192815 research outputs found
Sort by
A Multi-modal Large Language Model with Graph-of-Thought for Effective Recommendation
Get PDFChain-of-Thought (CoT) prompting has been shown to be effective in guiding Large Language Models (LLMs) to decompose complex tasks into multiple intermediate steps, and constructing a rational reasoning chain for inferring answers. However, the linear nature of CoT falls short from enabling LLMs to effectively handle graph structures, which are essential for personalized recommendation tasks that rely on user-item interaction graphs. To bridge this gap, we introduce GollaRec, which leverages a Graph-of-Thought (GoT) prompting technique in a Multi-modal LLM, namely LLaVA, to effectively exploit the complex structure of the interaction graphs. GollaRec enhances the recommendation effectiveness by integrating both visual and textual “thoughts” into a graph-structured prompt, using both item images and descriptions to produce richer multi-modal user/item representations. In our proposed approach, GollaRec leverages text-graph alignment and graph instruction tuning to allow the Multi-modal LLM to capture complex graph structures. In addition, GollaRec leverages a graph adaptor to integrate user-item interactions into the resulting user/item embeddings, therefore effectively adapting the model to the recommendation task. Our extensive experiments on 6 benchmark datasets demonstrate the superiority of our proposed GollaRec model over 12 existing state-of-the-art models in various multi-modal recommendation tasks, including general and multi-domain recommendation tasks
Cultivating a space for learning: a study of note-taking and sharing on a video-sharing platform Bilibili
Get PDFVideo-sharing platforms such as YouTube are popular today, not only for entertainment but also for learning, e.g., for understanding lecture content or acquiring everyday skills. Some platforms, like Bilibili, a popular video-sharing platform in China, even introduced note-taking and sharing features to further foster such learning practices. While note-taking and sharing have been extensively explored in environments specifically for learning, their use around these social media platforms such as YouTube and Bilibili remains under-studied. In this paper, we present a qualitative study with 15 participants who have used the note-taking and sharing feature called BNote launched on Bilibili. Our study reveals how note-taking and sharing are used as a way to cultivate a space for learning on such a general video-sharing platform by structuring, supplementing, and substituting user-generated videos and by leveraging social forces for motivation, self-discipline, and organization for learning. We end by discussing our findings and providing design recommendations to better support a learning space on social media platforms
Does team size affect Scottish male academy soccer player technical, locomotor and psychosocial outcomes during age and maturity bio-banded small-sided games
Get PDFThe timing of the adolescent growth spurt (i.e. peak height velocity [PHV]) varies greatly, causing significant differences in maturity-related anthropometric and physical development among chronologically age-categorised youth soccer players. These differences can lead to a maturity-selection bias favouring early maturing players. Using fifty-three players from two professional Scottish soccer academies, this study examined the effects of team size (4v4, 5v5, and 6v6) and bio-banding (i.e. grouping players by maturity status) on players technical (i.e. touches, releases, possessions), physical (i.e. distance covered, accelerations/decelerations), and psychological (i.e. confidence, competitiveness, positive attitude) performance during small-sided games, using foot-mounted inertial measurement units (F-IMU) and coach observations. Data were analysed using multivariate ANOVA. During maturity-matched games, technical actions decreased as team size increased. More mature players displayed greater technical actions in 4v4 compared to 6v6. Maturity-mismatched games revealed significant technical action differences for less mature players, particularly in 4v4 formats. More mature players covered greater high-intensity distances in maturity-matched and mismatched 6v6 games. Psychological scores were higher for more mature players in smaller team sizes (4v4), whereas less mature players showed consistent psychological scores across all formats. Smaller team sizes and maturity-matched bio-banding formats enhanced technical actions and psychological characteristics, particularly for more mature players
Health inequalities and contemporary youth: young people’s accounts of the social determinants of health in an 'austere meritocracy'
Get PDFYoung people coming of age amidst widespread socioeconomic uncertainty have a unique vantage from which to interpret how social, economic and environmental factors might influence health and the generation of health inequalities. Despite this, only a small number of existing studies of ‘lay’ understandings of health inequalities have focused on young people. This arts-based qualitative study builds on that body of research, in the context of the UK, to explore how young people make sense of health inequalities. Across two sites, Glasgow and Leeds, six groups of young people (39 in total) took part in online workshops to explore their perspectives. Throughout they engaged with population health research evidence; contributed to group discussions and responded creatively, via visual and performance art and by articulating their own views and experiences. In this paper, we explore how individual and structural explanations for health inequalities emerged, employing concepts from sociological studies of youth to shed light on these accounts. In particular, we argue that the concept of ‘austere meritocracy’, the persistence of narratives of aspiration and hard work as key to success against an increasingly hostile socio-economic backdrop, helps explain young people’s perspectives on health inequalities in the UK
Embracing inclusivity and flexibility in higher education: a study of Hyflex delivery and Its impact on diverse undergraduates in the post-pandemic era
Get PDFIn the wake of the COVID-19 pandemic, higher education has undergone a significant transformation, necessitating a re-evaluation of pedagogical approaches to meet the evolving needs of a diverse student body. This article delves into these changes, with a particular emphasis on the adoption and impact of Hybrid Flexible (Hyflex) delivery models in higher education. Conducted at a Scottish university, this study focuses on a mandatory management accounting course for accounting and finance students, exploring how the Hyflex model aligns with principles of inclusivity and Universal Design for Learning.
