using Peru (Piura-like coastal semi-arid contexts) as an anchor and Latin America as a comparative lens. We used a structured, traceable database-based workflow and synthesized studies reporting measurable outcomes across five application categories: drought/flood early warning, hydrometeorological forecasting, water quality surveillance, non-revenue water (NRW)/leakage, and allocation and compliance. Findings were organized into an application-oriented taxonomy spanning remote sensing (RS) and GIS, Internet of Things (IoT)/telemetry, analytics/AI-enabled decision support, and hybrid approaches. Evidence most consistently reports operational gains (coverage, timeliness, predictive performance), while governance outcomes are less frequently measured and appear contingent on interoperability, digital capacity, and sustainable operations and maintenance (O&M) conditions. We conclude with a territorial adoption agenda specifying minimum enabling conditions and a phased pathway from pilots to scalable, eco-efficient smart WRM in Peru and comparable semi-arid settings across Latin America. Full article

The rapid transition to Industry 4.0/5.0 has widened the gap between graduates’ skill sets and labor market expectations; this study aimed to profile student competencies and align academic pathways with inclusive and adaptive AI-driven learning. A quantitative design was applied: an online survey of n = 126 students (engineering and economics, February–March 2025), expert evaluations from 5 faculty and 5 employers on a 5-point scale, framed by T-shaped competencies, 4C skills, and Bloom’s taxonomy. Analysis was performed in Python 3.11; future demand until 2035 was forecasted using ARIMA and Prophet models trained on publicly available labor market data (OECD, WEF, Eurostat 2015–2024); competency prioritization employed K-Means clustering and Random Forest models. Strengths included cooperation 4.2, critical thinking 3.9, communication 3.8, and creativity 3.6. Deficits were programming 2.8, project management 3.2, and solution development 3.2; employers rated programming at 2.5 (−0.7 compared to faculty). Forecast 2025–2035 showed growth in demand for programming +56% (3.2 → 5.0), data analytics +39% (3.6 → 5.0), project management +34% (3.2 → 4.3), digital literacy +30% (3.7 → 4.8), and critical thinking +15% (3.9 → 4.5). Clustering identified critical (programming, analytics, project management), supporting (creativity, communication, teamwork), and optional (narrow theoretical depth) competencies. Curriculum adjustment with practice-oriented modules, AI-enabled adaptive learning, and systematic university–employer feedback is essential; the proposed AI-supported profiling model is scalable and enhances inclusiveness. Full article

Machine Learning (ML) models trained on large-scale datasets learn useful predictive patterns, but they may also memorize undesired information, leading to risks such as information leakage, bias, copyright violations, and privacy attacks. As these models are increasingly deployed in real-world and regulated settings, the consequences of such memorization become practical and high-stakes, reinforced by data-protection frameworks that grant individuals a Right to be Forgotten (e.g., the GDPR). Simply removing a record from the training dataset does not guarantee the elimination of its influence from the model, while retrain-from-scratch procedures are often prohibitive for modern architectures, including Transformers and Large Language Models (LLMs). In this work, we provide a perspective on Machine Unlearning (MU) in supervised learning settings, with a particular focus on Natural Language Processing (NLP) scenarios, grounded in a PRISMA-driven systematic review. We propose a multi-level taxonomy that organizes MU techniques along practical and conceptual dimensions, including exactness (exact versus approximate), unlearning granularity, guarantees, and application constraints. To complement this perspective, we run an illustrative benchmark evaluation using a standardized unlearning protocol on DistilBERT trained on a public corpus of news headlines for topic classification, contrasting the retraining gold standard with representative design-for-unlearning and approximate post hoc techniques. For completeness, we also report two oracle-assisted upper-bound baselines (distillation and scrubbing) that rely on a clean retrained reference model, and we account for their incremental cost separately. Our analysis jointly considers model utility, probabilistic quality, forgetting and privacy indicators, as well as computational efficiency. The results highlight systematic trade-offs between accuracy, computational cost, and removal effectiveness, providing practical guidance for selecting machine unlearning techniques in realistic deployment scenarios. Full article

