Search Results (2,575)

Artificial intelligence technology holds significant importance for building intelligent question-answering systems in the field of coal mine safety and enhancing safety management levels. Currently, there is a lack of specialized large language models and high-quality question-answering datasets in this field. To address this, this study proposes a two-stage fine-tuning method based on Low-Rank Adaptation (LoRA) and Group Sequence Policy Optimization (GSPO) for training a question-answering model tailored to the coal mine safety domain. The research begins by constructing a dedicated question-answering dataset based on domain-specific regulatory documents. Subsequently, using Qwen2.5-7B Instruct as the base model, the study fine-tunes the model through supervised learning with LoRA technology, followed by further optimization of the model’s performance using the GSPO reinforcement learning algorithm. Experiments show that the model trained with this method exhibits significant improvements in coal mine safety-related tasks, achieving superior results on multiple automated evaluation metrics compared to contrast models of similar scale. This study validates the effectiveness of the two-stage fine-tuning method in adapting large language models (LLMs) to specific domains, providing a new technical approach for the intelligentization of coal mine safety. It should be noted that due to the lack of external data, this study relies on a self-constructed dataset and has not yet undergone external independent validation, which constitutes the main limitation of the current work. Full article

Socioscientific issues (SSIs) provide meaningful contexts for developing students’ competencies in scientific evaluation and decision-making. This study developed an SSI-based instructional unit to support primary school students in making decisions about genome-edited fish by considering risks and benefits and proposing risk mitigation. The study aimed to examine the unit’s effectiveness in improving students’ risk-related decision-making and their attitudes toward critical thinking and risk. Sixty-three fifth-grade students participated in an 18-lesson unit comprising two phases: information gathering and risk management practice. Students completed three decision-making tasks and a post-unit questionnaire on related attitudes. Written arguments were analysed using a rubric based on claims, risk knowledge, benefit knowledge, and risk mitigation. The results indicated that the unit improved the quality of students’ socioscientific arguments. By the final task, about 60% of arguments reached the highest level, demonstrating integration of risk knowledge and corresponding mitigation. However, students’ risk–benefit emphasis ratings showed that their decisions remained predominantly risk-focused, and questionnaire data revealed a persistent zero-risk mindset. These findings provide empirical evidence that an SSI-based unit incorporating risk management practice can foster primary students’ risk-related socioscientific decision-making. Further refinement is needed to shift students’ risk attitudes and support more balanced risk–benefit reasoning. Full article

This paper presents a scalable and low-cost Retrieval Augmented Generation (RAG) architecture designed to enhance learning in university-level courses, with a particular focus on supporting students from economically disadvantaged backgrounds. Recent advances in large language models (LLMs) have demonstrated considerable potential in educational contexts; however, their adoption is often limited by computational costs and the need for stable broadband access, issues that disproportionately affect low-income learners. To address this challenge, we propose a lightweight, mobile, and friendly RAG system that integrates the LLaMA language model with the Milvus vector database, enabling efficient on device retrieval and context-grounded generation using only modest hardware resources. The system was implemented in a university-level Data Mining course and evaluated over four semesters using a quasi-experimental design with randomized assignment to experimental and control groups. Students in the experimental group had voluntary access to the RAG assistant, while the control group followed the same instructional schedule without exposure to the tool. The results show statistically significant improvements in academic performance for the experimental group, with p < 0.01 in the first semester and p < 0.001 in the subsequent three semesters. Effect sizes, measured using Hedges g to account for small cohort sizes, increased from 0.56 (moderate) to 1.52 (extremely large), demonstrating a clear and growing pedagogical impact over time. Qualitative feedback further indicates increased learner autonomy, confidence, and engagement. These findings highlight the potential of mobile RAG architectures to deliver equitable, high-quality AI support to students regardless of socioeconomic status. The proposed solution offers a practical engineering pathway for institutions seeking inclusive, scalable, and resource-efficient approaches to AI-enhanced education. Full article

