87191 research outputs found
Sort by
Reevaluating zero-shot information extraction: Sampling bias, prompting transferability and sensitivity in large language models
Get PDFLarge Language Models (LLMs) have advanced zero-shot Information Extraction (IE), particularly in Sentence-level Relation Extraction (SentRE), through in-context learning and instruction tuning. However, the current evaluation of LLMs’ zero-shot ability on IE tasks remains fragile and unreliable. In this work, we provide a systematic examination of the fragility underlying current evaluation practices across three interrelated levels. At the data level, we demonstrate that the commonly adopted random sampling strategy introduces significant biases in class-imbalanced datasets, whereas balanced sampling provides more stable and faithful assessments of LLMs performance. At the task level, we reveal that three domain prompt frameworks on SentRE transfer inconsistently to Document-level Relation Extraction (DocRE) and Named Entity Recognition (NER), showing partial effectiveness on NER but notable limitations on DocRE due to long contexts and complex entity structures. At the method level, through extensive experiments on three IE tasks and seven datasets, we conduct the first comprehensive comparison of five general prompt frameworks, including Chain-of-Thought, Self-Improvement, and Self-Debate, showing that prompt effectiveness is highly task-dependent, with no single strategy dominating across tasks. For each task, the CoT prompt framework achieves the best performance on SentRE, the Vanilla prompt framework performs best on DocRE, and the Self-Consistency prompt framework excels on NER. These insights challenge current landscape of information extraction, providing guidelines for robust evaluation and prompt designs
Enhancing Active Channel Delineation in Alluvial Rivers Using Monthly Aggregation of Sentinel‐2 Imagery
Get PDFPlain Language Summary: Active river channels show where the river has shaped its bed over a certain time. With today's frequent satellite observations, active channels can be monitored monthly. Yet, long‐term studies still use yearly maps built on summer images or annual medians to reduce processing time and smooth seasonal changes. The role of monthly information in improving annual active channel delineation and better understanding river processes remains unexplored. Here, we mapped yearly active channels of the Po River (Italy) by aggregating Sentinel‐2 monthly classifications of river water and sediment bars. Using the persistence in which a pixel was classified as active channel during the year, we automatically smoothed some classification errors and bias. Monthly data also revealed significant intra‐annual variability, with dynamic reaches changing their active channel area by nearly 50% over the year—dynamics that are missed in single annual images. In less dynamic reaches, results are instead similar whether using monthly or annual data. Overall, results show that in dynamic river reaches sub‐annual data better identify when and how sediment–vegetation interactions start shaping active channel reconfiguration and migration—a mirror to the river's ecological functioning and sediment transport dynamics
On the transport properties of K 2 ZnV 2 O 7
Get PDFK2ZnV2O7 has recently been reported as a promising oxide ion conductor. We have studied this material using a number of structure- and physical property-probing techniques. Our extensive characterisation using variable temperature synchrotron X-ray and neutron diffraction, impedance spectroscopy and tracer diffusion measurements of its transport properties, does not support the reports that K2ZnV2O7 undergoes partial reduction at high temperatures, leading to the creation of vacancies and oxide ion conductivity. In particular, the lack of oxide ion diffusion observed by isotope exchange definitively rules out oxide ion conductivity within K2ZnV2O7. Instead, we find that the high conductivity measured originates from the melting of a small amount of KVO3 impurity in the sample, which is detectable by synchrotron X-ray and neutron diffraction
Xenomusicology
No full textSpeculative xenomusicology explores alternative music theories, imagining the physical and cognitive affordances of alien musical life. Exoplanets are actively studied in astronomy, and though there is no direct evidence of xenobiology, particularly of more advanced musical intelligences, potential alien music may still be considered in advance in the same way that exobiologists speculate on the conditions for alien life. In particular, a generative system is presented which creates imagined xenomusic based on altering human memory constraints and links the organisation of the sound to the parallel generation of an alien language. Microtonal pitch, complex rhythm, timbral material and spatialisation within putative alien architectures are all considered. This alien ‘analysis by synthesis’ can provide new musical adventures and new understanding of the possibilities of music theoretical space, regardless of any eventual ontological resolution of xenocultures
Do Coding Club After-School Activities Improve Pupils’ Non-cognitive Skills and Performance in Coding?