The development of critical thinking from an early age is essential in science education, and despite its importance, there is still little research in early childhood education. This exploratory study presents a learning environment around a daily life problem faced by preschoolers, related to the human body and health, carried out through role-play and inquiry, aimed at developing critical thinking within the knowledge application domain in 4-year-old children. The study involved a sample of 9 children from a preschool in Málaga (Spain). Data were collected through observations, dialogues, field notes, and children’s productions. The assessment of progress in the application of scientific knowledge and understanding of science encompassed a comprehensive set of criteria aligned with Bloom’s revised taxonomy. The findings indicate greater progress in the remembering compared to understanding. Specifically, 76.18% of the children reached the achieved level in listing, 72.21% in explaining, 62.50% in relating, and 58.33% in identifying. This suggests that, at early ages, learning environments designed around daily-life health contexts can contribute to the development of certain aspects of critical thinking. Full article

Russian-language resources for medical natural language processing (NLP) are expanding rapidly; however, their fragmentation, uneven curation, and limited clinical reliability hinder the development of safe machine learning systems for prognosis, prevention, and precision medicine. We provide the first systematic survey of Russian medical NLP datasets and analyze their suitability for clinically meaningful tasks as defined by the MedHELM taxonomy. We additionally perform expert clinical validation of three representative public corpora—RuMedPrimeData (real outpatient notes), MedSyn (synthetic clinical notes), and RuMedNLI (translated natural language inference)—assessing clinical plausibility, diagnosis accuracy, and logical consistency. Experts identified substantial reliability issues: across randomly sampled subsets of each corpus, only approximately 20% of RuMedPrimeData records, fewer than 15% of MedSyn records, and approximately 55% of RuMedNLI pairs met essential quality criteria, which can hinder downstream ML systems built on these data. To support robust applications—ranging from medical chatbots and triage assistants to predictive and preventive models—we outline practical requirements for high-quality datasets: coordinated, expert-validated, machine-readable corpora aligned with clinical guidelines and insurance logic, standardized de-identification, and transparent provenance. Strengthening these data foundations will enable the development of reliable, reproducible, and clinically relevant AI systems suitable for real-world healthcare applications. Full article

Indonesia’s rapid digital growth has been accompanied by escalating cyber threats, with public discourse on social media emerging as a critical but underutilized source of threat intelligence. This discourse is characterized by informal language and local nuances that render existing international cybercrime taxonomies ineffective, creating a gap in scalable, locally relevant threat analytics. This study introduces the Indonesian Cybercrime Threat Taxonomy (ICTT), a novel five-dimensional framework tailored to Indonesian online environments. An end-to-end OSINT pipeline was developed to collect 2344 samples from X (formerly Twitter) and YouTube, employing weak supervision with 12 high-precision regex patterns to generate training labels. A state-of-the-art IndoBERT model was fine-tuned on this data, and its performance was compared against rule-based and hybrid classification models. On a manually annotated gold-standard dataset of 600 samples, both the IndoBERT and hybrid models achieved 96.8% accuracy, significantly outperforming the rule-based baseline (66.7%). The models demonstrated strong generalization across both social media platforms, and the hybrid approach provided an effective balance of high performance and interpretability. This research demonstrates that informal public discourse can be systematically transformed into structured threat intelligence. The ICTT and the accompanying hybrid classification system provide a scalable, interpretable, and locally relevant foundation for cyber threat analytics in Indonesia, establishing a methodological blueprint for other low-resource language contexts. Full article

The increasing complexity and uncertain system of modern discrete event system (DES) challenge traditional model-based control approaches, while artificial intelligence (AI) techniques offer powerful data-driven decision-making capabilities but lack formal guarantees. This review surveys recent research on the integration of AI with DES and supervisory control theory. Following a systematic literature mapping methodology, the literature is organized using a taxonomy based on three orthogonal perspectives: control and decision paradigm, system capability and property, and application and operational objectives. The review highlights how learning-based methods enhance adaptability and performance in DES, while also exposing persistent challenges related to safety, nonblocking behavior, data efficiency, and interpretability. By structuring existing approaches and identifying open issues, this review provides a coherent overview of the current research landscape and outlines key directions for future work on AI-enabled DES. Full article

