Open Research Exeter - University of Exeter
Not a member yet
41213 research outputs found
Sort by
Human-in-the-loop testing of AI agents for air traffic control with a regulated assessment framework
No full textWe present a rigorous, human-in-the-loop evaluation framework for assessing the performance of AI agents on the task of Air Traffic Control, grounded in a regulator-certified simulator-based curriculum used for training and testing real-world trainee controllers. By leveraging legally regulated assessments and involving expert human instructors in the evaluation process, our framework enables a more authentic and domain-accurate measurement of AI performance. This work addresses a critical gap in the existing literature: the frequent misalignment between academic representations of Air Traffic Control and the complexities of the actual operational environment. It also lays the foundations for effective future human-machine teaming paradigms by aligning machine performance with human assessment targets.</p
Echoes of automation: How bots shaped political discourse in Brazil, 2018-2022
No full textIn an era where social media platforms are central to political communication, the activity of bots raises pressing concerns about amplification, manipulation, and misinformation. Drawing on more than 315 million tweets posted from August 2018 to June 2022, we examine behavioural patterns, sentiment dynamics, and thematic focus of bot- versus human-generated content spanning the 2018 Brazilian presidential election and the lead-up to the 2022 contest. Our analysis shows that bots relied disproportionately on retweets and replies, with reply activity spiking after the 2018 election, suggesting tactics of conversational infiltration and amplification. Sentiment analysis indicates that bots maintained a narrower emotional tone, in contrast to humans, whose sentiment fluctuated more strongly with political events. Topic modelling further reveals bots’ repetitive, Bolsonaro-centric messaging, while human users engaged with a broader range of candidates, civic concerns, and personal reflections. These findings underscore bots’ role as amplifiers of narrow agendas and their potential to distort online political discourse.</p
Black sheep: Gossip and more-than-human racialization in French néo-ruralité
No full textThis article brings together French-language ethnological literature on néo-ruralité with interdisciplinary debates about more-than-human intersectional marginalization, to contribute to the existing scholarship on the anthropology of gossip. Using a rich ethnographic case study widely known as the “black ram-gate” in the Cotentin region, I argue that néo-ruraux (urbanites who relocate to a rural community as part of a life project) can be shown to use the gossip allegedly spread about them by ruraux de souche (individuals identified by néo-ruraux or by their own account as “born and raised in the area”) to discredit the latter’s suitability for modern rural life. Here, as outsiders, néo-ruraux are able to construct gossip as an outdated activity, synonymous to rural conservatism, xenophobia and social paralysis. However, a focus on animal breeding narratives reveals that, far from elevating themselves above gossip altogether, néo-ruraux use similar scandalizing devices to racialize ruraux de souche in their turn.</p
SHARE-U: Semantic-aware Human Avatar Reconstruction with Efficient Gaussians via Uncertainty-weighted Contrastive Learning
No full textWe present SHARE-U, a 3D Gaussian Splatting (3DGS)-based human model, represented by significantly fewer Gaussians, achieving high-fidelity quality with fast training and real-time rendering speed. To constrain 3D Gaussian optimization to adapt it for monocular human modeling, regularization-based models rely on rigid rules, while another trend is employing more flexible contrastive learning. However, existing methods typically require a large number of Gaussians to represent a human avatar, increasing memory consumption and reducing rendering efficiency, and struggle to achieve photo-realistic rendering with fewer Gaussians. This limitation stems from their uniform 3D constraints to all Gaussians, which fail to distinguish between well-observed and poorly-observed Gaussians, leading to over-constraints on poorly-observed ones with large reconstruction errors and thereby degrading local details. To address these challenges, SHARE-U relies on uncertainty-weighted contrastive learning and semantic-aware optimization. Specifically, SHARE-U applies Hessian-derived uncertainty to distinguish and weight each Gaussian sample in contrastive learning, strengthening supervision for low-uncertainty Gaussians and attenuating the influence of high-uncertainty ones. During optimization, we innovatively propose a semantic-aware densification strategy to adaptively control Gaussians’ density according to different body parts’ granularity. For modeling finer textures without increasing the Gaussian number, we further suggest a Fourier loss to align texture within each semantic part in the frequency domain. Therefore, this cohesive framework forms the key insight of SHARE-U: the semantic-aware optimization yields an efficient but under-constrained Gaussian representation, and the uncertainty-weighted contrastive learning specifically enables its robust optimization. Extensive experiments on the ZJU Mocap and MonoCap datasets demonstrate that SHARE-U outperforms previous state-of-the-art human models both quantitatively and qualitatively.</p
Explaining famine in the British Empire: Agricultural science, food security, and the rise of statistics
No full textThis book is about the famines and food shortages that struck India and Britain at the close of the eighteenth century, and it explores how these crises and episodes of scarcity gave rise to scientific efforts to explain and quantify ‘famine’. Focusing on the time period between the Bengal famine of 1770 and the food shortages in Britain in 1800, it explores the development of the concepts of ‘artificial scarcity’ (and ‘artificial famine’), and how statistical science and philosophy played a role in the naturalization of famine. During this time, Britain’s first ‘Board of Agriculture’ was established, creating political opportunities for a rising class of agriculturalists interested in the promotion of their science as a means of confronting and solving the empire’s food insecurity during a time of war and upheaval. Following the networks and collaboration between this Board of Agriculture and the East India Company, the book explores the careers and correspondence of agriculturalists, economists, Company officials, scientists, hack writers, and politicians. In conclusion, Explaining Famine shows how these debates over the anthropogenic and natural causes of scarcity and famine shaped the subsequent development of the field of food security and modern concerns over carrying capacity, environment, and population.</p
Unsupervised Geospatial Analysis of Terror in the UK
No full textThis paper aims to apply geospatial analysis via k-means clustering on the global terrorism database supplied by the Study of Terrorism and Responses to Terrorism (START) consortium. By performing parameter sweeps for the algorithm we optimise implementation and identify patterns and gain insight into the nature of terrorist attacks in a case study, focusing specifically on the UK to analyse terrorist events in a localised environment. The aim of this paper is to show unsupervised machine learning (ML) methods to meet multiple targets set by the United Nations Sustainable Development Goals (UNSDG) 16.</p
Navigating Special Educational Needs Support in Private, Voluntary and Independent Early Years settings: Practitioner Preparedness and Educational Psychologist Involvement
No full textThis thesis investigates how well-prepared Early Years Practitioners feel in identifying and supporting children with Special Educational Needs (SEN) in England, and the role that Educational Psychologists (EPs) have in this process. The study uses a qualitative approach, using semi-structured interviews to explore the experiences and perspectives of EYPs in Phase 1 and EPs in Phase 2. The research is grounded in a relativist ontological stance and social constructionist epistemology, acknowledging the subjective nature of reality and the socially constructed nature of knowledge in early years SEN identification and support.
