435032 research outputs found
Sort by
Closing the denitrification gap:applying the <sup>15</sup>N gas flux method with an artificial atmosphere in conventional and regenerative agriculture
Get PDFIntroduction: Denitrification is an elusive process that remains notoriously difficult to measure under field conditions, yet it plays a crucial role as the only natural terrestrial sink for reactive nitrogen, especially in agricultural systems where large amounts of fertilizer are applied. Direct measurements of N2 fluxes over extended periods remain rare in the literature due to technical challenges. Methods: In this study, we quantified and characterized denitrification emissions under two contrasting land-use practices—conventional and regenerative (unfertilized) agriculture—using a recently developed custom method combining a 15N isotopic tracer with an artificial atmosphere (improved 15N Gas Flux method). We conducted nine field campaigns over one year to (i) assess method applicability, (ii) derive a first annual estimate of denitrification, (iii) understand controls on denitrification dynamics, and (iv) trace denitrification‐driven losses of applied synthetic nitrogen fertilizer in conventional agriculture. Results: Our method successfully detected denitrified N2 fluxes in 90% of measurements and yielded annual budgets of 22.12 and 2.41 kg N ha‐1 yr‐1 in the conventional and regenerative fields, respectively. Soil moisture and nitrate availability (particularly under fertilized conditions) were the main controls on the denitrification product ratio (N2O/(N2O + N2)). We estimated that 11% of applied fertilizer nitrogen was lost via denitrification in the conventional field, with 7.3% of this loss emitted as N2O rather than N2. Discussion: These results underscore the role of fertilization management in shaping denitrification dynamics and its potential to act as a sink for reactive nitrogen, while modulating N2O emissions
Off Grid:The Problem of Early-Eighteenth-Century Caribbean Sinew Populations
Get PDFTo understand the early modern Caribbean, we must understand the societies that inhabited it. The parameters through which historians approach these societies have changed drastically in the last decade. While recent interventions have proven useful for framing our attitude to how populations in the Caribbean formed, they are less effective when applied to societies whose longevity was uncertain that, in some cases, fractured or collapsed. It is in this context that some historians have identified what they term “sinew populations”: communities whose “off-grid” nature necessitates different ways of thinking about how they functioned. Recent works have discussed how sinew populations ensured the long-term viability of their communities, but this approach also requires attention to the factors that could render a sinew population’s existence unviable. This article uses an eighteenth-century Caribbean population of pirates as a case study to illustrate the issue of viability within sinew populations. In particular, the article emphasizes the weak social foundations on which this sinew population was built and the lack of interest among the pirates themselves, after 1718, in maintaining a large pirate population. In thinking about how pirates related to one another and what this meant for the long-term survival of the pirate sinew population, this article demonstrates the importance of social maintenance for understanding how Caribbean societies operated
Mental health advice on TikTok
Get PDFIn this paper, we provide the first, large-scale corpus-pragmatic analysis of mental health advice by social media influencers on TikTok. We identify advice-giving in large datasets focusing on if-conditionals as a specific form that allows us to analyse how the audience is positioned relative to a need and the solution which is then proposed. To identify the different ways in which mental health issues are presented, we use an adapted version of the ‘mental health quotient’ (Newson and Thiagarajan, 2020), as a linguistically informed framework for differentiating between lay discussions of mental health and those that invoke specific disorders. We sample a corpus of over 27,000 TikTok videos from 85 mental health influencers, using corpus-scale identification to extract and analyse if-conditionals produced by mental health professionals and wellness influencers. Our analysis of the protasis shows how these two types of influencers use prompts that share some similarities but also rely on fundamentally different models of healthcare. The relationship between these prompts and the information and recommendations in the apodosis show how health professionals rely on diagnostic information and therapeutic advice, while wellness influencers recommend embodied practice and products to treat mental health issues. These findings set out the distinctive ecosystem of healthcare which is emerging within the algorithmically driven contexts of sites like TikTok