Meditation-related difficulties have been systematically documented in contemporary contemplative science, yet the prescriptive resources preserved in the ancient Buddhist commentarial literature remain underutilized in comparative research. This study analyzes the case of five hundred monks in the Paramatthajotikā I’s commentary on the Karaṇīya-metta-sutta. During intensive practice, these monks experienced complex psychosomatic symptoms—perceptual disturbances, fear, somatic distress, and cognitive impairment—and received from the Buddha an integrated prescription of five protective practices (pañca rakkhā). Through Pāli textual and comparative analysis with Lindahl et al.’s taxonomy of meditation-related difficulties, this study demonstrates that the monks’ symptoms correspond systematically to the perceptual, affective, somatic, and cognitive domains of the modern taxonomy, with the critical difference residing in interpretive frameworks rather than in the phenomena themselves. The five practices—loving-kindness meditation, protective chant recitation, contemplation of impurity, mindfulness of death, and the arousal of religious urgency—constitute a sequentially structured system progressing from the emotional reframing of fear to the deconstruction of bodily and existential attachment, culminating in the restoration of soteriological motivation. This study argues that Paramatthajotikā I’s prescriptive system provides a historically grounded, soteriologically oriented complement to contemporary contemplative science, particularly in bridging the gap between phenomenological classification and meaning-centered intervention. Full article

The genus Erysimum (Brassicaceae) comprises more than 150 species distributed mainly across Europe, Central Asia, East Asia, the Middle East, North Africa and North America, many of which are traditionally used for treating cardiovascular, respiratory, and inflammatory disorders. Plants of this genus are rich in various groups of secondary metabolites, including cardenolides, glucosinolates and isothiocyanates released from them, sterols, phenolic compounds such as flavonoids and tannins, and other secondary metabolites. This review synthesizes its unique phytochemical profile, characterized by the coexistence of ancestral glucosinolates and independently evolved cardenolides. Over 100 cardenolide structures based on 15 aglycones have been reported from Erysimum, although the structural characterization of several compounds remains inconsistent or incomplete, with some glycosides still absent in major chemical databases. A variety of pharmacological activities have been documented for extracts and isolated constituents, including cardiotonic, anti-inflammatory, antioxidant, antimicrobial, and cytotoxic effects, supporting the therapeutic potential of the genus. Ecologically, the genus employs a two-tiered defense strategy where strophanthidin-based compounds deter butterfly oviposition and digitoxigenin-based compounds repel larval feeding. This review summarizes current knowledge on the taxonomy, distribution, phytochemical composition, and biological activities of Erysimum species, with a focus on cardenolide diversity, structural ambiguities, and research gaps that require further investigation. Full article

Diffusion models have emerged as a powerful class of generative models, demonstrating impressive results across visual domains such as image and video synthesis. This survey provides a comprehensive taxonomy of generative models, with a particular focus on diffusion models and their applications in enhancing visual fidelity for text-to-image and text-to-video generation. We discuss the theoretical foundations of diffusion models, including their formulation through stochastic differential equations, and analyze the forward noising and reverse denoising processes that enable stable training and high-quality generation. The survey further categorizes diffusion architectures, including pixel-space and latent-space models, and examines their design choices, training strategies, and trade-offs across different resolution regimes. In addition, we review noise characteristics in real-world imaging domains and discuss their implications for diffusion-based models. Denoising strategies are analyzed by distinguishing between in-model denoising mechanisms and external denoising techniques used in preprocessing and post-processing pipelines. The survey also summarizes commonly used datasets and evaluation metrics for generative modeling, providing a practical perspective on benchmarking and model comparison. Finally, we discuss current challenges, including computational efficiency, scalability, and robustness to diverse noise distributions, and outline potential directions for future research. This survey aims to provide a structured reference for understanding diffusion models and their applications in visual generation tasks. Full article

