Search Results (782)

Biochar serves as an effective adsorbent for the heavy metal cadmium, with its performance significantly influenced by its physicochemical properties and various environmental features. Traditional machine learning models, though adept at managing complex multi-feature relationships, rely heavily on expertise in feature engineering and hyperparameter optimization. To address these issues, this study employs an automated machine learning (AutoML) approach, automating feature selection and model optimization, coupled with an intuitive online graphical user interface, enhancing accessibility and generalizability. Comparative analysis of four AutoML frameworks (TPOT, FLAML, AutoGluon, H₂O AutoML) demonstrated that H₂O AutoML achieved the highest prediction accuracy (R² = 0.918). Key features influencing adsorption performance were identified as initial cadmium concentration (23%), stirring rate (14.7%), and the biochar H/C ratio (9.7%). Additionally, the maximum adsorption capacity of the biochar was determined to be 105 mg/g. Optimal production conditions for biochar were determined to be a pyrolysis temperature of 570–800 °C, a residence time of ≥2 h, and a heating rate of 3–10 °C/min to achieve an H/C ratio of <0.2. An online graphical user interface was developed to facilitate user interaction with the model. This study not only provides practical guidelines for optimizing biochar but also introduces a novel approach to modeling using AutoML. Full article

While prior work has explored learner behavior using learning management systems (LMS) data, few studies provide week-level clustering validated against external engagement labels. To understand and assist students in online learning platforms and environments, this study presents a week-level engagement profiling framework for e-learning environments, utilizing K-means clustering and label-based validation. Leveraging log data from 127 students over a 13-week course, 44 activity-based features were engineered to classify student engagement into high, moderate, and low levels. The optimal number of clusters (k = 3) was identified using the elbow method and assessed through internal metrics, including a silhouette score of 0.493 and R² of 0.80. External validation confirmed strong alignment with pre-labeled engagement levels based on activity frequency and weighting. The clustering approach successfully revealed distinct behavioral patterns across engagement tiers, enabling a nuanced understanding of student interaction dynamics over time. Regression analysis further demonstrated a significant association between engagement levels and academic performance, underscoring the model’s potential as an early warning system for identifying at-risk learners. These findings suggest that clustering based on LMS behavior offers a scalable, data-driven strategy for improving learner support, personalizing instruction, and enhancing retention and academic outcomes in digital education settings such as MOOCs. Full article

In an era where online interactions increasingly shape social dynamics, the pervasive issue of cyberbullying poses a significant threat to the well-being of individuals, particularly among vulnerable groups. Despite extensive research on text-based cyberbullying detection, the rise of visual content on social media platforms necessitates new approaches to address cyberbullying using images. This domain has been largely overlooked. In this paper, we present a novel dataset specifically designed for the detection of visual cyberbullying, encompassing four distinct classes: abuse, curse, discourage, and threat. The initial prepared dataset (cyberbullying visual indicators dataset (CVID)) comprised 664 samples for training and validation, expanded through data augmentation techniques to ensure balanced and accurate results across all classes. We analyzed this dataset using several advanced deep learning models, including VGG16, VGG19, MobileNetV2, and Vision Transformer. The proposed model, based on DenseNet201, achieved the highest test accuracy of 99%, demonstrating its efficacy in identifying the visual cues associated with cyberbullying. To prove the proposed model’s generalizability, the 5-fold stratified K-fold was also considered, and the model achieved an average test accuracy of 99%. This work introduces a dataset and highlights the potential of leveraging deep learning models to address the multifaceted challenges of detecting cyberbullying in visual content. Full article

