Cognitive Load Theory (CLT) is an instructional design theory that aligns with human cognitive architecture for creating instructional materials, through the design guidelines of its 17 instructional effects. However, the Self-Management effect suggests that students can be instructed to manage their learning. The Collective Working Memory effect highlights how a group of students working together can foster a more effective learning environment than an individual student, resulting in better learning outcomes. This research explored applying the Self-Management effect of CLT alongside the Collective Working Memory effect learning data structures in basic programming and measuring their effectiveness regarding essential knowledge acquisition in declarative knowledge, knowledge transfer (near transfer) in procedural knowledge, and developing self-management skills. Cognitive load was measured to determine the difference between groups and to determine the correlation with learning outcomes. The study was carried out through a quasi-experimental design with homogeneous groups, involving students from the Autonomous University of Aguascalientes. The results suggest positive findings in knowledge transfer as well as the development of self-management skills. The cognitive load between the participating groups does not show any significant statistical difference, nor does it show any correlation with the learning results.

This research explores the connection between interventions for mental well-being and Virtual Reality (VR). A bibliometric analysis was conducted to assess the state of the field, comparing mental health applications to understand VR's significance. The study revealed a limited but growing body of literature examining VR's effects on mental health, primarily targeting mood, stress, and anxiety disorders. Various approaches, including immersive VR experiences, were identified, offering unique therapeutic benefits. The comparative investigation across disorders underscored the potential of VR therapy to create synergistic effects when combined with other interventions. Immersive VR experiences were found to offer innovative ways to address emotional regulation and stress management, enhancing traditional therapeutic methods. The applications and techniques developed so far provide significant insights into the transformative role VR could play in mental health care. The findings emphasize the importance of further research to optimize and expand VR interventions for mental well-being. Such advancements could pave the way for more personalized, engaging, and effective mental health solutions, particularly for conditions resistant to conventional therapies. By leveraging VR’s immersive and interactive capabilities, mental health practitioners can create interventions that not only alleviate symptoms but also foster long-term psychological resilience. This study highlights the critical need to develop and implement VR-based interventions systematically, ensuring their accessibility and efficacy across diverse populations. By doing so, VR can serve as a cornerstone in the evolution of mental health care, bridging gaps and unlocking new possibilities for well-being.

At Tay Nguyen University in Vietnam, the English for Medical Purposes (EMP) course is tailored to address the specific language learning needs of medical students. Despite its aims, the course exhibits various limitations that hinder both teaching and learning processes. This cross-sectional descriptive study was conducted among medical students in the Faculty of Medicine and Pharmacy, encompassing cohorts from 2017 to 2022, to assess the interest levels and influencing factors related to learning EMP. A total of 258 students participated in the survey, revealing that only 31% expressed a preference for learning EMP. The study identified several key factors impacting the learning of EMP, including future career prospects, the learning environment, study materials, students' self-study habits, and specific teaching activities for medical purposes. This research provides valuable insights for both students and educational administrators at Tay Nguyen University, facilitating the development of effective strategies to enhance the efficiency and quality of EMP education.

The growing complexity of academic and career decision-making requires intelligent systems that can deliver personalized recommendations while ensuring strict compliance with institutional policies and incorporating evolving contextual factors. This paper introduces NeSy-Guidance, a neuro-symbolic recommendation approach that combines symbolic rule reasoning with graph-based neural inference over a dynamic and time-aware academic Knowledge Graph (KG). The model encodes students, programs, and contextual entities including regions and emerging trends while integrating regulatory constraints derived from Moroccan admission policies. It applies a two-stage reasoning pipeline: a symbolic layer enforcing hard eligibility rules and soft preference-based adjustments mined automatically from the knowledge graph, and a Graph Convolutional Network (GCN) layer trained with a weighted loss to address class imbalance and capture latent student–program compatibility. A weighted score fusion mechanism integrates both inference outputs, achieving a balance between interpretability, adaptability, and predictive performance. Evaluated on a real-world dataset of 800 students and 325 academic programs, NeSy-Guidance outperforms three state-of-the-art baselines in both accuracy and policy compliance. It achieves 83.8% accuracy, 74.5% precision@5, 75.4% F1-score, and ensures 100% compliance with institutional eligibility rules. Furthermore, a qualitative survey confirmed positive student satisfaction regarding the clarity and relevance of recommendations. These results demonstrate the effectiveness of hybrid reasoning and validate NeSy-Guidance as a reliable, explainable, and regulation-aware academic guidance system capable of adapting to regional disparities, emerging academic trends, and evolving student preferences.

