Search Results (27,566)

Detecting deceptive statements is a complex challenge, particularly in non-English contexts where resources are often limited. This study addresses this problem by evaluating the performance of the Polish Large Language Model (LLM), Bielik-11B-v2.3-Instruct. We utilized a dataset of nearly 1500 true and false statements in Polish. In this data, labels reflect agreement with the author’s views. We assessed the model’s performance across diverse prompting variants, adapter fine-tuning, and their combination. The results demonstrate that, while adapter fine-tuning outperforms zero-shot prompting across both data modalities, the combined approach of prompting and fine-tuning is far superior. It achieves an accuracy of 0.82 on typed utterances, exceeding the previously best machine-learning result of 0.69 and human-level accuracy of 0.54. This highlights a growing disparity between humans’ limited detection capabilities and LLMs’ increasing ability to identify deceit. Full article

This study systematically examines the robustness of the Akaike Information Criterion (AIC) in determining the optimal order (p) of an autoregressive (AR) model applied to the RR interval time series of the PhysioNet healthy subject database. The AR approach is widely used to estimate the power spectral density (PSD) of heart rate variability (HRV), and accurate order selection is essential for model stability and reliable spectral estimation. Although the AIC is designed to balance model fit and complexity, it suffers from the problem of arbitrary model selection. This study provides a quantitative robustness analysis of information-criterion-based AR order selection under controlled expansion of the search space. Specifically, we investigated the behavior of the AIC using the PhysioNet database (N = 1257) under conditions where the maximum search order was set to an excessively high value (p = 50), far exceeding the commonly recommended range. Our analysis suggested that the AR model began to capture subtle noise and nonstationary components rather than the intrinsic HRV structure, leading to overfitting and excessive order selection, resulting in false peaks in the PSD and reduced robustness. In conclusion, order decisions based solely on information criteria such as the AIC become unstable when the search range is too large. To ensure robustness, it is recommended to complement the AIC with more stringent criteria such as the Bayesian Information Criterion (BIC) or Final Prediction Error (FPE), in addition to the traditional maximum order restriction. Full article

This study investigated sulphide textures and trace element chemistry from the Tomingley Gold Project (TGP) region of Central-West NSW, eastern Australia, using in situ techniques. In particular, the study focused on pyrite and arsenopyrite to gain insights into ore-forming processes and determine which trace elements within these minerals can be used as potential pathfinder elements for mineral exploration in the TGP. A total of 41 drill core samples from a variety of lithologies (volcaniclastic, monzodiorite, graphitic siltstone, dacite, andesite) were described and analysed using reflected light microscopy, high-resolution microscopy (via Scanning Electron Microscope or SEM), elemental mapping (via Electron Probe Micro Analysis or EPMA) and targeted trace element analysis of sulphide grains (via Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry or LA-ICP-MS). Findings show that pyrite and arsenopyrite are the major sulphides that host fracture-fill/inclusions of native gold and ‘invisible gold’. Pyrite rich in groundmass inclusions should be evaluated due to their characteristic high concentrations of both As and Au. Pyrite trace element chemistry (Sn, Bi, W, Sb, Au and Se) was able to delineate mineralised from unmineralised samples in volcaniclastics, graphitic siltstones and andesites but was much more challenging for lithologies like dacites and monzodiorites. The study also found that Au may have been introduced into the system earlier and existed as ‘invisible gold’ in earlier generations of pyrite. This study highlighted the utility of in situ techniques to discriminate mineralised signatures from unmineralised samples, and this has proven to be far more effective compared to whole-rock techniques, emphasising the benefits of such datasets in mineral exploration. Full article

New Testament textual criticism in the twenty-first century continues to refine principles and approaches that have been part of this scholarly discipline since the development of the modern era of textual criticism. At the same time, even as New Testament textual criticism remains connected to its roots in pursuing the (re)construction of a critical text of the New Testament, more recent approaches have expanded the set of questions being asked by textual critics to include far more than simply the words of a New Testament text. In all of these developments, there is both “Old” in the “New” and “New” in the “Old,” resulting in New Testament textual criticism in the twenty-first century having become not only one of the more vibrant fields in New Testament studies but also having captured the popular imagination. Full article

Traditional media have progressively lost electoral centrality, while social media platforms have become key arenas for political communication. Although digital campaigning has been widely studied, limited cross-platform research has examined how social media engagement relates to broader patterns of digital public attention, particularly in Southern European multi-party systems. This study analyses the digital strategies of Portuguese political parties during the 2024 Legislative Elections, drawing on an original dataset of 6251 posts and 8.5 million interactions across Facebook, Instagram, X, YouTube, and TikTok, combined with Google search trends data. The main findings show that ideologically extreme parties generate significantly higher engagement, especially the far-right. However, high engagement does not necessarily translate into broader digital attention. Televised debates remain decisive in structuring peaks of online interest, confirming the persistence of hybrid media dynamics. By integrating cross-platform engagement metrics with search data, this study demonstrates the limits of engagement as a proxy for political attention and electoral impact. Full article