The research examines student preferences for Hyflex education and its influence on academic performance, paying special attention to how these preferences and outcomes vary among students from different socio-economic backgrounds. The findings indicate that while a majority of students exhibit a preference for a blend of online and on-campus attendance, there is a notable inclination towards online engagement among students from more deprived areas. Factors such as travel challenges, work-life balance, and anxiety are identified as significant determinants in the choice of remote learning.
Additionally, the study reveals a moderate correlation between students' socio-economic status and their exam performance, with varying results across different modes of delivery. This suggests that while Hyflex education offers a flexible framework, its effectiveness and accessibility can differ based on individual student circumstances.
While further research is required for more definitive conclusions, this study contributes valuable insights into the post-pandemic educational landscape. It underscores the importance of adopting flexible and inclusive educational pathways, such as the Hyflex model, which not only adapt to technological advancements but also uphold principles of equity and student engagement, catering to the diverse needs of contemporary undergraduates
Narrow linewidth distributed feedback lasers utilizing distributed phase shift
Get PDFThis study proposes and experimentally demonstrates a distributed feedback (DFB) laser with a distributed phase shift
(DPS) region at the center of the DFB cavity. By modeling
the field intensity distribution in the cavity and the output spectrum, the DPS region length and phase shift values
have been optimized. Experimental comparisons with lasers
using traditional π-phase shifts confirm that DFB lasers
with optimized DPS lengths and larger phase shifts (up to
15π) achieve stable single longitudinal mode operation over a
broader current range, with lower threshold current, higher
power slope efficiency, and a higher side mode suppression
ratio (SMSR). Furthermore, the minimum optical linewidth
is reduced significantly, from 1.3 MHz to 220 kHz
Effects of GLP-1 receptor agonists on kidney and cardiovascular disease outcomes: a meta-analysis of randomised controlled trials
Get PDFBackground: GLP-1 receptor agonists reduce the risk of major adverse cardiovascular events (MACE) and can also have
kidney benefits. However, whether GLP-1 receptor agonists improve clinically important kidney outcomes remains
uncertain. We aimed to comprehensively assess the effects of GLP-1 receptor agonists on kidney and cardiovascular
disease outcomes by performing a meta-analysis of randomised controlled trials.
Methods: For this meta-analysis, we searched MEDLINE, Embase, and the Cochrane Central Register of Controlled
Trials for randomised controlled trials that included at least 500 participants with type 2 diabetes, compared a GLP-1
receptor agonist with placebo with at least 12 months of follow-up, and reported a primary clinical kidney or
cardiovascular outcome, from database inception to March 26, 2024. Post hoc, we included the SELECT trial
(NCT03574597), which enrolled participants with cardiovascular disease and a BMI of 27 kg/m² or more without
diabetes. Study-level summary data were extracted independently by two authors for inclusion in this random-effects
analysis. The main kidney outcome was a composite outcome, consisting of kidney failure (kidney replacement
therapy or a persistent estimated glomerular filtration rate [eGFR] <15 mL/min per 1·73 m²), a sustained reduction in
eGFR by at least 50% or the nearest equivalent, or death from kidney failure. The main cardiovascular outcome was
MACE, consisting of cardiovascular death, non-fatal myocardial infarction, or non-fatal stroke. This study is registered
with PROSPERO, CRD42024528864.