The focus of the research is on the analysis of the effectiveness of different forms of educational activities in developing youth’s information and media literacy (IML), based on the results of the Ukrainian project “MEDIA & CAPSULES”, implemented within IREX’s “Learn and Discern” initiative. The study compared the impact of webinar sessions, masterclasses, and information and media workshops on three key IML indicators: information literacy, media literacy, and digital security. An empirical pre-post design was used to assess changes in participants’ competencies before and after each type of educational intervention. Statistical analysis revealed that information and media workshops had the strongest overall impact, particularly enhancing media literacy and digital security. Masterclasses were most effective in improving information literacy, while webinars showed moderate improvements across all indicators. The findings highlight the importance of aligning instructional formats with specific educational goals and provide practical implications for educators and curriculum developers working to strengthen youth resilience against misinformation and digital threats.

Traditional educational institutions prioritize theoretical education over hands-on practical skills which produce a gap between classroom learning and industry requirements particularly in the fast-growing blockchain sector. A structured case study demonstrates how Project-Based Learning (PBL) was implemented in an undergraduate engineering course which focused on blockchain technology. The educational approach evolved through four stages that combined theoretical instruction with collaborative solution creation and DApp programming and assessment evaluation. Performance metrics from students including testing coverage, GitHub contributions, documentation quality and research paper output are carefully analyzed through algorithmic guidance of each phase. The paper demonstrates the development of teaching methods through traditional practices and outcome-based instruction up to project-based learning supported by visual timeline comparisons. Student feedback demonstrates that the education methods led to enhanced technical abilities together with teamworking and increased student confidence. The case study demonstrates how PBL functions as an educational connection between academic learning and practical blockchain development needs because most teams (over 75%) finished functional DApps alongside several groups producing research suitable for publication.

Massive Open Online Courses (MOOCs) and micro-credentials have emerged as key innovations in modern education, offering scalable, flexible access to learning and skill development. Despite their potential, challenges such as low learner engagement, high dropout rates, and uncertainty over the value of digital credentials remain. This study analyzes 3,743 publications from 1970 to 2024 using bibliometric and text analytics to uncover research trends, influential studies, and dominant themes in the field. Results show a surge in research from 2014 to 2020 driven by digital technology adoption and the COVID-19 pandemic followed by a decline as hybrid learning models became normalized. Key themes include learner motivation, engagement strategies, digital badges, and ethical concerns tied to data-driven education. While advancements in learning analytics and personalization show promise, the study underscores the need for standardized credentialing, scalable engagement frameworks, and ethical governance in online education. Critical gaps remain, particularly in evaluating the long-term impact of micro-credentials on employability and understanding adoption differences across regions and socio-economic groups. Limitations include reliance on the Web of Science and author-provided keywords, which may narrow the scope. Despite this, the study provides a systematic overview and offers practical insights for improving MOOCs and micro-credentials as tools for lifelong learning and global educational equity.