Canine babesiosis is a vector-borne disease of dogs with a worldwide distribution, presenting varying clinical signs depending on the host, parasite strain and climatic factors. Beyond companionship, dog meat serves as delicacy and is also used for zootherapy in some African communities. In this study, we collated and analyzed molecular biology-based diagnostic data on Babesia species of dogs in Africa in order to elucidate the epidemiological factors of the disease on the continent. Four Babesia species—B. rossi, B. vogeli, B. gibsoni and B. canis—were detected in Africa based on the results from 40 studies that involved the screening of 9435 dog blood samples from 83 study locations. Babesia rossi was the most commonly reported (aggregate detection rate = 7.7%) and was detected in samples from all the African regions except northern Africa. Babesia vogeli was the second most commonly reported (aggregate detection = 4.8%) and was detected in all of the African sub-regions. There were few reports of B. gibsoni (0.6%) in the southern and western African regions, and a single case of B. canis in an untraveled Nigerian dog. So far, there were no reports of Babesia coco, Babesia conradae or Babesia vulpes (Babesia annae, Babesia microti-like) in any of the African countries that have been confirmed by a molecular method. This study presents a synopsis of canine babesiosis in Africa, and provides an overview of common clinical signs, etiologies and risk factors that will serve as a quick guide to veterinarians to achieve timely tentative diagnosis. Full article

This paper proposes an innovative research algorithm “measurement—pattern—driving force—synergy” that determines the eco-efficiency of 83 Russian federal subjects (2000–2019) using the Slacks-Based Measure (SBM) model with non-desired outputs (incorporating comprehensive input indicators such as water resources and electricity input, and dual non-desired outputs of waste gas and wastewater). Combined with hot spot analysis, a gravity center model, and panel Tobit regression, we reveal the temporal-spatial evolution and driving mechanisms of eco-efficiency in resource-based economies. The research finds that the overall eco-efficiency of Russia is at a medium level and shows a dynamic correlation with the economic development stage. In the early stage of the period under review, there was a high degree of synergy, but the efficiency declined during the period of rapid economic growth. Later, it rebounded somewhat in tie with technological progress. Spatially, it presents a special pattern of low efficiency in the western European industrialized regions and high efficiency in the Arctic and Far East peripheral regions, reflecting the spatial heterogeneity of resource-dependent economies and the survival-constrained efficiency feature. The analysis of influencing factors indicates that per capita GDP has a significant positive driving effect on eco-efficiency, but the expansion of residents’ consumption, the improvement of education level and the dependence on foreign trade all have inhibitory effects, highlighting the path dependence of the current growth model on the structure of resource consumption. The research suggests that Russia should implement differentiated spatial governance in the future, promote the green transformation of consumption and trade structures, and strengthen the ecological orientation of the education and scientific research system to achieve a fundamental transformation of regional sustainable development from survival constraints to innovation-driven. Full article

Applications for machine learning (ML) and deep learning (DL) are constantly growing and have already been adopted in the field of particle measurement technology. Even though analytical (ultra-)centrifugation (AC/AUC) is a widely used technique for characterizing dispersed particle systems, ML and DL have not yet been applied in this area. Data evaluation and interpretation in AC/AUC can be challenging and often requires expert knowledge. DL models can help, but their development is limited by a lack of annotated training data. One solution is to generate and use synthetic data instead. In the first part of this study, a model was trained to synthesize data from experiments using a combination of Variational Autoencoder (VAE) and Generative Adversarial Networks (GANs). The results appear highly realistic. Novice users could distinguish real from synthetic samples with only 63% accuracy. Then, a classifier was trained on experimental AC data to categorize real-world examples based on their underlying separation kinetics, testing different DL architectures. After initial training, the models were further fine-tuned with synthetic AC data. ResNet34 models achieved the best performance with 94% accuracy, comparable to an AC expert (91%), while inexperienced users reached only 53%. In the second part of our study, a regression model was trained for the analysis of sedimentation coefficients. Therefore, various generative models were developed and evaluated for synthesizing AUC data based on numerically simulated sedimentation boundaries. The best results were achieved by combining VAE and GAN architectures with embedded physical constraints. However, the generative networks have so far led to additional smearing of the profiles, resulting in a broadening of the sedimentation coefficient distribution and indicating that further refinement is necessary. Full article