Findings: Of the 5140 records identified through the literature search, 11 trials, involving 85 373 participants
(29 386 female, 55987 male), were included in the meta-analysis. In participants with type 2 diabetes (67769), GLP-1
receptor agonists reduced the composite kidney outcome by 18% compared with placebo (hazard ratio [HR] 0·82,
95% CI 0·73–0·93; I²=26·41%), kidney failure by 16% (HR 0·84, 0·72–0·99; I²=0%), MACE by 13% (HR 0·87,
0·81–0·93; I²=49·75%), and all-cause death by 12% (HR 0·88, 0·83–0·93; I²=0%). The effect on the composite kidney
outcome (HR 0·81, 95% CI 0·72–0·92; I²=23·11%), kidney failure (HR 0·84, 0·72–0·98; I²=0%), MACE (HR 0·86,
0·80–0·92; I²=48·9%), and all-cause death (HR 0·87, 0·82–0·91; I²=0%) was similar when the SELECT trial was
included, with no evidence of heterogeneity between this trial and those including participants with type 2 diabetes
(pheterogeneity >0·05). There was no difference in the risk of serious adverse events, including acute pancreatitis and severe
hypoglycaemia, between the GLP-1 receptor agonist and placebo groups (risk ratio [RR] 0·95, 95% CI 0·90–1·01;
I²=88·5%). However, treatment discontinuation due to adverse events occurred more frequently in the GLP-1 receptor
agonist groups (RR 1·51, 95% CI 1·18–1·94; I²=96·3%).
Interpretation: We found evidence that GLP-1 receptor agonists significantly reduce clinically important kidney events,
kidney failure, and cardiovascular events.
Funding: Non
Zirconium–based MXenes: synthesis, properties, applications, and prospects
Get PDFZirconium (Zr) based MXenes are a new type of two–dimensional (2D) transition metal carbides and carbonitrides that have attracted significant research interest in recent years. These materials exhibit a unique combination of physicochemical properties, making them attractive for a wide range of applications. Despite this growing attention, a systematic review of their synthesis methods, material properties, and applications is still lacking. This review provides a comprehensive overview of the state of research on Zr–MXenes, covering various aspects from synthesis to applications. The discussion includes an in–depth analysis of the different wet–chemical etching protocols used to obtain Zr–MXenes from Zr–containing MAX phases and the impact of these methods on the morphology of materials. Detailed characterization techniques have revealed important properties of Zr–MXenes, such as hydrophilicity, electrical conductivity, and ion storage capability. Further, this review examines the potential applications of Zr–MXenes in various fields, including energy storage, electromagnetic interference shielding, corrosion prevention, and biomedical applications. While, Zr–MXenes offer promising prospects, challenges related to large–scale production and property optimization must be addressed to facilitate their widespread adoption. By providing a comprehensive overview of Zr–MXene synthesis, properties, and applications, this review aims to inspire and guide future research and development efforts toward the rational design and utilization of these promising 2D nanomaterials
Toward a generic model for phosgene synthesis catalysis over activated carbon
Get PDFA reaction model was recently been postulated to describe the phosgene synthesis process over an activated carbon catalyst. That body of work exclusively considered a single grade of activated carbon (Donau Supersorbon K40), raising the issue that the mechanistic insight gleaned could be restricted to the single substrate. Therefore, to evaluate the generality of the proposed sequence of elementary reactions that lead to phosgene production, this article concentrates on Norit RX3 Extra, a representative activated carbon that finds wide application in industrial applications. Several phosgene synthesis traits were associated with Donau Supersorbon K40 and are reproduced with Norit RX3 extra over the reaction conditions studied, such as exclusive phosgene selectivity and a degree of reagent retention. However, temperature-programmed reaction studies show the reaction profile of the Norit RX3 extra to exhibit deviations from that observed for Donau Supersorbon K40. A modified reaction model is proposed to account for these differences. The revised reaction model is thought to represent a generic model for phosgene synthesis that can accommodate disparities in the surface chemistry of different activated carbon formulations
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! 👇