Sarcasm, a subtle form of expression, is challenging to detect, especially in modern communication platforms where communication transcends text to encompass videos, images, and audio. Traditional sarcasm detection methods rely solely on textual data and often struggle to capture the nuanced emotional inconsistencies inherent in sarcastic remarks. To overcome these shortcomings, this paper introduces a novel multimodal framework incorporating text, audio, and emoji data for more effective sarcasm detection and emotion classification. A key component of this framework is the Contextualized Semantic Self-Guided BERT (CS-SGBERT) model, which generates efficient word embeddings. Primarily, frequency spectral analysis is performed on the audio data, followed by preprocessing and feature extraction, while text data undergoes preprocessing to extract lexicon and irony features. Meanwhile, emojis are analyzed for polarity scores, which provide a rich set of multimodal features. The fused features are then optimized using the Camberra-based Dingo Optimization Algorithm (C-DOA). The selected features and the embedded words from the preprocessed texts are given to Entropy-based Robust Scaling - Gated Recurrent Units (E-RS-GRU) for detecting sarcasm. Experimental results on the MUStARD dataset show that the proposed E-RS-GRU model achieves an accuracy of 76.65% and F1-score of 76.9%, with a relative improvement of 2.18% over the best-performing baseline and 1.25% over the best-performing state-of-the-art model. Additionally, KLKI-Fuzzy model is proposed for emotion recognition, which dynamically adjusts membership functions through Kullback-Leibler Kriging Interpolation (KLKI), enhancing emotion classification by processing features from all modalities. The KLKI-Fuzzy model exhibits enhanced emotion recognition performance with reduced fuzzification and defuzzification times.

Most programmers and users resort to find individual solution per problem depending on the data and nature of problem, that will lead to solve a specific problem using an algorithm without the ability of this solution to solve a new problem. This variance comes from the difference in algorithm parameters from one problem to another, as these parameters related to data nature, its size, and values it carried that can affect the way algorithm work. Individual solutions lead to increase in time cost and effort spent on solving a new problem, which the new problem requires to work on programming new criteria for algorithm solution. That is prompted us to highlight necessaries to develop main components for algorithms used in practical life, such as data mining algorithms so that a solution designed for one problem can be more easily adapted to new problems with different data structures, within the general scope of decision tree applicability. These algorithm components need control mechanism settings, so when using component to solve problem, there is no need to develop algorithm settings again, regardless data size and data structure. We found that the dynamic solution saves effort and time needed to solve problems with same algorithm. In this paper, we present our methodology for using ID3 decision tree algorithm to mine data dynamically, and the mechanism used to achieve the dynamic solution, that provides a flexible and reusable solution for a wide range of problems that require decision tree support, reducing the need to redesign or reimplement models for each new task The proposed model was tested on three datasets. The proposed model achieved an accuracy of 97%, 97%, and 93% on the breast cancer, heart disease, and diabetes datasets, respectively.

The classification of scientific articles faces challenges due to the complexity and diversity of academic content. In response to this issue, a new approach is proposed, utilizing Ensemble Learning, specifically Decision Tree, Random Forest, AdaBoost, and XGBoost, along with Convolutional Neural Network (CNN) techniques. This study utilizes the arXiv dataset, comparing the effectiveness of Term Frequency-Inverse Document Frequency (TFIDF) and Sentence-BERT (SBERT) for text representation. To further refine feature extraction, vectors derived from SBERT are integrated into the CNN framework for dimensionality reduction and obtaining more representative feature maps named latent feature vectors. The study also observes the impact of incorporating both the title and abstract on performance, demonstrating that richer textual information enhances model accuracy. The hybrid model (CNN + Ensemble Learning) demonstrates a substantial improvement in classification accuracy compared to traditional Ensemble Learning. The evaluation shows that CNN + SBERT with XGBoost achieved the highest accuracy of 94.62%, showcasing the benefits of combining advanced feature extraction techniques with powerful models. This research emphasizes the potential of integrating CNN within the Ensemble Learning paradigm to enhance the performance of scientific article classification and provides insights into the crucial role of CNN in improving model accuracy. Additionally, the study highlights the superior performance of SBERT in feature extraction, contributing beneficially to the overall model.