Sydney eScholarship
Not a member yet
31878 research outputs found
Sort by
Freshwater-induced coral bleaching following extreme rainfall in Pioneer Bay, Orpheus Island, Great Barrier Reef
No full textThis dataset provides benthic cover and coral assemblage data from three sites at Pioneer Bay, Orpheus Island, assessed across three time points (July 2023, July 2024, and April 2025
Hearing Educator Perspectives: From ‘Evidence-Based Practice’ to Valuing the ‘Enriched Evidence-Based Practice’ of Education
No full textThe intention of this discussion paper is to open conversation with educators to raise awareness to the problems of evidence-based practice, the value in their nuanced and complex work to rebuild teacher professionalism across the community, and provide support for educators to enrich their use of evidence to inform practice. In opening conversation on evidence-based practice, we sought to hear from educators to understand how their voices and experiences may be valued in moving forward in the use of evidence to inform schooling practice
Three Essays in Advanced Graph Neural Networks: From Multivariate Temporal Forecasting to Heterophilous Graph Learning
No full textIn recent decades, the rapid advancement of computing power and data collection techniques has led to an unprecedented surge in structured and semi-structured data, necessitating the development of deep learning techniques to extract meaningful insights. While traditional deep learning methods have demonstrated remarkable power across various domains, they struggle to model the complex inter-relational structures inherent in real-world data, such as graphs. Graph Neural Networks (GNNs) have emerged as a powerful framework to address this limitation, enabling the effective representation of interdependent entities and achieving success in numerous real-world applications.
However, several challenges remain for GNNs, particularly in areas such as multivariate time series forecasting and heterophilous graph learning. For instance, while GNNs have been widely applied to multivariate time series forecasting, their potential in probabilistic forecasting remains under-explored. Most existing GNN-based forecasting models focus on deterministic predictions, failing to quantify predictive uncertainty, which is crucial for decision-making. Developing a GNN-based framework that effectively integrates probabilistic modelling into temporal forecasting remains an open challenge.
Additionally, GNNs are often incorporated into time series models to capture inter-series dependencies, while intra-series temporal dependencies are typically modelled separately using temporal models. However, in real-world systems, intra- and inter-temporal dependencies are inherently intertwined, making it challenging to fully exploit complex entangled relationships using existing methods. This raises the question of whether time series data can be transformed into a pure graph representation to unify these dependencies under a pure graph paradigm..
Computational Measure & Evaluation Of Visual Information In Communication Artefacts In Urban Design
No full textDesign visualisations do not often convey information efficiently to stakeholders in urban design. While visual artefacts are intended to convey key aspects of a project, the process of tailoring them to support efficient collaboration and decision-making, particularly in the early phase of urban design, remains largely underexplored. A visual communication artefact is an object containing graphical information used in design for visual communication. This can be plans, sections, 2D, 3D or higher dimensional graphical information. Artefacts are constructed and manipulated graphical data. However, there is no universally accepted standard for the appropriate level of detail, leading to continued reliance on traditional formats. This results in a mismatch between information and needs of stakeholders. Despite ongoing advancements in information and communication technology, artefacts are often treated as standard one-size-fits-all tools. Without a deliberate process for calibrating the level and types of detail suitable for different stakeholders, these artefacts may present excessive or irrelevant data, delaying shared understanding, particularly when priorities, technical expertise or perspectives vary widely. As such, this research presents an algorithmic system for measuring visual information and evaluating this based on its fitness to meet the specific information requirements of individual stakeholders. The research highlights inefficiencies in current practice and aims to reduce these by exploring how graphical information can be more aligned with stakeholder needs. Additionally, the research contributes to emerging areas of inquiry focused on measuring aesthetics, and analysing and evaluating graphical data using computation. Aesthetics in this research leans on George Stiny and James Gips’ definition in Algorithmic Aesthetics to describe how a piece of work can be descried, interpreted and evaluated and with questions about how new works can be created
Towards transparent and replicable flower selection in agricultural flower strips
No full textFlower strips are increasingly employed as ecologically sustainable methods for attracting beneficial insects to agricultural landscapes. Since flowers differ substantially in an array of characteristics including morphology, nutritional content, and phenology, selecting ‘the right flowers for the right job’ requires the designer to make careful decisions about which flowers to include. These decisions ultimately determine the success of flower strips. Given the importance of flower selection, how do researchers choose which flowers to plant? We conducted a quantitative content analysis to examine how researchers make flower selection decisions in studies aimed at supporting or attracting natural enemies or pollinators to target crops. We collected data on the criteria used to justify selection decisions and we recorded the type of evidence (experiential evidence such as “expert recommendations” or “personal communication” vs peer reviewed evidence) used to support selection decisions. We found that the majority of studies relied on experiential data to inform their flower choices. Specific selection criteria varied, but 'attractiveness' and 'native to region' were most frequently cited for both natural enemies and pollinators. Overall, we found that the prevalent use of experiential evidence for flower selection may limit the replicability and transparency of studies. To address this concern, we introduce a flexible framework aimed at ensuring the clear communication of the flower selection process. Our framework emphasizes evidence-based practices to enhance both replicability and transparency in flower strip design, offering a path toward more rigorous and effective ecological interventions.Paper selection and screening
Between August and September 2023, we searched Web of Science for studies that investigated the use of supplemental flower plantings using the following search terms in the ‘Topic’ field:
Flower strip OR insectary insectary OR wildflower planting wildflower planting Or wildflower strip wildflower strip OR Conservation biological control
Conservation biological control OR floral enhancement
We refined results to the subjects entomology, zoology, animal ecology, crop science, crop protection, agriculture multidisciplinary horticulture, environmental studies, plant sciences, agronomy, biodiversity conservation, and agricultural policy.