Tutoring has played a significant role in pandemic-related learning recovery, supporting student learning and engagement. This paper follows up on a recent randomized controlled trial (RCT) estimating that one-on-one virtual early literacy tutoring was nearly twice as effective as two-on-one tutoring for improving student learning. To better understand this gap, we analyze transcripts from 16,629 tutoring sessions from this RCT—which included over 3.7 million tutor utterances—using natural language processing and machine learning techniques. We explore how tutors allocate attention across content instruction, relationship building, and classroom management between one-on-one and two-on-one formats. While tutors dedicate similar time to content instruction and relationship building across both formats, students receiving one-on-one tutoring receive more attention and personalized support. To improve the effectiveness of two-on-one tutoring, it may be beneficial to equip tutors with strategies that engage multiple students simultaneously, thereby reducing downtime and minimizing the potential for disengagement. Full article

This study investigates the factors influencing Chinese K-12 teachers’ adoption of generative artificial intelligence (GenAI) for instructional purposes by extending the Technology Acceptance Model (TAM) with pedagogical beliefs, perceived intelligence, perceived ethical risks, GenAI anxiety, and demographic moderators. Drawing on a theory-driven framework, survey data were collected from 218 in-service teachers across K-12 schools in China. The respondents were predominantly from urban schools and most had prior GenAI use experience. Eight latent constructs and fourteen hypotheses were tested using structural equation modeling and multi-group analysis. Results show that perceived usefulness and perceived ease of use are the strongest predictors of teachers’ intention to adopt GenAI. Constructivist pedagogical beliefs positively predict both perceived usefulness and intention, whereas transmissive beliefs negatively predict intention. Perceived intelligence exerts strong positive effects on perceived usefulness and ease of use but has no direct effect on intention. Perceived ethical risks significantly heighten GenAI anxiety, yet neither directly reduces adoption intention. Gender, teaching stage, and educational background further moderate key relationships, revealing heterogeneous adoption mechanisms across teacher subgroups. The study extends TAM for the GenAI era and highlights the need for professional development and policy initiatives that simultaneously strengthen perceived usefulness and ease of use, engage with pedagogical beliefs, and address ethical and emotional concerns in context-sensitive ways. Full article

Textual revision is a recursive process integral to writing. However, less experienced writers often struggle to select effective strategies, underuse self-regulation, and evaluate their work without metacognitive control. This study examined the effectiveness of instructional programs focused on textual revision, incorporating dynamic assessment as a means to promote self-regulation. A total of 88 secondary school students (aged 13–15) participated, randomly assigned by class group to one of four conditions: mechanical revision, substantive revision, combined revision, or rewriting. A quasi-experimental design with repeated measures was used to assess the revisions carried out by the students Each intervention focused on distinct revision strategies: surface-level corrections, content and structure, or a combination of both. The rewriting group received no specific instruction beyond the weekly practice of rewriting the text that the other groups worked on. Findings revealed that students in the substantive revision group achieved the greatest gains in their revisions. The study concludes that instructional approaches focused on deep, content-oriented revision are particularly effective in improving students’ writing performance and fostering self-regulatory skills. These findings highlight the value of embedding metacognitive support in revision-focused instruction. Full article

Supervised fine-tuning is key for model alignment, but its mechanisms are debated, with conflicting evidence supporting either a superficial alignment hypothesis or significant task improvements. This paper examines supervised fine-tuning’s impact from the perspective of verbatim memorization. Using the open-source OLMo-2 model series and test datasets (instruction format, safety-sensitive, and factual knowledge) constructed from its pre-training corpus, we analyzed changes across memorization, linguistic styles, and task performance. We found that supervised fine-tuning significantly weakens the model’s verbatim memorization of pre-training data. Simultaneously, it improves generated text in terms of alignment objectives, such as polite expression and structured organization. However, this process also leads to performance degradation on knowledge-intensive downstream tasks. Further representation analysis reveals that these changes are mainly concentrated in the later layers of the model. We conclude that supervised fine-tuning acts as a continuation of the learning process on new data. By adjusting model representations, supervised fine-tuning induces a learning tilt toward the styles and content of the instruction-tuning dataset. This inclination successfully instills alignment objectives while consequently reducing the effective accessibility of previously learned knowledge, which indicates the observed degradation in both pre-training data memorization and factual task performance. The source code is publicly available. Full article

