85000 research outputs found
Sort by
Exploring sex differences in the longitudinal association between streetlighting and transport walking
No full textRegular physical activity, such as transport walking, is essential for health, yet many adults do not meet recommended physical activity levels. The built environment, particularly streetlighting, plays a critical role in promoting transport walking. However, existing evidence on the relationship between streetlighting and transport walking is largely cross-sectional, with limited longitudinal research exploring potential sex differences. This study aimed to assess the longitudinal association between objectively measured streetlight count and transport walking among mid-to-older-aged adults in Brisbane, Australia, and to explore whether this association varies by sex. Data were from the How Areas in Brisbane Influence HealTh and AcTivity (HABITAT) study, a multilevel cohort study with five waves over nine years (2007–2016). The analytical sample included participants who did not move during the study. Transport walking was self-reported and dichotomized into walkers and non-walkers. Streetlight counts within a 1 km road network buffer around participants’ homes were measured alongside transport walking. Generalized linear mixed-effects models were used, adjusting for sociodemographic factors, length of stay, neighbourhood preference, and built environment attributes. Higher streetlight counts were positively associated with transport walking (OR 1.002; 95% CI 1.001, 1.002), and the association remained significant after adjusting for residential density, street connectivity, and land-use mix. No significant sex differences were found. This study provides longitudinal evidence that well-lit environments promote transport walking among mid-to-older-aged adults. This finding can inform urban planning and public health policies aimed at encouraging transport walking to help reduce the risk of chronic disease.</p
Designing an AI Artefact to Detect Greenwashing in Sustainability Reports: A Design Science Research Method
No full textGreenwashing, i.e., the misrepresentation of environmental, social, and governance (ESG) performance,
poses major challenges for regulators, investors, and the public. As sustainability reporting becomes
central to corporate accountability, regulatory scrutiny underscores the need for reliable mechanisms to
detect misleading claims. Existing AI tools are either opaque “black boxes” or lack regulatory and
contextual sensitivity. This study applies a Design Science Research (DSR) methodology to develop an
AI-enabled artefact that detects greenwashing in sustainability reports. Grounded in the Product and
Service Performance Information Quality (PSP/IQ) model, the artefact incorporates explainable AI
techniques such as natural language processing, anomaly detection, and semantic similarity analysis to
evaluate ESG disclosures against standards and stakeholder expectations. The research contributes
design principles, architecture, and evaluation criteria for transparent, regulator-aligned detection
tools. Outcomes include enhanced credibility, accountability, and compliance in ESG reporting,
advancing both academic knowledge and practical solutions to greenwashing in Australia and beyond.</p
Living or Leaving: Life in the Mekong Delta Region of Vietnam
No full textThe Mekong Delta has long been one of Viet Nam’s most vital economic and agricultural hubs. Home to nearly 18 million people and the source of most of the country’s rice, fruit, and fish exports, it has played a central role in national development. Yet its economic weight has diminished as Viet Nam’s growth has shifted away from agriculture. Even so, the region made remarkable strides in reducing poverty in recent decades, buoyed by high agricultural productivity, expanding wage employment in the non-agricultural sector, and rising remittances. That momentum has now faltered. A succession of environmental shocks, both climate-driven and human induced, has eroded these gains and exposed the fragility of the Delta’s development model. This report draws heavily on the extensive body of academic and policy research that has documented the Mekong Delta’s evolving social, economic, and environmental landscape. The analysis builds on these foundational insights to understand the structural forces shaping livelihoods and mobility in the region. These findings are complemented by new and old evidence from the 2024 Life in the Mekong Survey and the Viet Nam Household and Living Standards Survey (VHLSS), which together provide a snapshot of household welfare, migration patterns, and perceptions of risk and opportunity. By integrating established research with new empirical data, the report aims to present a grounded and current account of the transitions underway in the Mekong Delta, anchored in evidence and oriented toward actionable policy recommendations.</p
Stretchable, adhesive and conductive cellulose nanofiber-based for multifunctional wearable electronics
No full textConductive hydrogels have shown significant potential in wearable sensors and energy harvesting due to their distinctive conductive network and excellent flexibility. In this work, a conductive CNF/PAM/PANI (CPP) hydrogel with excellent performance as wearable sensors and supercapacitor was successfully prepared by free radical polymerization, multiple hydrogen bonding and the combination of ionic/electronic dual-conductive network. The incorporation of Li+ has modulated the interaction between the cross-linked networks, endowing the hydrogel with excellent tensile properties (1200 %) and high tensile strength (119 kPa). The synergistic effect of Li+ ion conduction sites and the polyaniline conjugated π-electron system has greatly enhanced the conductivity. This ionic/electronic dual-conductive hydrogel exhibited high sensitivity to tensile and compressive stress, thus the as-fabricated sensors possessed a high sensitivity (GF = 8.95), fast response and recovery time (250 ms and 200 ms) and a wide sensing range (400 %). In addition, the triboelectric nanogenerator based on the CPP conductive hydrogel system has a maximum power density of 69 mW/m2, enabling energy harvesting from human movements to power small devices and monitor joint movements. The flexible supercapacitor with CPP hydrogel as the electrolyte showed a high specific capacitance of 309.41 mF/cm2 at a scanning speed of 0.5 mA/cm2, and the capacitance retention rate during charging and discharging at −20 °C was still 77 %. This novel ionic/electronic dual-conductive hydrogel offers new opportunities for developing wearable electronics and energy harvesting systems.</p
