Monash University Research Portal
Not a member yet
41460 research outputs found
Sort by
Vulgarity in online discourse around the English-speaking world
No full textThis paper takes a corpus-based approach to study vulgar language in online communication across 20 English-speaking regions based on the Global Web-Based English Corpus (GloWbE). The identification of vulgarity combines word lists used in profanity detection with regular expressions to identify a wide range of vulgar elements including spelling variants and obscured forms. The results show a notable trend for inner circle L1-varieties to exhibit higher rates of vulgarity online compared to outer circle and L2-varieties. The results also show that inner circle varieties have lower adapted corrected type-token rations which indicates that inner circle variety speakers use more varied English vulgar forms compared with speakers from other circle varieties. In addition, there is a general register difference with vulgarity being more common in blog data compared with general web content. Finally, the results show that different regions exhibit preferences for specific vulgar lemmas feck being preferred in Ireland, cunt, in Britain, and ass(hole) in the United States. The findings are interpreted to show that cultural differences are reflected in region-specific preferences for vulgarity and that the creativity observed in inner circle varieties is linked to norm-setting compared to norm-reception associated with outer circle varieties.</p
Psychometric analysis of the Italian Doomscrolling Scale:Associations with problematic social media use, psychological distress, and mental well-being
No full textDoomscrolling is a fairly new concept in mental health research which has attracted significant attention in recent years. Doomscrolling involves individuals spending excessive time online reading unpleasant news, and leading to negative emotional states (e.g., sadness, anxiety, anger, etc.). Several studies have found that doomscrolling is associated with lower quality of life, poorer mental well-being, and problematic technology use. In Italy, there is a lack of instruments to assess this construct. Therefore, the main objective of the present study was to validate both the 15-item and four-item Doomscrolling Scale (DSS) to fill this gap. The sample comprised 300 Italians (70.7% females), with a mean age of 38.02 years (SD = ± 13.08). Participants completed an online survey comprising the DSS, Depression Anxiety Stress Scale (DASS-21), Warwick-Edinburgh Mental Well-Being Scale (WEMWBS), Satisfaction With Life Scale (SWLS), and Bergen Social Media Addiction Scale (BSMAS). The results of confirmatory factor analysis supported a first-order one-factor scale, with satisfactory fit indices. The DSS showed good internal consistencies (Cronbach alpha = 0.96 and McDonald omega = 0.96). Additionally, the DSS score was positively associated with scores on the BSMAS and DASS-21, and negatively associated with scores on the WEMWBS and SWLS. The short version of the DSS also demonstrated very good psychometric characteristics. The findings indicate that the both versions of the DSS are psychometrically reliable and valid measures for assessing doomscrolling activity among Italian adults. The study expands the literature regarding factors related to doomscrolling behavior.</p
Combating confirmation bias:a unified pseudo-labeling framework for entity alignment
No full textEntity alignment (EA) aims at identifying equivalent entity pairs across different knowledge graphs (KGs) that refer to the same real-world identity. It has been a compelling but challenging task that requires the integration of heterogeneous information from different KGs to expand the knowledge coverage and enhance inference abilities. To circumvent the shortage of prior seed alignments provided for training, recent EA models utilize pseudo-labeling strategies to iteratively add unaligned entity pairs predicted with high confidence to the seed alignments for model training. However, the adverse impact of confirmation bias during pseudo-labeling has been largely overlooked, thus hindering entity alignment performance. To systematically combat confirmation bias, we propose a new Unified Pseudo-Labeling framework for Entity Alignment (UPL-EA) that explicitly alleviates pseudo-labeling errors to boost the performance of entity alignment. UPL-EA achieves this goal through two key innovations: (1) Optimal Transport (OT)-based pseudo-labeling uses discrete OT modeling as an effective means to determine entity correspondences and reduce erroneous matches across two KGs. An effective criterion is derived to infer pseudo-labeled alignments that satisfy one-to-one correspondences; (2) Parallel pseudo-label ensembling refines pseudo-labeled alignments by combining predictions over multiple models independently trained in parallel. The ensembled pseudo-labeled alignments are thereafter used to augment seed alignments to reinforce subsequent model training for alignment inference. The effectiveness of UPL-EA in eliminating pseudo-labeling errors is both theoretically supported and experimentally validated. Our extensive results and in-depth analyses demonstrate the superiority of UPL-EA over 15 competitive baselines and its utility as a general pseudo-labeling framework for entity alignment.</p
Greening up their act:corporate carbon emissions reduction in response to political risk