Exclusion criteria
Since we were interested in contemporary approaches to flower strip design, we restricted our search to papers published in or after the year 2000. We excluded studies that focused on woody plantings such as trees and hedges as we were interested in studies that used perennial or annual flowers. We only included papers which provided a list of flower species (or which had a citation to a species list). This led to the exclusion of studies that provided only the name of a flower seed mix, without access to the list of species therein.
We only included studies that were targeted at increasing pollinator or pest control services in an adjacent crop and we excluded studies that only sampled insect biodiversity in the general landscape (rather than on a specific crop). While such papers are of high interest, the selection criteria for plantings where the primary goal is biodiversity enhancement will likely differ from those being selected for the explicit role of promoting beneficial insects.
We excluded papers that examined the impact of natural flower diversity, such as the impact of crops being adjacent to native forest or bushland again as we were interested in plantings done specifically for the purposes of attracting natural enemies to crops Note that we did not screen for study quality as our goal was to understand the flower selection strategy and evidence base, rather than formally assessing intervention efficacy.
For each paper, we collected general information about the study such as the country and continent the study was conducted in, the number of flower species used in the supplemental planting, and the target functional group (pollinators or natural enemies, where ‘natural enemies’ includes both parasitoids and predators). Studies that investigated both pollinators and natural enemies were classified as “both”
Achieving High-Performance Dynamic Wireless Charging through Model-Free Predictive Control: Theoretical Insights and Practical Implementations
No full textDynamic Wireless Charging (DWC) for Electric Vehicles (EVs) is constrained by a critical, dual-sided
control challenge. A receiver-side voltage regulation bottleneck, caused by severe disturbances,
severely hampers the transmitter's ability to achieve Maximum Energy Efficiency Tracking (MEET) on
a non-monotonic landscape, especially at high speeds. This thesis develops a systematic solution
using Model-Free Predictive Control (MFPC), advancing both its theory and practical implementation
for DWC systems.
First, a novel unified modelling framework is developed to analyze and generalize MFPC strategies.
This systematic tool predicts controller behavior, expedites the design process, and reveals a new
structural design criterion for enhancing performance robustness.
Second, a specific MFPC controller, MFC1, is proposed for the power receiver to eliminate the
voltage regulation bottleneck. Small-signal analysis and comprehensive experiments validate that
MFC1’s cascaded control structure inherently enhances disturbance rejection compared to
Proportional-Integral (PI) control.
Finally, a multi-layered adaptive MFPC (ad-MFPC) is developed for ultra-fast MEET. It overcomes
the fundamental limitations of standard MFPC for non-monotonic tasks by integrating two key
innovations: an adapted Ultra-Local Model (ULM) with gradient-sign detection and a dual-factor
adaptive step-size mechanism. Experimental validation demonstrates the superiority of the ad-MFPC
in achieving both rapid transient and robust performance
Identities in Practice and Discourse: Transnational English Language Teachers
No full textThis thesis examines the language teacher identities (LTIs) of transnational teachers who are completing a language teacher education (LTE) program in Australia. Using an integrated framework of identity-in-practice and identity-in-discourse based on Trent (2015) and Varghese et al. (2005), the thesis, primarily, explores how these teachers represent their LTIs and, secondarily, whether and how teacher educators integrate identity into their teaching.
Three qualitative studies were conducted. The first study employed photo-elicitation to explore the LTIs of teachers in terms of claimed, assigned, practised and imagined identities. The second study employed the talk-around-text method to investigate how the teachers discursively represented their LTIs in their written assignments. The third study used interviews to examine the teacher educators’ integration of identity into their teaching.