Distribution characteristics and ecotoxicological risks of typical organic flame retardants and plasticisers in the Amazon River basin around Manaus, Brazil
Get PDFCurrent understanding of potential anthropogenic/environmental drivers and ecotoxicological risks of emerging contaminants in the Amazon Basin, particularly organic flame retardants and plasticisers, is limited. We collected sediment samples from Amazonian Rivers near Manaus city/Amazonas State, Brazil, and measured concentrations of 39 organic flame retardants and plasticisers, namely: polybrominated diphenyl ethers (PBDEs), novel brominated flame retardants (NBFRs), dechlorane plus (DP), hexabromocyclododecane (HBCDD), organophosphate esters (OPEs), and polychlorinated biphenyls (PCBs). Concentrations of Σ9PBDEs, Σ9NBFRs, Σ2DPs, Σ3HBCDDs, Σ7OPEs, and Σ9PCBs ranged between <0.26 and 5.1, 2.0 – 18, <0.075 – 290, <0.0014 – 0.29, 2.2 – 15, and 0.18 – 3.0 ng/g dw in sediments, respectively. Distribution of legacy BFRs (PBDEs and HBCDD), DPs, and OPEs in sediments was primarily driven by domestic sources; distribution of NBFRs was mainly driven by industrial sources; while distribution of PCBs was primarily modulated by organic matter. We also measured PBDEs in paired surface water samples collected from the Amazon basin near Manaus, with concentrations of Σ9PBDEs ranging between 0.18 and 3.4 ng/L. In contrast to sediments, PBDE concentrations in water were primarily controlled by short-term aggregation and sedimentation enhanced by organic matter. Comparisons of our observations with PNEC (predicted no-effect concentration) values imply high ecotoxicological risks presented by DP, bis(2-ethyl hexyl) tetrabromophthalate (BEH-TEBP), PCB-11, and tris(1-chloro-2-propyl) phosphate (TCIPP) in sediments, and low to medium risks by other pollutants. This is especially striking for DP for which the maximum RQ (Risk Quotient) value was 24.7 at sampling point 1, possibly due to the impact of domestic waste disposal and intensive boat traffic
Unilateral red eye, consider short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing
Get PDFMetaphor identification using large language models:A comparison of RAG, prompt engineering, and fine-tuning
No full textMetaphor is a pervasive feature of discourse and a powerful lens for examining cognition, emotion, and ideology. Large-scale analysis, however, has been constrained by the need for manual annotation due to the context-sensitive nature of metaphor. This study investigates the potential of large language models (LLMs) to automate metaphor identification in full texts. We compare three methods: (i) retrieval-augmented generation (RAG), where the model is provided with a codebook and instructed to annotate texts based on its rules and examples; (ii) prompt engineering, where we design task-specific verbal instructions; and (iii) fine-tuning, where the model is trained on hand-coded texts to optimize performance. Within prompt engineering, we test zero-shot, few-shot, and chain-of-thought strategies. Our results show that state-of-the-art closed-source LLMs can achieve high accuracy, with fine-tuning yielding a median F1 score of 0.79. A comparison of human and LLM outputs reveals that most discrepancies are systematic, reflecting well-known grey areas and conceptual challenges in metaphor theory. We propose that LLMs can be used to at least partly automate metaphor identification and can serve as a testbed for developing and refining metaphor identification protocols and the theory that underpins them
Hardware-in-the-loop real-time simulation of an active bypass SOC/Voltage balancing topology with ultrafast charging and comparative power loss analysis
Get PDFCell balancing is a crucial part of the battery management system to optimise the charging process of the electric vehicle battery pack. Recently, Electric vehicles have been using ultra-fast charging that can initially charge the battery at a 2C rate. However, basic SOC/voltage balancing topologies cannot provide SOC/voltage balance during this charging time. This paper presents the modular bypass balancing technique, which can balance the SOC/voltage during ultra-fast charging. The voltage-based controller is designed to eliminate the need for accurate SOC estimation. The proposed topology uses MOSFETS as a switching device to connect or disconnect the module according to the voltage measurement and control algorithm. The constant current constant voltage (CCCV) charge controller is designed, as the bypass topology requires an adaptive control to adjust the charger supply voltage according to