Resource allocation in sixth-generation (6G) networks must meet throughput, latency, and reliability targets while network conditions keep changing. At the same time, the telemetry needed to train good models is distributed across many devices and edge nodes, so sending it to a central server can violate privacy or data-sharing constraints. Federated learning (FL) helps, but two practical concerns usually determine whether it works in practice: how much communication is needed to achieve strong performance, and whether weaker (tail) clients benefit-not only the average client. In this study, we run large-scale FL on 6G telemetry with 200 clients and quantify the communication fairness trade-off. We evaluate FedAvg and FedProx under multiple settings and benchmark them against a strong centralized model and a local-only baseline. Results are reported as mean ± 95% confidence intervals over five random seeds. We measure the accuracy, macro-F1, AUC, and AP, and we also focus on tail behavior using the worst eligible client accuracy, p10 client accuracy, and fairness gap. By plotting the accuracy/macro-F1 against cumulative communication (bytes), we show that some configurations match the average performance while transmitting far fewer data. Finally, we find that the worst client performance improves early and then stabilizes, and a sensitivity study suggests that FedProx’s $\mu$ has a limited impact in this setup. These findings offer actionable guidance for 6G operators and system designers by quantifying how participation and dropout policies translate into concrete communication budgets and tail client behavior. Full article

The accurate measurement and optimization of spatial vitality inside commercial complexes has become crucial for sophisticated urban governance as urban growth moves from rapid expansion to quality-oriented stock augmentation. This research creates a multifaceted assessment methodology that incorporates systemic connectedness (transportation synergy), spatial performance (public activity and social efficacy), and spatial supply (human–land linkages and arrangement). We used a stratified purposive sample of 20 business complexes spread across eight districts in Guangzhou, a typical high-density megacity. In order to understand the underlying mechanisms of spatial vitality, we measured important indicators including the Polycentricity Index (α) and the Spatial Performance Index (β) using a mixed-methods approach that included K-means clustering, multinomial logit regression, and Structural Equation Modeling (SEM). Four important insights are shown by our findings. 1. The paradox of density and efficiency: The notion that high-density development inevitably ensures lively public space is called into question by the lack of a significant linear correlation between the Floor Area Ratio (FAR) and spatial performance (r = 0.32, p > 0.05), despite a core–periphery gradient in development intensity. 2. Structural Supply Demand Mismatch: Although overall spatial performance is strong (β = 0.81 ± 0.07), there is a notable shortfall in cultural and artistic venues, where young adults’ demand (0.27) is 145% greater than supply (0.11). 3. Polycentric Networking vs. Transport Polarization: While spatial structures show a networked polycentric pattern (mean α = 6.40), transportation synergy is affected by core–periphery polarization, which results in “vitality islands” in the periphery. 4. Dual-Path Driving Mechanisms: According to SEM results, cultural spaces have a considerable indirect impact (39.7% mediation) by boosting brand uniqueness and “cultural capital,” while composite plaza spaces have a strong direct effect on commercial performance (γ = 0.682). Based on these findings, we suggest distinct optimization strategies: aging projects need climate-responsive design interventions; growing areas should create family-oriented consumption ecosystems; and core districts should give priority to cultural “IP” integration. For the planning and revitalization of commercial land use in high-density global environments, this study offers a solid analytical framework and practical insights. Full article

During gas production, the wellbore operates under high temperature and high pressure conditions, where gas–liquid two-phase flow is commonly present. Thus, it is important to understand the holdup of different flow patterns of gas–liquid two-phase flow in the wellbore. Currently, prediction models for flow patterns and holdup are mainly based on experimental data obtained under normal temperature and pressure, judged by researchers’ subjective observations, resulting in the lack of a unified flow pattern boundary so far. This article uses CFD simulation to study the characteristics of high temperature and high pressure gas–liquid two-phase flow in wellbores and obtains data such as flow pattern, probability density distribution of gas content, and liquid holdup. The results indicate that pressure is the dominant factor affecting the transition of gas–liquid two-phase flow patterns in wellbores. Under high temperature and pressure, each flow pattern has unique probability density distribution characteristics of gas content, which can be used as a basis for flow pattern identification. Flow pattern diagrams of gas–liquid two-phase flow at room temperature and pressure as well as high temperature and high pressure were drawn, and a boundary model for the transition of gas–liquid two-phase flow patterns in wellbores under the influence of variable pressure was established, verifying the rationality of the flow patterns’ transition boundaries. Based on the simulation of high temperature and high pressure gas–liquid two-phase flow in the wellbore, a calculation model for gas–liquid two-phase flow in the wellbore was established. The absolute percentage error between this model and the Fluent simulation results is within 10%, which can achieve the calculation of gas–liquid two-phase flow in the wellbore under different temperature and pressure conditions. Full article

