AUT Research Repository (Auckland Univ. of Technology)
Not a member yet
14707 research outputs found
Sort by
Rituals of Violent Masculinity: A Feminist Comparative Historical Analysis of Male-Male Fighting, Shame and Misogyny
No full textThis article uses a combination of two feminist research methods to further understanding of the enduring nature of men’s use of ritualised forms of violence. In particular, this article examines men fighting other men to mitigate the effects of feminized shame and to stabilise masculine honour. Using a feminist comparative historical analysis alongside a feminist systematic review, two manifestations of ritualised honour-based fighting will be explored: men’s duelling of the eighteenth and nineteenth century and today’s (hetero)romantic and homosocial practice of territory marking: men claiming ownership over their (hetero)romantic partner by threatening to fight other men who appear to be romantically interested in her. By looking at the relationship between two types of ritualised fighting from different time-periods, the enduring nature of why men fight other men to mitigate feminized shame can be discussed in new ways. This type of analysis helps shed light on inherent fragilities within these violent practices, signalling how men’s ritualised fighting could be destabilised in the future
Prompt-based Few-Shot Text Classification With Multi-Granularity Label Augmentation and Adaptive Verbalizer
No full textFew-Shot Text Classification (FSTC) aims to classify text accurately into predefined categories using minimal training samples. Recently, prompt-tuning-based methods have achieved promising results by constructing verbalizers that map input data to the label space, thereby maximizing the utilization of pre-trained model features. However, existing verbalizer construction methods often rely on external knowledge bases, which require complex noise filtering and manual refinement, making the process time-consuming and labor-intensive, while approaches based on pre-trained language models (PLMs) frequently overlook inherent prediction biases. Furthermore, conventional data augmentation methods focus on modifying input instances while overlooking the integral role of label semantics in prompt tuning. This disconnection often leads to a trade-off where increased sample diversity comes at the cost of semantic consistency, resulting in marginal improvements. To address these limitations, this paper first proposes a novel Bayesian Mutual Information-based method that optimizes label mapping to retain general PLM features while reducing reliance on irrelevant or unfair attributes to mitigate latent biases. Based on this method, we propose two synergistic generators that synthesize semantically consistent samples by integrating label word information from the verbalizer to effectively enrich data distribution and alleviate sparsity. To guarantee the reliability of the augmented set, we propose a Low-Entropy Selector that serves as a semantic filter, retaining only high-confidence samples to safeguard the model against ambiguous supervision signals. Furthermore, we propose a Difficulty-Aware Adversarial Training framework that fosters generalized feature learning, enabling the model to withstand subtle input perturbations. Extensive experiments demonstrate that our approach outperforms state-of-the-art methods on most few-shot and full-data splits, with F1 score improvements of up to +2.8% on the standard AG’s News benchmark and +1.0% on the challenging DBPedia benchmark.</jats:p
Beyond the Frontline: Exploring Indirect Trauma and Organisational Stress in Emergency Communication Centre Employees
No full textEmergency communication centre employees are indirectly exposed to potentially psychologically traumatic events (PPTEs), yet research on their mental health is limited. This national census survey (73.4% response rate, n=58) examines symptoms of Major Depressive Disorder (PHQ-9), Generalised Anxiety Disorder (GAD-7), and PTSD (SPRINT) among Fire and Emergency New Zealand personnel. It also explores PPTE exposure, organisational stressors, employee experiences, and coping mechanisms (AUDIT-C, emotional numbing scale) using mixed methods. Results show 64% screened positive for at least one mental disorder, with high rates of emotional numbing (31%) and hazardous alcohol use (51%). Qualitative analysis, including interviews and open-text survey responses, highlights stressors such as inadequate staffing, excessive workload, and lack of support. As the first dedicated study on this group in New Zealand, findings highlight the urgent need for targeted interventions.</jats:p
The Antithesis of Hospitality: Unpacking Workplace Bullying and Advancing a Māori-centric Response
No full textThis paper examines workplace bullying in the hospitality sector – an industry paradoxically defined by welcoming others – through a mixed-method approach integrating large-scale quantitative analysis with an in-depth qualitative case study. Study 1 draws on survey data from 2,302 hospitality employees in Aotearoa, New Zealand, to identify the prevalence, patterns, and perpetrators of bullying, and employees’ confidence in employer responses. Over half (56%) reported experiencing or witnessing bullying, with women and supervisors most affected. Study 2 explores a Māori hospitality business guided by manaakitanga (care), whanaungatanga (relationships), and tika (fairness), illustrating how Māori values can counter bullying behaviours. Together, the studies reveal the gap between hospitality’s ideals and workplace realities, proposing Māori-informed approaches as a pathway towards more respectful, inclusive, and restorative organisational environments. The paper contributes to management and hospitality scholarship by demonstrating how Indigenous relational ethics can operationalise organisational care as an antidote to workplace harm
Category-dependent Preferences and Stochastic Choice
No full textWe introduce a generalisation of (Aguiar’s 2017) random categorisation rule (RCR) that relaxes (Aguiar’s 2017) Acyclicity axiom to an Asymmetry condition. Unlike other alternatives to the RCR, such as the models of Brady and Rehbeck (2016) and Cattaneo et al., (2020), our generalised random categorisation rule (GRCR) relaxes the linearity of preference, rather than varying the random consideration process. We show that the GRCR is also equivalent to allowing linear preferences to be category-dependent, with the requirement that the preferences associated with overlapping categories agree on the intersection. This allows the GRCR to capture intransitivities across categories, which may naturally arise when categorisation is used to frame choices. We provide a characterisation for the GRCR and compare it to the random utility model, as well as to the BR model, the RAM, and the model of Manzini and Mariotti (2014)