This study investigates the interplay between adaptive online learning, students’ digital intelligence, sustainable learning, and digital mental well-being among female university students in Saudi Arabia. In response to the growing reliance on digital platforms in higher education, a structured questionnaire was distributed via social media to capture student perceptions of their online learning experiences. Using Partial Least Squares Structural Equation Modelling (PLS-SEM), the analysis revealed that while adaptive online learning is a critical enabler, its influence is most effective when mediated by students’ digital intelligence. The findings highlighted that students with higher digital intelligence are more likely to engage in sustainable learning practices and maintain better mental well-being in digital environments. Furthermore, innovative teaching practices were shown to strengthen these relationships, underscoring the importance of interactive and adaptive pedagogies. This research contributes to the growing discourse on digital education by emphasizing the importance of indirect pathways and learner-centred dynamics in shaping positive educational and psychological outcomes. This study offers practical and theoretical implications for educators, institutions, and policymakers aiming to create inclusive, resilient, and psychologically supportive digital learning environments. Future research is encouraged to examine these relationships across different cultural and institutional contexts and explore the longitudinal impacts of digital learning strategies. Full article

This study introduces an educational framework that merges 3D modeling with AI-assisted narrative interaction to apply digital technology in cultural heritage education, exemplified by an ancient carriage culture. Through immersive tasks and contextual narratives, the course notably improved learners’ professional skills and cultural awareness. Experimental results revealed significant knowledge acquisition among participants post-engagement. Additionally, the user experience improved, with increased satisfaction in the narrative interaction design course. These enhancements led to heightened interest in cultural heritage and deeper knowledge acquisition. Utilizing Norman’s three-layer interaction model, Ryan’s contextual narrative theory, and Falk and Dierking’s museum learning experience model, the study developed a systematic course for multi-sensory design and contextual interaction, confirming the positive impact of multimodal interaction on learning outcomes. This research provides theoretical support for the digital transformation of cultural education and practical examples for educational practitioners and cultural institutions to implement in virtual presentations and online learning. Full article

Every day humans interact with smartphones that have embedded sensors that enable the tracking of changing physical activities of the device owner. However, several problems arise with the recognition of multiple activities (such as walking, sitting, running, and other) on smartphones. Firstly, most of the devices do not recognize some activities well, such as walking upstairs or downstairs. Secondly, recognition algorithms are embedded into smartphone software and are static, unless updated. In this case, a recognition algorithm must be re-trained with training data of a specific size. Thus, an adaptive (also known as, online or incremental) learning algorithm would be useful in this situation. In this work, an adaptive learning and classification algorithm based on hidden Markov models (HMMs) is applied to human activity recognition, and an architecture model for smartphones is proposed. To create a self-learning method, a technique that involves building an incremental algorithm in a maximal likelihood framework has been developed. The adaptive algorithms created enable fast self-learning of the model parameters without requiring the device to store data obtained from sensors. It also does not require sending gathered data to a server over the network for additional processing, making them autonomous and independent from outside systems. Experiments involving the modeling of various activities as separate HMMs with different numbers of states, as well as modeling several activities connected to one HMM, were performed. A public dataset called the Activity Recognition Dataset was considered for this study. To generalize the results, different performance metrics were used in the validation of the proposed algorithm. Full article

Cantonese Automatic Speech Recognition (ASR) is hindered by tonal complexity, acoustic diversity, and a lack of labelled data. This study proposes a phoneme-aware hierarchical augmentation framework that enhances performance without additional annotation. A Phoneme Substitution Matrix (PSM), built from Montreal Forced Aligner alignments and Tacotron-2 synthesis, injects adversarial phoneme variants into both transcripts and their aligned audio segments, enlarging pronunciation diversity. Concurrently, a semantic-aware SpecAugment scheme exploits wav2vec 2.0 attention heat maps and keyword boundaries to adaptively mask informative time–frequency regions; a reinforcement-learning controller tunes the masking schedule online, forcing the model to rely on a wider context. On the Common Voice Cantonese 50 h subset, the combined strategy reduces the character error rate (CER) from 26.17% to 16.88% with wav2vec 2.0 and from 38.83% to 23.55% with Zipformer. At 100 h, the CER further drops to 4.27% and 2.32%, yielding relative gains of 32–44%. Ablation studies confirm that phoneme-level and masking components provide complementary benefits. The framework offers a practical, model-independent path toward accurate ASR for Cantonese and other low-resource tonal languages. This paper presents an intelligent sensing-oriented modeling framework for speech signals, which is suitable for deployment on edge or embedded systems to process input from audio sensors (e.g., microphones) and shows promising potential for voice-interactive terminal applications. Full article