Existing general image inpainting works require the user to customize a mask to indicate the region to be inpainted. However, the mask is often hard to calibrate accurately in real-world applications, e.g., graffiti removal. Blind image inpainting aims to automatically restore the degraded image into the visually reasonable one without a priori mask to indicate the area to be repaired. So far, most proposed blind inpainting methods convert the task into general inpainting by predicting the mask before inpainting. However, these methods are highly dependent on mask prediction results, which may produce inferior inpainting results if the prediction is inaccurate. To address this issue, we propose a two-stage blind inpainting framework with two novel designs: (1) cross-space contrastive learning, to remove the noise in the degraded images and realize the automatic inpainting in the latent space by reducing the distance of the degraded images and the corresponding complete images in the latent space; and (2) mask-aware adversarial training, to minimize the mutual information between the inpainted feature and the noise. Extensive experiments prove that our blind inpainting framework performs better on multiple datasets than the state-of-the-art methods. Full article

Under global climate change, frequent droughts threaten ecosystem functions, but how drought characteristics affect ecosystem resilience remains unclear. Focusing on the Dongjiang River Basin, China, we identified drought events at an 8-day scale from 2000–2024 using multi-source remote sensing and reanalysis data. The water use efficiency-based resilience index (R_de) was calculated, and a random forest model quantified the contributions of 21 potential driving factors. The model explained 68% of R_de variance (R² = 0.68, RMSE = 0.12). Downward shortwave radiation was the primary factor, followed by antecedent water use efficiency and soil moisture anomaly, with drought intensity and air temperature ranking fourth and fifth. All dominant factors exhibited nonlinear threshold effects: R_de decreased significantly after radiation exceeded ~110 W·m⁻²·(8d)⁻¹; R_de declined when standardized soil moisture anomaly fell below −2.0; and R_de increased sharply when drought intensity exceeded 12%. Drought intensity far outweighed duration and severity, establishing it as the key drought attribute. This study reveals the dominant drivers and their thresholds governing ecosystem resilience in the Dongjiang River Basin, providing quantifiable indicators for ecological drought early warning. Full article

Rapid and accurate identification of water inrush sources is essential for the prevention and control of coal mine water hazards. Fisher discriminant analysis and random forest are widely applied, but their performance comparison and applicability under single-mine and multi-mine scenarios remain to be investigated. This study takes the Tunlan Mine in Shanxi Province, China, as an example and evaluates both models using accuracy, precision, recall, F1-score, and confusion matrix. A joint discrimination scheme is used to explore their generalization ability. In the single-mine scenario, the improved Fisher algorithm achieves an overall accuracy of 93% and the random forest model achieves 87%, indicating that the former has greater advantages when data distribution is relatively linear. In the multi-mine joint discrimination scenario, the random forest model yields accuracies of 77–98%, far exceeding those of the Fisher algorithm and demonstrating clear superiority in handling complex nonlinear data. The results show that model performance depends primarily on data quality and feature distribution rather than solely on sample size. This study provides a scientific basis for selecting water source identification algorithms in different scenarios and has practical value for improving coal mine water hazard prevention and control. Full article

Background: Tourists often travel within their own country or abroad for business, leisure or to receive planned healthcare. However, they are often not prepared for unexpected medical emergencies that occur far from home. Seeking emergency healthcare during travel may pose various barriers and challenges to tourists. Aims: This systematic review aimed to identify the challenges and barriers tourists face while seeking emergency healthcare during travel. Methods: A comprehensive search was performed in PubMed, Scopus, Web of Science and ScienceDirect from 1st January 1995 to 31 October 2025. The review included studies focusing on tourists who sought emergency healthcare abroad. Due to the methodological heterogeneity of the studies making meta-analysis impossible, a narrative synthesis of the results was conducted. The review protocol was registered with PROSPERO (ID CRD420251156975). Results: From 608 initial titles (603 from database searches and 5 additional from similar articles), 10 studies were selected—5 cross-sectional and 5 retrospective. Most (7/10) were conducted in Asian countries, while others were conducted in Europe (1), the U.S.A. (1) and multiple countries (1). The participant number ranged from 37 to 2333. All studies included both genders, apart from one that focused exclusively on pregnant women. The most common challenges identified were language and cultural barriers, limited access to healthcare services in terms of appropriateness and timeliness of care and financial and insurance coverage issues. Conclusions: The findings underscore that tourists face multiple barriers when seeking emergency healthcare abroad, resulting in negative tourist travel experiences. Once identified, specific strategies should be adopted to improve accessibility and the overall quality of care for tourists. Full article