Integrating Lean Construction (LC), the Circular Economy (CE), and Construction 5.0 (C5.0) remains challenging in emerging delivery contexts. This difficulty increases when procurement routines determine which practices become enforceable across tendering, contracting, and site execution. This study prioritized barriers to LCCE5.0 implementation in Morocco and translated expert judgments into actionable recommendations. A structured literature review informed the barrier inventory and conceptual framing. The study proposed a three-layer, life-cycle LCCE5.0 framework that links governance, operational routines, and digital enablers. It operationalized 40 critical barrier factors across six dimensions and five life-cycle macro-phases. A two-round Delphi study was conducted with 22 Moroccan experts using a 7-point Likert scale. Barriers were ranked using Round 2 (T2) medians with ties resolved using the interquartile range. Top-box agreement (ratings of 6–7) and consensus tiers were reported. The ranking showed strong stability across rounds, with 92.5% of barrier factors remaining stable. Kendall’s W at T2 equaled 0.817 (p < 0.001), indicating high panel consensus. Results indicated that constraints clustered in upstream governance. Three procurement-centered regulatory and contractual barriers topped the ranking (Mdn_T2 = 7). These barriers reflected missing CE procurement guidelines, limited weighting of environmental criteria, and the absence of circularity and digital requirements in tenders. Six additional barriers reinforced this procurement bottleneck. They included limited owner commitment, weak enforcement authority, limited top-management commitment, and regulatory instability. They also included low interorganizational trust, limited risk-sharing contracts, and tool-centered deployment of LCCE5.0 practices. These findings support procurement-focused recommendations to institutionalize auditable circular requirements and data-enabled verification in tendering and contracting routines. The proposed LCCE5.0 mechanism and the resulting recommendations require empirical validation beyond this Delphi-based prioritization. Full article

This study presents a systematic comparative benchmark between two distinct paradigms for solving nonlinear partial differential equations (PDEs): the data-driven Physics-Informed Neural Networks (PINNs) and the analytical Homotopy Analysis Method (HAM). We apply both methods to a unified family of canonical PDEs, the Burgers, Huxley, Fisher, Burgers–Huxley, and Burgers–Fisher equations, under identical problem setups, domain discretization, and validation metrics. PINNs incorporate physical laws directly into neural network training by minimizing a loss function that enforces PDE residuals, yielding physically consistent solutions even for strongly nonlinear problems. HAM provides approximate analytical solutions using a unified framework, and the same initial guess, auxiliary linear operator, and auxiliary function across all equations despite their distinct nonlinearities. The controlled, consistent application of both methods enables a fair, reproducible comparison across this equation family. The results provide a quantitative performance map under identical conditions, delineating when PINNs (high accuracy, long-term stability, and generalization capability) are preferable, versus when HAM (computational speed, short-term analytic approximation, and lower memory footprint) offers advantages. While the finite radius of convergence of the truncated HAM series is theoretically expected, our controlled comparison quantifies for the first time how this degradation varies across equation types, revealing that the choice between methods depends on specific problem requirements including error tolerance, available computational resources, and temporal horizon. The novelty lies not in solving each equation individually, but in deriving a performance taxonomy that systematically connects equation features (shocks, stiffness, and reaction–diffusion coupling) to optimal solver choice—providing previously unavailable, evidence-based guidance for the scientific computing community. This study establishes the first rigorous, controlled comparative benchmark between analytic and data-driven PDE solvers across a spectrum of nonlinearities, providing a reproducible baseline for future hybrid scientific machine learning solvers. Full article

Nematodes in the genus Oscheius (Rhabditidae) have traditionally been regarded as free-living bacteriophagous or necromenic associates of insects. Over the past two decades, however, multiple Oscheius species and isolates have been shown to express facultative pathogenicity toward insects and, in some cases, parasitism of mollusks. This has stimulated interest in Oscheius as a complementary group of biological control agents that may function under conditions limiting classical entomopathogenic nematodes (EPNs) of the genera Steinernema and Heterorhabditis. Here, we synthesize current knowledge on Oscheius taxonomy and diversity, life-history strategies, bacterial associations and virulence mechanisms, evidence for control of insect and mollusk pests, and recent advances in chemo-ecology relevant to host finding. We emphasize that Oscheius represents a continuum of ecological strategies, and we adopt conservative terminology in which “entomopathogenic” is reserved for Oscheius species/isolates that meet operational criteria of insect pathogenicity. Finally, we highlight key barriers to wider implementation—strain variability, bacterial partner instability, non-target and community effects, and production/quality control needs—and propose research priorities for the development of robust, field-reliable Oscheius-based biocontrol. Full article