The first two studies revealed that the LTIs of the transnational teachers – such as research-informed and empowered multilingual teachers – were highly idealised, likely influenced by common educational discourses, such as evidence-based teaching and multilingualism. Yet, tensions emerged in their LTI representations due to perceived contradictions between teacher preparation in Australia and expectations in their teaching contexts. The third study found three possible levels of identity integration in LTE: explicit, implicit and theoretical. This study also found three practices emphasised by the educators: bridging theory and practice, shifting away from deficit-based discourses, and developing critical reflexivity. Although not necessarily intended as identity pedagogy, these practices were perceived as contributing to the LTIs of teachers. These findings are discussed in light of calls to make LTI a leitmotif in LTE. Based on the findings from the three studies, the thesis presents an extension to Trent’s (2015) framework of identity-in-practice and identity-in-discourse
NETS NSW Paediatric Trauma Inter-hospital Retrieval: A 10-year review of patient outcomes
No full textIn New South Wales (NSW) there are multiple trauma transport providers. With only 3 paediatric trauma centres (PTC), nearly half of paediatric major trauma cases require inter-hospital transfer to reach definitive care. International literature reports reduced hospital length of stay (LOS), reduced mortality and reduced adverse events during transport for paediatric trauma patients retrieved by paediatric specialist teams compared to adult teams and paramedics. This study examined the quality of care delivered to paediatric trauma patients requiring inter-hospital transfer in NSW.
This is a retrospective observational study of trauma patients aged 0-15 years, referred to the Newborn and paediatric Emergency Transport Service (NETS NSW) and were medically retrieved or transported to a PTC between 1 January 2010 and 31 December 2019. The primary outcome was acute morbidity, measured by admission to paediatric intensive care unit (PICU), and lengths of stay in PICU and hospital.
Nine-hundred and thirty-two patients were included in the final analysis. We found no difference in PICU admission rate; PICU LOS; or hospital LOS between the paediatric retrieval team (NETS NSW) and the adult medical retrieval service (AMRS). There was a higher PICU admission rate for the regional retrieval service in this study, and this was associated with a shorter PICU and hospital LOS. NETS NSW teams on average spent the same amount of time stabilising patients at the referring hospital as AMRS teams. However, this study identified potentially modifiable delays related to teams reaching the patient’s and in NETS NSW teams’ departure once the patient was stabilised.
Within the current trauma system, the inter-hospital transport process and providers are delivering an equitable level of service as measured in mortality, morbidity and time-based outcomes. However, opportunities exist to improve upon the medical retrieval response times
The Context of Voice: How Group, Organisational and Digital Contexts Shape Employee Voice
No full textEmployee voice is vital for organisational improvement, yet most research has focused on dyadic interactions, typically between employee and manager. This thesis adopts a broader view of voice as a multilevel, contextually embedded phenomenon shaped by social actors, group dynamics, and digital affordances. Three empirical studies build sequentially to advance this perspective. Study 1, guided by role theory (Biddle, 1968), reviewed 268 articles (370 studies) and identified six social actors: voicers, endorsers, solicitors, implementers, allies, and bystanders. These social actors play an important role in shaping the voice process. My findings highlight a broader a system of interdependent actors whose interactions influence whether, how and when voice is heard, endorsed or implemented. Building on this, Study 2 examined voice in shared group contexts using Parker et al.’s (2010) proactive motivational states framework. An experimental vignette study tested how information redundancy and group diversity affect promotive and prohibitive voice. Building on previous research, I found that information redundancy had significant indirect effects on both voice types via diffusion of responsibility. There was a significant moderated mediation such that group diversity moderated the influence of information redundancy on both voice types via voice self-efficacy. Extending to a broader shared context, Study 3 applied affordance theory (Khan et al., 2023) and organisational social capital theory (Leana & van Buren, 1999) to investigate e-voice at the organisational level. Using machine learning, I analysed 347,305 Glassdoor posts from 301 U.S.-based organisations. Higher e-voice quantity and positive sentiment predicted stronger organisational performance, mediated by positive word of mouth. Together, these studies show voice is a multilevel process shaped by roles, group dynamics and digital platforms, offering new theoretical and methodological insights for enabling voice
Targeted foundational supports for people with psychosocial disability outside of the NDIS: model proposal
No full textTo provide a draft model for delivering targeted foundational supports for people with psychosocial disability1 who are outside the NDIS. The aim is to offer a model derived from a research perspective as a basis for critique, refinement and to inform the wider sector’s approach to foundational supports