Background: Pelvic floor dysfunction, more prevalent in women but affecting both genders, impairs sphincter control and sexual health, and causes pelvic pain. Pelvic floor muscle (PFM) training is the first-line treatment for urinary incontinence, supported by robust evidence. Rehabilitative ultrasound imaging (RUSI) serves as a visual biofeedback tool, providing real-time imaging to enhance PFM training, motor learning, and treatment adherence. Aim: This narrative review evaluates the role and efficacy of RUSI in pelvic floor rehabilitation. Method: A comprehensive search of PubMed, Cochrane, and MEDLINE was conducted using keywords related to pelvic floor rehabilitation, ultrasound, and biofeedback, limited to English-language publications up to July 2025. Systematic reviews, meta-analyses, and clinical trials were prioritized. Results: Transperineal and transabdominal ultrasound improve PFM function across diverse populations. In post-prostatectomy men, transperineal ultrasound-guided training enhanced PFM contraction and reduced urinary leakage. In postpartum women with pelvic girdle pain, transabdominal ultrasound-guided biofeedback combined with exercises decreased pain and improved function. Ultrasound-guided pelvic floor muscle contraction demonstrated superior performance compared to verbal instruction. Notably, 57% of participants who were unable to contract the pelvic floor muscles with verbal cues achieved a correct contraction with ultrasound biofeedback, and this approach also resulted in more sustained improvements in PFM strength. Compared to other biofeedback modalities, RUSI demonstrated outcomes that are comparable to or superior to those of alternative methods. However, evidence is limited by a lack of standardized protocols and randomized controlled trials comparing RUSI with other modalities. Conclusions: RUSI is an effective visual biofeedback tool that enhances outcomes of PFM training in pelvic floor rehabilitation. It supports clinical decision-making and patient engagement, particularly in cases where traditional assessments are challenging. Further research, including the development of standardized protocols and comparative trials, is necessary to optimize the clinical integration of this method and confirm its superiority over other biofeedback methods. Full article

This study developed and evaluated an adaptive e-learning environment for selected number theory topics using a mixed-methods research design, conducted over an eleven-month period across secondary and early tertiary education contexts. The evaluation focused on three primary outcome domains: (1) learning-related outcomes (problem-solving accuracy and task success rate), (2) learner engagement and activity indicators (daily logins and tasks completed per day), and (3) system usability, assessed according to Jakob Nielsen’s usability dimensions. Quantitative data were collected through student and teacher questionnaires (N = 264 students; N = 52 teachers) and large-scale logfile analytics comprising more than 825,000 recorded system interactions. Qualitative feedback from students and teachers complemented the quantitative analyses. The results indicate statistically significant increases in learner activity, task completion rates, and problem-solving success following the introduction of the adaptive system, as demonstrated by inferential statistical analyses with confidence intervals. Post-use evaluations further indicated high levels of learner motivation and self-confidence, along with positive perceptions of system usability. Teachers evaluated the system positively in terms of learnability, efficiency, and instructional integration. Logfile analyses also revealed sustained growth in daily engagement and task success over time. Overall, the findings suggest that adaptive e-learning environments can effectively support engagement, usability, and learning-related performance in number theory education, although further research is required to examine the sustainability of learning-related outcomes over extended periods and to further refine error-handling mechanisms. Full article