the number of bypass modules. The results prove that the CC to CV mode is switched when the battery SOC reaches 85 %. To check the performance and stability of the proposed topology, it has been tested with dynamic initial SOC conditions and charging rates (0.5C and 1C). The hardware in the loop setup is adopted to verify the reliability and feasibility of the proposed topology and controller design. The experimental results showed that the proposed topology successfully achieved SOC/voltage balancing by reducing the SOC difference to 0 % while the voltage difference is reduced from 2 V to 250 mV. The results also verified that the proposed topology gives a shorter balancing time as compared with the other methods. The balancing time is compared according to the initial SOC imbalance and the balancing current. It shows that the proposed method can provide SOC/Voltage balancing in 30 min while using 2C rate charging and 80 % initial imbalance. Since the power loss is directly proportional to the charging current thus the compression is performed under 2C rate charging. The proposed topology produces a power loss of 12.5 W, while other topologies produce 12.85 W to 51.41 W, considering the number of switching devices used in each method
Effect of thermal aging on the mechanical properties of oxide/oxide composites manufactured by a prepreg technique
Get PDFThe effectiveness of an oxidation resistant monazite coating on the mechanical properties of oxide composites after long term thermal exposure at 1200°C has been investigated. The monazite coating was produced by a dip coating process and is believed to have reduced the strength of the fibre / matrix bonding. As a result, the monazite-coated oxide composites showed non-brittle failure behaviour with extensive fibre pull out and good flexural strength, even after thermal aging. In contrast, after similar thermal aging the uncoated composite showed significant strength degradation and brittle failure behaviour, with no fibre pull out observed
FedPAC:A Federated Semi-Supervised Learning Approach for Non-IID Data with Feature Shift
Get PDFFederated Semi-Supervised Learning (FSSL) enables collaborative model training across distributed clients with limited labeled data while preserving data privacy. However, a critical challenge in FSSL is feature shift, where clients exhibit diverse feature distributions despite sharing the same task. To address this issue, we propose FedPAC, a novel FSSL framework that integrates Contrastive Mean-Teacher Regularization and Perturbation-Aware Gradient Descent. Our framework enhances feature representation learning by aligning feature distributions between teacher and student models and mitigates optimization challenges caused by feature heterogeneity through controlled gradient perturbations. Extensive experiments on benchmark datasets demonstrate that FedPAC outperforms existing FSSL methods in feature shift scenarios, making it a practical solution for real-world applications such as medical imaging and industrial fault diagnosis
The developmental trajectories of implicit and explicit metacognitive monitoring and control in cued recall
Get PDFAdults are adept at metacognitively monitoring their memory accuracy—both explicitly and implicitly—and at using metacognitive control to maintain high memory accuracy. However, the development of monitoring and control is less well understood. We administered an episodic cued recall task with children aged five to 11 years (N = 106). Participants watched two video clips of everyday episodic events before answering cued recall memory questions. For each memory question, participants provided a confidence rating (explicit monitoring), sorted their answer into show/hide boxes (control), and chose to volunteer/withhold their response (control). Multiple behavioural gestures of cognitive effort (implicit monitoring; e.g., looking to carer, non-word fillers) were recorded and later coded by blind raters. Children were less accurate and less able to assign confidence to reflect their memory accuracy when they were forced to generate a response after previously saying “I don’t know”. But on volunteered trials, explicit, implicit monitoring and control measures predicted memory accuracy. There were age-related improvements in explicit monitoring for predicting memory accuracy, but there were no age differences in implicit monitoring or control processes. We found evidence for both a direct and an indirect link between confidence and memory accuracy. Our findings suggest that explicit and implicit monitoring have different developmental trajectories in cued recall and that children can be adaptive to control their memory accuracy to a similar extent from early- to mid-childhood
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