No full textThe development of coding skills is considered important for a digital age, linked to broader learning and non-cognitive abilities. Schools in England integrate coding into the curriculum, perhaps supplemented by after school activities. “Code Club” is a structured learning programme offered as an after-school activity for pupils aged 9 to 13. This study evaluates Code Club using a quasi-experimental design assessing the impact on student attitudes to learning and coding skills. Pupils attending the 22-week programme were compared to non-participants in the same schools. Initially, 412 pupils (Years 4–9) participated, but only 239 from 13 schools remained for full analysis. The findings suggest positive effects for resilience (+0.22), confidence (+0.47), sense of belonging (+0.09), and coding skills (+0.24). 40% dropout rate limits the generality of the conclusions. Code Club, run voluntarily in schools, was well-received, with participants enjoying coding and digital projects. The paper discusses the findings and implications
Are You in the Zone when Working from Home? How Remote Workers' Daily Flow Experiences Promote Daily Functioning and Well-Being Through Reduced Work-Home Interruption Behaviors
No full textThe surge in telework after the COVID-19 pandemic has raised debates between employers and employees about how teleworkers can balance flexibility and productivity while working remotely. Our study examines how flow experiences as a volatile personal resource can facilitate teleworkers' functioning and well-being at work and at home through effortless self-regulation. Drawing on the work-home resources model, we argue that daily flow experiences promote teleworkers’ work-domain functioning (e.g., work engagement and need for recovery) and home domain well-being (e.g., subjective vitality and regulatory resource depletion) through reducing work-home interruption behaviors. Furthermore, we propose that states of high daily morning mindfulness act as another volatile personal resource that can support teleworkers' functioning and well-being on days with lower levels of flow experiences by attenuating the daily relationship between flow experiences and work-home interruption behaviors. Results from an experience sampling study with teleworkers during the COVID-19 pandemic (N=87 individuals across N=607 days) support our hypotheses that reduced work-home interruption behaviors mediate the daily relationships between daily flow experiences and work engagement, need for recovery, subjective vitality, and regulatory resource depletion. Our results further highlight that these indirect relationships became weaker on days when teleworkers experience higher, as compared to lower, mindfulness in the morning. Thus, on days with lower levels of flow experiences, mindfulness can be an effective way to facilitate functioning and well-being across domains. Our findings offer theoretical insights and practical implications by revealing how teleworkers can remain both productive and health
Automated surface damage detection with an embodied intelligence robotic platform
No full textRegular inspections of civil structures and infrastructure, performed by professional inspectors, are costly and demanding in terms of time and safety requirements. Additionally, the outcome of inspections can be subjective and inaccurate as they rely on the inspector’s expertise. To address these challenges, autonomous inspection systems offer a promising alternative. However, existing robotic inspection systems often lack adaptive positioning capabilities and integrated crack labelling, limiting detection accuracy and their contribution to long-term dataset improvement. This study introduces a fully autonomous framework that combines real-time crack detection with adaptive pose adjustment, automated recording and labelling of defects, and integration of RGB-D and LiDAR sensing for precise navigation. Damage detection is performed using YOLOv5, a widely used detection model, which analyzes the RGB image stream to detect cracks and generates labels for dataset creation. The robot autonomously adjusts its position based on confidence feedback from the detection algorithm, optimizing its vantage point for improved detection accuracy. Experiment inspections showed an average confidence gain of 18% (exceeding 20% for certain crack types), a reduction in size estimation error from 23.31% to 10.09%, and a decrease in the detection failure rate from 20% to 6.66%. While quantitative validation during field testing proved challenging due to dynamic environmental conditions, qualitative observations aligned with these trends, suggesting its potential to reduce manual intervention in inspections. Moreover, the system enables automated recording and labeling of detected cracks, contributing to the continuous improvement of machine learning models for structural health monitoring
Featureless stars: flux calibration for extremely large telescopes
No full textThe spectrophotometric flux calibration of recent spectroscopic surveys has reached a limiting systematic precision of approximately 1 − 3 per cent, and is often biased near the wavelengths associated with H I Balmer absorption. As we prepare for the next generation of imaging and spectroscopic surveys, and high-precision cosmology experiments, we must find a way to address this systematic. Towards this goal, we have identified a global network of 29 bright (G < 17.5) featureless white dwarf stars that have a spectral energy distribution consistent with an almost pure blackbody form over the entire optical and near-infrared wavelength range. Based on this sample, we have computed the systematic uncertainty and AB magnitude offsets associated with Gaia, SDSS, SMSS, Pan-STARRS, DES, and 2MASS, and we have also checked the consistency of our objects with both GALEX and WISE. The magnitude range of the featureless stars reported here are ideally suited to observations taken with the forthcoming generation of extremely large telescopes, as well as calibrating the survey data acquired by the Rubin, Euclid, and Roman observatories. Finally, all of the high-precision spectrophotometric standard stars reported here have been included in the latest release of the PYPEIT data reduction pipeline
Surviving the Black Death in Medieval England: Recovering from Illness at Warboys, Huntingdonshire
No full textThe Black Death swept through Europe between 1347 and 1353, killing perhaps half of the population. Yet, there is remarkably little information about those who contracted plague and recovered. Who fell ill? How long did their recovery take? Why did some survive, and others succumb to the disease? This article provides the first evidence of a group of twenty-two individuals on the manor of Warboys who missed their labour services due to illness over the summer of 1349. Most returned to work within three or four weeks, providing fresh insights into how long it took survivors to reintegrate into society
The social learning and development of intra- and inter-ethnic sharing norms in the Congo Basin
No full textCompared to other species, the extent of human cooperation is unparalleled. Such cooperation is coordinated between community members via social norms. Developmental research has demonstrated that very young children are sensitive to social norms, and that social norms are internalized by middle childhood. Most research on social norm acquisition has focused on norms that modulate intra-group cooperation. Yet around the world, multi-ethnic communities also cooperate, and this cooperation is often shaped by distinct inter-group social norms. In the present study, we investigated whether intra-ethnic and inter-ethnic social norm acquisition follows the same, or distinct, developmental trajectories. Specifically, we worked with BaYaka foragers and Bandongo fisher-farmers who inhabit multi-ethnic villages in the Republic of the Congo. In these villages, inter-ethnic cooperation is regulated by sharing norms. Based on our ethnographic knowledge of the participating communities, we predicted that children’s intra-ethnic sharing choices would match those of adults at an earlier age than their inter-ethnic sharing choices. To test this prediction, children (5–17 years) and adults (17 + years) participated in a modified Dictator Game to investigate the developmental trajectories of children’s intra- and inter-ethnic sharing choices. Contrary to our prediction, both intra- and inter-ethnic sharing norms were acquired in middle childhood. Interviews with adult participants suggested that intra- and inter-ethnic sharing norms are acquired from multiple sources, including parents and peers. Further, Bandongo adults primarily reported learning sharing norms via Instruction, whereas BaYaka adults primarily reported learning via Observation/Imitation. These cross-cultural differences may reflect variation in norm complexity. Together, these findings suggest that when social contexts regularly expose children to out-group collaboration, inter-ethnic norms are acquired at similar timelines to intra-ethnic ones, as part of children’s broader cooperative repertoire
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! 👇