No full textThis study explores whether firms decrease their carbon emissions in response to heightened political risk. Political risk often imposes significant economic consequences, prompting companies to engage in costly strategies such as political lobbying. Alternatively, firms may opt to manage their emissions to partially mitigate political risk. We posit that high carbon emissions can attract unwanted political scrutiny, making emissions reduction a strategic response to escalating political risk. Our findings demonstrate that firms indeed lower their emissions in response to political risk. Moreover, we find that specific firm characteristics can affect the relationship between political risk and carbon emissions. Finally, we show that firms achieve emissions reductions through investments in environmental innovation. To investigate robustness, we employ multiple identification strategies, including entropy balancing, instrumental variable analysis, and quasi-natural experiments. Our findings highlight the strategic importance of emissions reduction as a response to political risk, offering insights for policymakers, investors, and corporate decision-makers on aligning sustainability efforts with risk management strategies.</p
Unraveling the mechanisms and effectiveness of AI-assisted feedback in education:A systematic literature review
No full textRapid advancements in Artificial Intelligence (AI) have prompted growing interest in leveraging AI for educational feedback. Yet, the centrality of the learner in this process is often overshadowed by technological excitement, and a broad understanding of AI-assisted feedback (AIFB) in education remains evolving. To address this gap, we conducted a systematic review of 129 peer-reviewed journal articles (2014–2023) based on widely used AI-related search terms to examine how AI, especially generative AI, supports feedback mechanisms and influences learner perceptions, actions, and outcomes. Our analysis identified a sharp rise in AIFB research after 2018, driven by modern large language models. We found that AI tools flexibly cater to multiple feedback foci (task, process, self-regulation, and self) and complexity levels (basic, intermediate, and elaborated). Our findings demonstrate that AIFB can effectively enhance targeted learning outcomes. By employing a transparent and field-aligned methodology, we synthesized recent advances and offers actionable insights for both research and practice. While the focus on widely recognized AI-related search terms ensures strong comparability and relevance, some specialized subfields (e.g., Automated Writing Evaluation), are less prominent in this synthesis. The study also highlights the ongoing need for clearer reporting of underlying AI algorithms. Building on these findings, we propose an original conceptual model that synthesizes current progress and offers a roadmap for future explorations. By illuminating the affordances and constraints of AIFB, we highlight the necessity for transparent methodological reporting and underscores the importance of integrating pedagogical and technological insights to promote meaningful, learner-centered feedback.</p
Latent classes of self-reported feedback experiences:exploring students’ challenges, motivations, and action-taking behaviours in feedback processes
No full textAlthough students often acknowledge the importance of feedback, they generally struggle to engage with it and act upon it. Specific pedagogical factors, such as poorly structured feedback, unsuitable tone, and weak educator-student relationships, can impede effective utilisation of feedback. Students also exhibit varying degrees of comprehension, engagement, and action in response to feedback. Despite these observations, there is a lack of empirical studies systematically investigating diverse feedback experiences, practices, and action-taking behaviours of students. This paper addresses this gap by reporting on a study that aimed to explore students’ current feedback practices, self-reported action-taking behaviours, and perceived challenges related to students’ sensemaking and action-taking processes. A sample of 641 students from higher education was surveyed to investigate: (a) their feedback experiences, including practices, attitudes and beliefs; (b) variations in their motivations and emotional responses to feedback; and (c) variations in students’ perceived challenges in understanding and acting on feedback. The study employed 29 Likert scale items and latent class analysis (LCA) to identify four distinct classes of students based on their feedback experiences, aiming to uncover heterogeneity in their inclination to act upon feedback and challenges experienced in the feedback process. Additionally, thematic analysis of four open-ended questions captured a comprehensive understanding of their challenges, motivations, and emotional responses to feedback. The analysis revealed that students showed various levels of feedback experiences, engagement, and challenges in the feedback process across different classes. The paper concludes by highlighting the importance of self regulation skills and the social-affective component of a dialogic feedback process. This process can potentially be facilitated by technology-enhanced feedback tools, such as learning analytics (LA) tools.</p
Light-PTNet:A lightweight parallel temporal network for smartphone-based human motion classification
No full textThe increased popularity of smartphone-based human activity recognition (HAR) in recent decades has been driven by its low computational requirements and user privacy protection. Yet, developing a reliable smartphone-based HAR still presents several challenges. For example, handcrafted feature-based approaches highly depend on laborious feature engineering/selection techniques that require human intervention. Implementing conventional Convolutional Neural Networks may result in unsatisfactory performance in time series classification as they cannot effectively extract time-dependent features. Although recurrent models excel at extracting temporal information, they require extensive computational resources to attain high performance, limiting their practicality for real-time applications. Thus, we propose a lightweight smartphone-based HAR architecture called Lightweight Parallel Temporal Network (Light-PTNet) for reliable classification. Light-PTNet comprises parallelly organised Light Spatial-Temporal Heads (LSTC Heads) that capture underlying patterns at various scales of the inertial signals. These heads utilise dilations and residual connections to preserve longer-term dependencies without increasing the model parameters. This work assesses the proposed Light-PTNet’s performance on open-access HAR datasets: UCI HAR, WISDM V1, and UniMiB SHAR, following a user-independent protocol. The results reveal that our proposed Light-PTNet achieves 98.03% accuracy on UCI HAR, 81.58% on UniMiB SHAR and 97.02% on WISDM V1 with fewer model parameters (lower than 0.1 million parameters).</p