Skeletal High‐Strength Nanoporous Copper and Metamaterials: The Hakka Tulou Design Heritage
No full textNanoporous metals (NPMs) are pivotal for next-generation technologies, yet their inherent mechanical fragility has long hindered practical implementation. Drawing inspiration from the ingenious architecture of the ancient Hakka Tulou walls, a novel class of nanoporous copper (NPCu) materials—skeletal NPCu—is developed to overcome this limitation. To achieve this, we leverage the principles of solidification and dealloying in alloy design to engineer a unique two-phase precursor alloy microstructure. This microstructure comprises a strong, ductile, non-dealloyable skeletal phase (minor phase) and a readily dealloyable principal phase. Upon dealloying, the precursor transforms into skeletal NPCu, achieving exceptional yield strength (200.4 ± 15.2 MPa)—significantly surpassing that of conventional NPMs. Extending this design concept, we fabricate skeletal NPCu lattice metamaterials that surpass the Gibson-Ashby strength model predictions by 800%, while offering an outstanding specific surface area (27.6 ± 1.2 m2 g−1) that enhances multifunctionality. By marrying ancient architectural wisdom with modern materials science, this innovation unlocks the vast potential of advanced NPMs for applications spanning energy, aerospace, and beyond.</p
Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors
No full textLarge language models (LLMs) have achieved remarkable performance across many tasks, yet aligning them with desired behaviors remains challenging. Activation intervention has emerged as an effective and economical method to modify the behavior of LLMs. Despite considerable interest in this area, current intervention methods exclusively employ a fixed steering vector to modify model activations, lacking adaptability to diverse input semantics. To address this limitation, we propose Semantics-Adaptive Dynamic Intervention (SADI), a novel method that constructs a dynamic steering vector to intervene model activations at inference time. More specifically, SADI utilizes activation differences in contrastive pairs to precisely identify critical elements of an LLM (i.e., attention heads, hidden states, and neurons) for targeted intervention. During inference, SADI dynamically steers model behavior by scaling element-wise activations based on the directions of input semantics. Experimental results show that SADI outperforms established baselines by substantial margins, improving task performance without training. SADI's cost-effectiveness and generalizability across various LLM backbones and tasks highlight its potential as a versatile alignment technique.</p
SEED: A Synthesized Multi-Dimensional Environmental Exposure Database for Neighborhood Profiling
No full textThe Synthesized Multi-Dimensional Environmental Exposure Database (SEED) has been developed to address the need for integrated, high-resolution data that captures the complexity of human–environment interactions. While much of the existing literature focuses on single exposures, SEED provides a comprehensive framework for examining the cumulative and interrelated effects of multiple environmental factors on health and well-being.The SEED includes a broad set of spatially harmonized indicators covering physical environmental, urban built environment, socioeconomic environment and population context. Physical environmental exposures are represented through measures of air pollution (e.g., AOD, NO2, CO), urban heat (e.g., land surface temperature), access to green spaces (e.g., normalized difference vegetation index, green view factor and canopy height), and built environment attributes (e.g., POI density, street connectivity). To account for population vulnerability and contextual disparities, socio-demographic variables and ethnicity profiles are systematically integrated. Human perception factors provide street-view–derived human perception indicators that represent the subjective dimension of environmental exposure.All indicators are standardized to a common spatial resolution, allowing fine-grained examination of exposure heterogeneity across neighborhoods. By combining these diverse datasets, SEED facilitates the construction of multidimensional neighborhood profiles that highlight both dominant exposure characteristics and underlying structural patterns. This database provides a robust empirical foundation for advanced spatial modelling, the assessment of environmental inequalities, and the design of policy-relevant interventions aimed at fostering healthier and more equitable urban environments.</p
Multi-objective Optimisation of Two-Way Charging of Electric Vehicles
No full textThe increasing uptake of electric vehicles (EVs) presents both new opportunities and challenges for power networks. Uncontrolled and erratic charging and discharging patterns could introduce multiple challenges to the network, including increased power losses and voltage fluctuations beyond acceptable thresholds. This thesis investigates the challenges of rapid EV adoption on power networks, with a focus on coordinated charging and discharging strategies. A multi-objective optimisation (MOO) framework is developed to balance the interests of EV owners, charging station operators, the environment, and network performance, while integrating a carbon emission reduction incentive program and a dynamic economic dispatch model. Furthermore, the research addresses long-term planning for large-scale EV integration by optimising the timing, location, and capacity of electric vehicle charging station (EVCS) and battery energy storage system (BESS) investments.