This paper investigates the knowledge, attitudes, and practical adherence to the Mediterranean Diet (MD) among students of the Teacher Education Study Program in Split. Recent trends indicate a decline in adherence within Mediterranean regions, a phenomenon known as the Mediterranean paradox. Studying the relationship between students’ knowledge and practice is critical within the context of SDG 3 and SDG 4, as it highlights the role of future educators in promoting healthy communities. A quantitative approach was employed using the MDNK methodology, supplemented with the MEDAS test, to assess adherence to the Mediterranean Diet. Statistical analysis included p-values and effect size measures to assess practical relevance. Students averaged 13.39/18 on the MDNK test, with most showing moderate MEDAS adherence and only 5 reaching high adherence. The year of study or employment status has not been shown as an influential factor. While most students possess a high level of knowledge on the MD’s key components and express a positive attitude toward this dietary pattern, a significant knowledge-practice gap exists, confirming the Mediterranean paradox among future teachers. The need for nutritional and food education within the university curriculum is essential to move beyond theoretical instruction and actively promote food literacy and practical skills. Full article

The broader aim of this research is to examine how Learning Analytics (LA) can become ethically sound, pedagogically actionable, and realistically adopted in educational practice. To address this overarching challenge, the study investigates three interrelated research questions: ethics by design, learning impact, and adoption conditions. Methodologically, the research follows an exploratory sequential multi-method design. First, a meta-synthesis of 53 studies is conducted to identify key ethical challenges in LA and to derive an ethics-by-design framework. Second, a quasi-experimental study examines the impact of interface-based LA guidance (strong versus minimal) on students’ self-regulated learning skills and academic performance. Third, a mixed-methods adoption study, combining surveys, focus groups, and ethnographic observations, investigates the factors that encourage or hinder teachers’ adoption of LA in K–12 education. The findings indicate that strong LA-based guidance leads to statistically significant improvements in students’ self-regulated learning skills and academic performance compared to minimal guidance. Furthermore, the adoption analysis reveals that performance expectancy, social influence, human-centred design, and positive emotions facilitate LA adoption, whereas effort expectancy, limited facilitating conditions, ethical concerns, and cultural resistance inhibit it. Overall, the study demonstrates that ethics by design, effective pedagogical guidance, and adoption conditions are mutually reinforcing dimensions. It argues that LA can support intelligent, responsive, and human-centred learning environments when ethical safeguards, instructional design, and stakeholder involvement are systematically aligned. Full article

Artificial Intelligence (AI) systems typically inherit biases from their training data, leading to discriminatory outcomes that undermine equity and inclusion. This issue is particularly significant when popular Generative AI (GAI) applications are used in educational contexts. To respond to this challenge, the study evaluates the effectiveness of dialogic reflection-based training for educators in identifying and mitigating biases in AI. Furthermore, it considers how these sessions contribute to the advancement of algorithmic justice and inclusive practices. A key component of the proposed training methodology involved equipping educators with the skills to design inclusive prompts—specific instructions or queries aimed at minimizing bias in AI outputs. This approach not only raised awareness of algorithmic inequities but also provided practical strategies for educators to actively contribute to fairer AI systems. A qualitative analysis of the course’s Moodle forum interactions was conducted with 102 university professors and graduate students from diverse regions of the Dominican Republic. Participants engaged in interactive activities, debates, and practical exercises addressing AI bias, algorithmic justice, and ethical implications. Responses were analyzed using Atlas.ti across five categories: participation quality, bias identification strategies, ethical responsibility, social impact, and equity proposals. The training methodology emphasized collaborative learning through real case analyses and the co-construction of knowledge. The study contributes a hypothesis-driven model linking dialogic reflection, bias awareness, and inclusive teaching, offering a replicable framework for ethical AI integration in higher education. Full article