Communal to Individual Midwifery Care: Cultural Practices and the Maternity Journey of Sub-Saharan African Women in New Zealand
No full textMaternal health disparities persist globally, including among Sub-Saharan African immigrant women in high-income countries. Many come from contexts where pregnancy, childbirth, and the postpartum period are embedded in communal traditions. In Aotearoa New Zealand, the birthplace of cultural safety, limited research has examined African women’s maternity experiences. This study forms part of a midwife-led qualitative exploration of the maternity journeys of women from Sub-Saharan Africa in New Zealand, using interpretive description informed by cultural safety and structural competency. Semi-structured interviews were conducted with eleven women between July 2024 and January 2025. Data were analysed inductively using Braun and Clarke’s reflexive thematic analysis. Participants described a repertoire of cultural practices, including herbal and dietary remedies, postpartum rituals, and newborn care customs. Herbal medicine was used to ease labour and promote physiological birth, while cultural nutrition supported recovery and breastfeeding. The extended family played a vital role in postpartum recovery and breastfeeding support. Migration, however, disrupted this communal model, leaving women socially isolated in New Zealand. Participants reported loneliness, lack of family care, and, in some cases, a history of postpartum depression. Despite these challenges, women demonstrated resilience, adapting practices and advocating through transnational family ties and community networks. Participants' cultural practices strongly shape maternity expectations yet often conflict with New Zealand’s individualised model of care. Addressing these gaps requires culturally safe, structurally competent maternity models that integrate positive cultural traditions and reduce the risk of isolation. The next phase of this project describes women’s clinical maternity care experiences, highlights how structural barriers, misdiagnoses rooted in cultural assumptions, and limited recognition of traditional practices further compromise the delivery of woman-centred care
A Hierarchical Federated Continual Learning Framework for Dynamic and Heterogeneous IoV
No full textTraditional federated learning (FL) architectures face challenges in handling heterogeneous data and dynamic tasks, often resulting in catastrophic forgetting when new training tasks are continuously introduced. Federated continual learning (FCL) integrates the privacy-preserving capabilities of FL with the knowledge retention and incremental update mechanisms of continual learning, effectively mitigating catastrophic forgetting and protecting user privacy. However, existing FCL solutions largely overlook the unique requirements of Internet of Vehicles (IoV) scenarios, such as data heterogeneity and dynamic task management. To address these challenges, we propose a novel framework, hierarchical federated continual learning (Hier-FCL), which incorporates local continual learning via optimized experience replay and meta-knowledge distillation, along with dynamic client clustering to tackle data heterogeneity. Additionally, a hierarchical aggregation mechanism is employed to enhance scalability and adaptability in diverse IoV scenarios. Experiments conducted in mixed-task environments using multiple datasets demonstrate that Hier-FCL outperforms baseline algorithms in terms of retained accuracy and backward transfer impact, validating its effectiveness in mitigating catastrophic forgetting and managing heterogeneous client data
Longitudinal, Multi-cycle Evaluation of Passive Function Improvement in People With Arm Spasticity Treated With Botulinum Toxin A
No full textImprovement in passive function (i.e., ease of caring for a limb) is a common goal for treatment of spasticity in the arm with botulinum toxin. A large international, observational, 2-year longitudinal study (ULIS-III, N = 953) was conducted in real-life practice. This original secondary analysis examines whether improvement in passive function goals were met over repeated injection cycles. We report changes by cycle measured by the Passive Function sub-scale of the Arm Activity measure (ArmA-PF) and examine predictors of improvement and injection occurrence. Inclusion in this analysis was based on passive function being selected as a primary or secondary goal for one or more cycle of treatment (n = 542/953). Goals were assessed at the start and end of each cycle using the Goal Attainment Test score and the ArmA-PF. Over all cycles of treatment, goals were set for 1641/2187 injections (75.0%) and achieved in 1250 (76.2%). Significant improvements in ArmA-PF score were identified for at least six cycles (p < 0.001) with evidence of cumulative benefit over successive cycles. This occurred regardless of patient-related baseline characteristics, with the possible exception of some relationship with injection localization techniques. In conclusion, repeated botulinum toxin injections provide significant improvement in passive function, which was sustained over repeated cycles of treatment
Ideological Isolation in Online Social Networks: A Survey of Computational Definitions, Metrics, and Mitigation Strategies
No full textThe proliferation of online social networks has significantly reshaped the way individuals access and engage with information. While these platforms offer unprecedented connectivity, they may foster environments where users are increasingly exposed to homogeneous content and like-minded interactions. Such dynamics are associated with selective exposure and the emergence of filter bubbles, echo chambers, tunnel vision, and polarization, which together can contribute to ideological isolation and raise concerns about information diversity and public discourse. This survey provides a comprehensive computational review of existing studies that define, analyze, quantify, and mitigate ideological isolation in online social networks. We examine the mechanisms underlying content personalization, user behavior patterns, and network structures that reinforce content-exposure concentration and narrowing dynamics. This paper also systematically reviews methodological approaches for detecting and measuring these isolation-related phenomena, covering network-, content-, and behavior-based metrics. We further organize computational mitigation strategies, including network-topological interventions and recommendation-level controls, and discuss their trade-offs and deployment considerations. By integrating definitions, metrics, and interventions across structural/topological, content-based, interactional, and cognitive isolation, this survey provides a unified computational framework. It serves as a reference for understanding and addressing the key challenges and opportunities in promoting information diversity and reducing ideological fragmentation in the digital age.preprin