First, an MOO problem is formulated to balance three key objectives: maximising EVCS profit, minimising network power losses, and reducing EV owners' costs. The EV charging/discharging model integrates diverse load profiles from various EV user categories, reflecting their distinct driving patterns. The analysis also examines the effects of varying charging/discharging rates, as well as the impact of load profiles and Photovoltaic (PV) generation uncertainties on costs, power losses, and EVCS profitability. The proposed framework is applied to a sample network under different scenarios, utilising Non-dominated sorting genetic algorithm-II (NSGA-II) to identify trade-offs between objectives. This approach allows the system to optimise power resource allocation, manage charging and discharging cycles, and improve the operational efficiency of EVCS. Furthermore, coordinated EV charging through vehicle-to-grid (V2G) and grid-to-vehicle (G2V) technologies provides substantial economic benefits and mitigates network issues associated with unmanaged EV charging.
Second, two programs are introduced, each serving a distinct purpose: the first aims to reduce overall stakeholder costs and compensate EV users for expenses incurred when participating in V2G services, while the second seeks to lower carbon emissions and address broader environmental objectives. The contribution is twofold: existing studies have not applied compensation mechanisms to minimise charging/discharging frequency in MOO problems, creating a gap that this work addresses, and the proposed programs collectively reduce costs for stakeholders while incentivising V2G participation by offering financial benefits to both EV users and EVCS owners. The key findings demonstrate that the proposed scheduling method increases participation in V2G services by over 10%, boosts EVCS benefits by more than 20%, and reduces network losses, while higher charging/discharging rates combined with greater carbon-revenue benefits help offset battery degradation costs. Furthermore, since the proposed planning balances EVCS profit with network loss costs, a trade-off is established along the Pareto non-dominated front, providing a spectrum of near-optimal solutions between these conflicting objectives. To support decision-making, a fuzzy metric is introduced to incorporate weighted preferences, enabling stakeholders to select the most suitable solution based on their priorities. This integrated approach simplifies complex decision processes, enhances transparency and flexibility in planning, and empowers stakeholders—including DNSPs, DNOs, policymakers, and investors—to evaluate trade-offs effectively, ultimately supporting more efficient investment strategies, improving EV infrastructure integration, and ensuring scalable, economically viable grid expansion aligned with future EV adoption trends
Lastly, this study introduces a dynamic expansion planning approach to tackle the technical challenges of a large-scale transition to an expanded EV fleet and the integration of renewable energy sources (RESs). The contribution is twofold: first, it addresses long-term planning by determining the optimal locations and capacities of EVCSs annually over the planning horizon, including the timing of constructing or upgrading existing infrastructure and the pace of EV adoption. Additionally, a technique is proposed based on drivers' travel patterns to estimate the distribution of EVs across different areas and calculate their annual growth rate. The main finding is that the EV charging/discharging schedule can significantly impact the total capacity of EVCS over the planning period. This not only enhances EVCS benefits by 25% over a decade but also reduces overall costs by preventing unnecessary capacity expansion or construction delays by 5%. Moreover, increased charging/discharging rates, combined with cost-reduction programs for EV users and EVCS, help to better offset investment costs by reducing the required size and optimising the location of EVCS.</p
Queerbaiting Strategy in Vietnamese and Taiwanese cinemas: An analysis of Song Lang and A Balloon’s Landing
No full textThis study analyzes the queerbaiting phenomenon that expresses an ambiguous queer relationship without confirmation in two films, Song Lang (Vietnam) and A Balloon’s Landing (Taiwan). Combining narrative analysis, visual discourse, and fan community feedback, the research indicates that elements (light, dialogues, and character motivations) are used to evoke romantic emotions between male characters. In a different sociopolitical context, Song Lang reflects the reticence of a conservative environment through cai luong , dim lighting, and silence. A Balloon’s Landing represents freer expression after Taiwan’s legalization of same-sex marriage but still maintains ambiguity to appeal to wider audiences. Findings show that queerbaiting is not simply a narrative device, but a cultural strategy influenced by law, censorship, market expectations, and active viewer interpretation. The study expands the queerbaiting framework into East and Southeast Asian context, emphasizing the role of meaning-making by the audience community in shaping a contemporary Asian queer cinema space.</p
A Comparative Analysis of Linguistic and Retrieval Diversity in LLM-Generated Search Queries
No full textLarge Language Models (LLMs) are increasingly used to generate search queries for various Information Retrieval (IR) tasks. However, it remains unclear how these machine-generated queries compare to human-written ones, particularly in terms of diversity and alignment with real user behavior. This paper presents an empirical comparison of LLM- and human-generated queries across multiple dimensions, including lexical diversity, linguistic variation, and retrieval effectiveness. We analyze queries produced by several LLMs and compare them with human queries from two datasets collected five years apart. Our findings show that while LLMs can generate diverse queries, their patterns differ from those observed in human behavior. LLM queries typically exhibit higher surface-level uniqueness but rely less on stopword use and word form variation. They also achieve lower retrieval effectiveness when judged against human queries, suggesting that LLM-generated queries may not always reflect real user intent. These differences highlight the limitations of current LLMs in replicating natural querying behavior. We discuss the implications of these findings for LLM-based query generation and user behavior simulation in IR. We conclude that while LLMs hold potential, they should be used with caution.</p