This study explores the integration of digital media tools—specifically TikTok, online press news analysis, and digital portfolios—into pre-service chemistry teacher education to enhance student engagement, foster conceptual understanding, and highlight the relevance of chemistry in society. The educational intervention involved 138 pre-service teachers who analysed digital news articles to reflect on the societal and environmental implications of chemistry, promoting media literacy and awareness of socioscientific issues. Additionally, they created short-form TikTok videos, using social media to communicate scientific concepts creatively and interactively. All participants compiled their work into digital portfolios, which served as both a reflective and integrative tool. A post-course Likert-scale questionnaire (N = 77) revealed high overall satisfaction with the methodology, with 94.8% valuing the news analysis activity and 59.7% finding TikTok particularly engaging. Despite some limitations regarding access to technical infrastructure, the findings indicate that incorporating Information and Communication Technology (ICT) in this manner supports motivation, meaningful learning, and the development of key teaching competencies. This case study contributes practical insights into ICT use in science education. Full article

Advances in information technology have enabled virtual museums to transcend traditional physical boundaries and become important tools in education. Despite their growing use, the factors influencing the effectiveness of virtual museums in enhancing students’ learning motivation remain underexplored. This study investigates key factors that promote learning motivation among secondary school students using the National Museum of Nature’s Online Virtual Exhibition as a case study. Grounded in the Stimulus–Organism–Response (S-O-R) theoretical framework, a conceptual model was developed and empirically tested using Structural Equation Modeling (SEM) to examine relationships among stimulus variables, psychological states, and learning motivation. Results reveal that affective involvement, cognitive engagement, and perceived presence significantly enhance learning motivation, while immersion shows no significant effect. Among the stimulus factors, perceived enjoyment strongly promotes affective involvement, perceived interactivity enhances cognitive engagement, and content quality primarily supports cognitive processing. Visual aesthetics contribute notably to immersion, affective involvement, and perceived presence. These findings elucidate the multidimensional mechanisms through which user experience in virtual museums influences learning motivation. The study provides theoretical and practical implications for designing effective and engaging virtual museum educational environments, thereby supporting sustainable digital learning practices. Full article

This study aims to develop and validate a user experience scale to construct an Actual System Use model for online courses based on the Technology Acceptance Model, allowing for a comprehensive assessment of the multidimensional factors affecting Learning Outcomes and Actual System Use in the context of online courses. The scale includes six core dimensions: Interactive Experience, Content Quality, Learning Outcomes, Teaching Quality, Technical Support, and Learning Motivation. Through a literature review, pre-survey, exploratory factor analysis, and confirmatory factor analysis, the reliability and validity of the developed scale were verified. A second-order complex Structural Equation Model was used to measure users’ Actual System Use with respect to online courses. The results demonstrate that the Interactive Experience and Learning Motivation dimensions play crucial roles in enhancing learners’ engagement and learning satisfaction, while Perceived Usefulness and Perceived Ease of Use significantly influence system usage behaviors. This study provides a systematic theoretical basis and empirical data for the design of online courses, offering valuable insights for optimizing course design and enhancing user experiences. Full article

This study aims to explore the interrelationships among key factors influencing learning burnout, such as motivation and negative emotions (depression, anxiety, and stress) along with other factors influencing including problematic mobile phone use, nomophobia, and interactive learning, as well as whether their pathways of influence on learning burnout differ between online and offline learning contexts. Using the convenience sampling method, data from 293 college students were collected. Measurements were carried out using the Nomophobia Scale, the Problematic Mobile Phone Use Scale, the Depression Anxiety Stress Scale (DASS), the Interactive Learning Scale, the Learning Burnout Scale, and the Scale of Motivation for Activity Participation. By applying network analysis and network comparison methods, and based on the Social Comparison Theory and the Affective Socialization Heuristics Model, it was found that under the online learning condition the motivation to pursue value directly affects learning burnout. In contrast, under the offline learning condition learning motivation indirectly affects learning burnout through negative emotions. This study posits that this difference is caused by peer comparison. In a collective learning atmosphere, students’ comparison with their peers triggers negative emotions such as anxiety and stress. These negative emotions weaken the learning motivation to pursue value, ultimately resulting in an elevated level of learning burnout. Full article