The findings reveal significant disparities in the training and qualifications of EYPs. Many EYPs rely heavily on personal experience and informal learning strategies due to limited access to formal SEN-specific training; these findings align with the literature review. This variability in preparation raises concerns about the consistency and quality of SEN support across different settings. The study also highlights a disconnect between EYPs' self-perception as knowledgeable professionals and the recognition they receive from external professionals, impacting their professional identity and effectiveness in supporting children with SEN.
The findings also reveal that EPs' involvement in early years settings is often limited to statutory assessments, with both EYPs and EPs expressing a strong desire for more regular engagement, particularly for staff training and ongoing support. The research highlights the need for enhanced multi-agency collaboration and systemic support to ensure equitable and high-quality SEN provision. By addressing these gaps, this study contributes to the ongoing dialogue on SEN support, advocating for policy changes and professional development initiatives to equip EYPs better and integrate EPs more effectively into early years settings.</p
“Her divers ways and paths”: Edna O’Brien’s “Sabiola” columns and Irish Women’s Mobility at Mid-Century
No full textThis article uncovers and analyses the 123 columns Edna O’Brien published under the pseudonym “Sabiola” between 1950 and 1953 to make some larger claims about O’Brien’s work and about mid-twentieth century Irish writing more broadly. The Sabiola columns were published in The Link, a weekly newspaper for employees of Córas Iompair Éireann (CIÉ, the Republic of Ireland’s public transport authority). The article argues that the Sabiola apprenticeship equipped O’Brien with concrete knowledge of Irish transport and the metaphorical possibilities of the same, concerns to which she would return throughout her long and illustrious career. Modes of transport – walking, bicycles, trains, buses, cars and airplanes – loom large in O’Brien’s early fiction and are often bound up with place, class, labour and, most particularly, gender. Moreover, mobility is central to the Sabiola column itself, both thematically and formally. In it, O’Brien foregrounds her peregrinations in (and beyond) Dublin, adopting a kind of “woman about town” persona usually reserved for male writers of the period. Ultimately, O’Brien’s writings for The Link confirm the ongoing significance of two vehicles of modernity – the railway and the newspaper – in what Neal Alexander and James Moran term the “long afterlife” of “regional modernism [that] extends well into the second half of the twentieth century.”</p
Using A Sui Generis System to Protect Traditional Chinese Medicine
No full textThere has been a global increase in the practice and use of traditional medicine and this trend has resulted in the flourishing of conflicts between the legal issues on protecting the relative resources of traditional medicine and the current intellectual property framework. This thesis focuses on the legal issues and policies on the management and protection of traditional Chinese medicine within China’s legal framework. This thesis comprises a systematic study on the protection of traditional Chinese medicine, as an independent research object, on the basis of re-identifying the legal characteristics of traditional Chinese medicine.
This thesis analyses the status quo of the protection of traditional Chinese medicine and the underlying causes of the conflicts with China’s intellectual property system, especially the patent system. The thesis further assesses the feasibility and legitimacy of establishing a sui generis system to regulate and protect traditional Chinese medicine resources in China's legal framework.
In line with the legal characteristics of traditional Chinese medicine, this thesis provides legislative suggestions for establishing the sui generis system, including competent authorities, a national database of traditional Chinese medicine, procedures of prior informed consent, a framework of access and benefit sharing and provisions on disclosure requirements of origins. Finally, the thesis assesses the possible impact of the sui generis system on the rights and interests of all parties concerned with the resources of traditional Chinese medicine.</p
A Long-Run Perspective on Unconditional Convergence in Manufacturing: Evidence from U.S. Industrialization
No full textThis paper examines long-run unconditional convergence of labour productivity in manufacturing across 48 contiguous U.S. states. We construct a detailed panel data set of state-industry pairs with over 120 industries covering the period 1880-2007. The remarkable finding is a rapid rate of unconditional β-convergence: 7.6% per year in the period 1880-2007, and about 6.5% in each period 1880-1940 and 1958-2007. We also find a steep rise in unconditional convergence from 5.5% in 1930-1940 to 10.6% in 1940-1947 and 9.4% in 1947-1958, followed by a considerable drop thereafter. In the years 1958-2007, the rate of convergence in ICT industries (8.3%) is higher than in 1st-Industrial-Revolution (6.3%) or 2nd-Industrial-Revolution (7.0%) industries. We interpret our results as a process of technological catch up between initially laggard Southern manufacturing and manufacturing industries in the North.</p