Transitioning from theoretical learning to practical application remains a significant challenge for students in medical and veterinary science education, particularly in the context of medical imaging and ultrasound interpretation. Traditional lecture-based methods offer limited support for developing the spatial reasoning and technical skills required for ultrasound pregnancy diagnosis. This study evaluates the effectiveness of an interactive mixed reality (MR) training tool, Ewe Scan, delivered through the Apple Vision Pro, compared to traditional lecture-based instruction. Forty-two undergraduate students were randomly assigned to either a lecture-trained or MR-trained group and assessed immediately after training and again after six weeks. Results showed that MR-trained students significantly outperformed their lecture-trained peers in both immediate comprehension and retention over time, particularly in ultrasound interpretation skills. The MR-trained group also reported higher levels of engagement, confidence, and satisfaction with their training experience. These findings suggest that MR-based learning enhances educational outcomes by improving spatial understanding, increasing active engagement, and supporting knowledge retention. Integrating MR simulations into ultrasound education offers a scalable, ethical, and effective alternative to traditional training methods, contributing to advancements in medical imagery education. Full article

Growing interest in food sustainability education aims to increase awareness of food distribution systems, environmental degradation, and the connectivity of sustainable and ethical food practices. However, recent scholarship has questioned whether such pedagogical efforts are meaningfully internalized by students or lead to sustained behavioral change. Prior studies document persistent gaps in students’ understanding of sustainability impacts and the limited effectiveness of existing instructional approaches in promoting transformative engagement. To address these concerns, the Food Ethics, Sustainability and Alternatives (FESA) course was implemented with 21 undergraduate and graduate students at Montclair State University (Montclair, NJ, USA). Course outcomes were evaluated using a mixed-methods design integrating qualitative analysis with quantitative measures informed by the Theory of Planned Behavior, to identify influences on students’ attitudes, and a Transtheoretical Model (TTM) panel survey to address progression from awareness to action, administered pre- and post-semester. Qualitative findings revealed five central themes: increased self-awareness of food system contexts, heightened attention to animal ethics, the importance of structured classroom dialogue, greater recognition of food waste, and increased openness to alternative food sources. TTM results indicated significant reductions in contemplation and preparation stages, suggesting greater readiness for change, though no significant gains were observed in action or maintenance scores. Overall, the findings suggest that while food sustainability education can positively shape student attitudes, the conversion of attitudinal shifts into sustained behavioral change remains limited by external constraints, including time pressures, economic factors, culturally embedded dietary practices, structural tensions within contemporary food systems, and perceptions of limited individual efficacy. Full article

Background/Objectives: Metabolism is fundamental to all living organisms. It comprises a highly complex network of fine-tuned chemical reactions that sustain life but also generate by-products that damage cellular biomolecules, including DNA, thereby contributing to aging and disease. As metabolism can be largely modified by dietary alterations, it has the potential to positively or negatively affect health and disease. Interestingly, many aging-associated illnesses known to be influenced by diet also show a causal relation with DNA damage. As DNA keeps all instructions for life, and DNA lesions, if unrepaired, interfere with vital processes such as DNA replication and transcription, DNA damage may be an important mediator of the impact of nutrition on health and aging. Methods: Here, we discuss the genome-protective effects of various oral interventions in mice, aiming to elucidate which nutritional alterations lower DNA damage and promote overall health. Results: Our analysis covers a wide range of interventions with reported positive impacts on genomic stability, including modified diets (e.g., dietary restriction, probiotics, micronutrients, fatty acids, and hormones), NAD+ precursors (e.g., nicotinamide riboside), plant derivatives, and synthetic drugs. Among these, caloric and dietary restriction emerge as the most potent, generic modulators of DNA damage and repair processes, enhancing aspects of repair efficiency through metabolic recalibration and improved cellular resilience. Other interventions, like NAD+ precursors, activate partly similar pathways without necessitating reduced food intake. Conclusions: While many interventions show promise, their effects are often less pronounced or are process-specific compared to caloric or dietary restriction. Additionally, many substances lack comprehensive exploration of their genome-protective effects in mice, with often only a small number of studies examining their impact on genome stability. Moreover, the heterogeneity between studies limits direct comparison. However, the observed overlap in mechanistic effects between treatments lends credibility to their potential efficacy. Ultimately, a deeper understanding of these mechanisms could pave the way for translating these findings into, e.g., combination treatments to promote healthy aging in humans. Full article