Codesigning rights-based recordkeeping for childhood out-of-home care
No full textDecades of inquires in Australia and internationally into child welfare and protection systems have highlighted the importance of quality recordkeeping to the lifelong identity, memory and accountability needs for those who experience Alternative Care. Having learnt about the devastating impacts of poor recordkeeping and challenges in accessing records of childhood Care experiences, it is clear that transformative approaches are needed to tackle what Australia’s Royal Commission into Institutional Responses to Childhood Sexual Abuse in December 2017 has labelled a ‘systemic and enduring’ problem. In this paper, we report on our involvement in a co-design team of recordkeeping researchers, Care leaver advocates, and digital technologists working together to develop a deep understanding of Care recordkeeping needs and realise them in the functionality of a digital prototype. Developed as a design provocation–a blueprint for participatory recordkeeping infrastructure–we will explain how it enables those currently cast as the client of services and ‘subject’ of records, to participate in their Care recordkeeping on equitable terms. We will discuss how having knowledge of, and appropriate say over, who has access to personal information and records can be re-imagined through digital technologies in support of lifelong rights to identity, privacy, autonomy and accountability.</p
Future schools and the energy implications of AI in education:A review of scenarios and method for engaging young people in futures thinking
Get PDF‘School of the future’ scenarios remain a popular means of animating policy, industry and public debates around issues relating to technological, economic, societal and environmental change. To date, these scenarios rarely involve the perspectives of school students. Purpose of the research: This study explores how scenarios can be used to engage school children in futures thinking, particularly regarding the uncertainties surrounding artificial intelligence and its environmental and energy implications. The research aims to develop participatory tools to help diversify future narratives about schools and foster young people’s ‘futures literacy’ and critical thinking about the future. Major findings: Analysis of 70 ‘future schools’ scenarios from 18 existing industry reports revealed limited approaches to climate change, energy and environmental implications of AI technology. These findings informed the design of scenario cards for engaging young people in imagining their own future schools, challenging dominant policy and industry expectations. Conclusions: The study contributes to knowledge in education and energy by combining scenario development from both sectors. It progresses all stakeholders towards desirable and resilient ‘AI energy futures’ by involving children and young people in the development of futures scenarios, addressing a gap in current scenario-building practices which have typically excluded student perspectives.</p
Turning Real-Time Analytics into Adaptive Scaffolds for Self-Regulated Learning Using Generative Artificial Intelligence
No full textIn computer-based learning environments (CBLEs), adopting effective self-regulated learning (SRL) strategies requires sophisticated coordination of multiple SRL processes. While various studies have proposed adaptive SRL scaffolds (i.e. real-time advice on adopting effective SRL processes) and embedded them in CBLEs to facilitate learners' effective use of SRL strategies, two key research gaps remain. First, there is a lack of research on SRL scaffolds that are based on continuous assessment of both learners' SRL processes and learning conditions (e.g., awareness of learning resources) to provide adaptive support. Second, current analytics-based scaffolding mechanisms lack the scalability needed to effectively address multiple learning conditions. Integration of analytics of SRL with generative artificial intelligence (GenAI) can provide scalable scaffolding for real-time SRL processes and evolving conditions. Yet, empirical studies implementing and evaluating effects of this integration remain scarce. To address these limitations, we conducted a randomized control trial, assigning participants to three groups (control, process only, and process with condition groups) to investigate the effects of using GenAI to turn insights from real-time analytics about students' SRL processes and conditions into adaptive scaffolds. The results demonstrate that integrating real-time analytics with GenAI in adaptive SRL scaffolds - addressing both SRL processes and dynamic conditions - promotes more metacognitive learning patterns compared to the control and process-only groups. In addition, the learners showed varying levels of compliance with analytics-based GenAI scaffolds, and this was also reflected in how the learners coordinated their SRL processes, particularly in the performance phase of SRL. This study contributes to the literature by designing, implementing, and evaluating the impact of adaptive scaffolds on learners' SRL processes using real-time analytics with GenAI.</p