With the development of robotics, robots are playing an increasingly critical role in complex tasks such as flexible manufacturing, physical human–robot interaction, and intelligent assembly. These tasks place higher demands on the force control performance of robots, particularly in scenarios where the environment is unknown, making constant force control challenging. This study first analyzes the robot and its interaction model with the environment, highlighting the limitations of traditional force control methods in addressing unknown environmental stiffness. Based on this analysis, a variable admittance control strategy is proposed using the deep deterministic policy gradient algorithm, enabling the online tuning of admittance parameters through reinforcement learning. Furthermore, this strategy is integrated with a quaternion-based nonlinear model predictive control scheme, ensuring coordination between pose tracking and constant-force control and enhancing overall control performances. The experimental results demonstrate that the proposed method improves constant force control accuracy and task execution stability, validating the feasibility of the proposed approach. Full article

This study aims to explore the mechanisms through which social media influences the cooperative learning attitudes and academic performance of design students in the context of China’s collectivist culture, providing a basis for the application of social media in design education. Using the Extended Technology Acceptance Model (TAM) as the theoretical framework, a questionnaire survey of 305 students was conducted. Structural equation modelling and moderation effect analysis revealed that perceived usefulness, ease of use, enjoyment, and interactivity significantly influence students’ attitudes toward social media-based collaborative learning. This attitude directly enhances academic performance and is positively moderated by knowledge-sharing willingness and academic self-efficacy. This study validated the applicability of the extended TAM in online collaborative learning, revealing that positive attitudes toward collaborative learning can only effectively translate into academic outcomes when students possess sufficient knowledge-sharing willingness or self-efficacy. This provides empirical evidence for strategically leveraging social media in educational design. Full article

The detection of abusive language in Roman Urdu is important for secure digital interaction. This work investigates machine learning (ML), deep learning (DL), and transformer-based methods for detecting offensive language in Roman Urdu comments collected from YouTube news channels. Extracted features use TF-IDF and Count Vectorizer for unigrams, bigrams, and trigrams. Of all the ML models—Random Forest (RF), Logistic Regression (LR), Support Vector Machine (SVM), and Naïve Bayes (NB)—the best performance was achieved by the same SVM. DL models involved evaluating Bi-LSTM and CNN models, where the CNN model outperformed the others. Moreover, transformer variants such as LLaMA 2 and ModernBERT (MBERT) were instantiated and fine-tuned with LoRA (Low-Rank Adaptation) for better efficiency. LoRA has been tuned for large language models (LLMs), a family of advanced machine learning frameworks, based on the principle of making the process efficient with extremely low computational cost with better enhancement. According to the experimental results, LLaMA 2 with LoRA attained the highest F1-score of 96.58%, greatly exceeding the performance of other approaches. To elaborate, LoRA-optimized transformers perform well in capturing detailed subtleties of linguistic nuances, lending themselves well to Roman Urdu offensive language detection. The study compares the performance of conventional and contemporary NLP methods, highlighting the relevance of effective fine-tuning methods. Our findings pave the way for scalable and accurate automated moderation systems for online platforms supporting multiple languages. Full article

High dropout rates remain a significant challenge in Massive Open Online Courses (MOOCs), making early identification of at-risk students crucial. This study introduces a novel approach called In-Session Stacked Ensemble Learning for Dropout Prediction (ISELDP), which predicts student dropout during course sessions by combining multiple base learners—Adaptive Boosting (AdaBoost), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Gradient Boosting—into a stacked ensemble with a Multi-Layer Perceptron (MLP) serving as the meta-learner. To optimise model performance, hyperparameters were tuned using Grid Search. The proposed method was evaluated under two scenarios using in-session student interaction data, one with imbalanced data and another with balanced data. Results demonstrate that ISELDP achieves an average accuracy of 88%, outperforming individual baseline models with improvements of up to 2% in accuracy and 2.4% in F1-score. Full article