Mobile terminals serve as the primary carriers of self-directed mobile learning among university students. Interaction data generated from these terminals can authentically reflect learners’ dynamic engagement levels. Existing research, however, inadequately accommodates the non-uniform temporal characteristics of behavioral sequences, struggles to capture long-range dependencies, suffers from poor fusion quality of heterogeneous features, and exhibits limited model interpretability—thereby hindering practical deployment in instructional intervention contexts. To address these issues, a unified framework for behavioral analysis and engagement prediction was proposed. Relying on non-intrusive sensing techniques, multimodal learning behavior data were collected, and learning engagement was quantified through the fusion of subjective and objective evaluation metrics. Temporal behavioral topological associations were uncovered by constructing horizontal visibility graphs integrated with a graph attention network. A cross-modal attention mechanism was employed to adaptively fuse multi-source heterogeneous features, while multitask learning was introduced to enhance model generalization, enabling accurate prediction of learning engagement trajectories. Furthermore, a gradient-weighted attribution strategy was incorporated to identify key influencing factors, and a mobile intervention mechanism was designed to optimize learners’ behavioral states. This study provides a technical reference for mobile learning behavior mining, learning state prediction, and the development of intelligent educational applications.
Low-latency and resource-efficient predictive analytics are essential for mobile edge computing applications, including smartphone-based human activity recognition (HAR). This study presents the Adaptive Latency Prediction and Scheduling (ALPS) framework, which aims to minimize inference latency, hardware energy consumption, and computational overhead while maintaining predictive accuracy. The ALPS framework comprises three primary modules: an adaptive principal component analysis (PCA) mechanism for real-time feature reduction, a lightweight ridge classifier optimized with an L regularization loss 2 function for rapid multi-class activity prediction, and a dynamic task scheduler that minimizes a combined latency-energy cost function to allocate processing tasks across mobile, edge, and cloud layers. Evaluation on the high-dimensional UCI-HAR smartphone dataset (10,299 samples, 561 features) demonstrated that the adaptive feature reduction module reduced the feature space to 68 principal components, resulting in an 87.9% reduction in dimensionality while maintaining 95% cumulative explained variance. Relative to conventional standalone models and isolated optimization baselines, ALPS achieved a 32% reduction in average end-to-end latency (120 ms), a 25% decrease in energy consumption (180 J), and a classification accuracy of 96.4% across six physical activities. The primary contribution of this study is the unified integration of adaptive data compression and distributed infrastructure scheduling into a scalable and energy-efficient pipeline for real-time edge intelligence.
The primary aim of this study is to provide a comprehensive and structured synthesis of existing research to understand how synthetic data is conceptualized, generated, and utilized within educational contexts. By analyzing 29 peer-reviewed articles, the research identifies seven primary dimensions of application: privacy and data sharing, data augmentation, NLP/ text generation, predictive modeling, pedagogical design, methodological analysis, and synthetic data in mobile, interactive, and adaptive learning systems. A significant finding is the increasing integration of artificial intelligence (AI) and machine learning technologies, such as generative adversarial networks (GANs) and large language models (LLMs), which are now central to generating high-fidelity artificial records and augmenting qualitative datasets. Across these analytical, predictive, and pedagogical domains, synthetic data offers a viable response to persistent challenges related to data scarcity, privacy constraints, and limited data accessibility in education. The findings indicate a growing reliance on synthetic generation as an emerging methodological response to data-intensive demands. While synthetic data supports advanced modeling, adaptive learning systems, and instructional design, its epistemological legitimacy and methodological robustness remain contingent on rigorous validation practices. The study concludes that the field currently lacks standardized validation protocols, particularly regarding subgroup equity and fairness. Establishing transparent, equity-aware frameworks remains essential for the future integration of synthetic data into applied educational systems.
This study explores behavioral patterns of generative artificial intelligence (GenAI) usage in higher education using a data-driven segmentation approach. Moving beyond intention-based models such as the technology acceptance model and the Unified Theory of Acceptance and Use of Technology, the study focuses on actual usage behavior and user heterogeneity. Data were collected through a structured questionnaire measuring AI usage behaviors and psychological factors, including trust and AI anxiety. A machine learning–based clustering approach, specifically K-means clustering, was employed to identify user segments, while ANOVA examined differences across clusters. The findings reveal three distinct groups: skeptical users, pragmatic users, and power users, with significant differences in usage frequency, trust, anxiety, and behavioral intention (p < 0.001). Notably, power users exhibit both high trust and elevated anxiety, highlighting the complex psychological dynamics of intensive AI usage.
The excessive use of artificial intelligence (AI) and mobile learning apps in Chinese universities has changed how students experience their studies and has led to higher levels of psychological distress. Using Self-Determination Theory, this study examines how various aspects of mobile empowerment relate to psychological distress. This study used a quantitative, non-experimental design. Researchers collected data through an online survey from 472 undergraduate and postgraduate students at four universities in Fujian Province, China. The data were analyzed with multiple linear regression and structural equation modelling (SEM) using AMOS. The results show that Confirmatory Factor Analysis (CFA) indicated a strong model fit and good construct validity. The RMSEA values were 0.023 for the independent-variable model and 0.011 for the dependent-variable model. The SEM results showed that digital competence significantly lowered generalised anxiety (B = −0.435, p < .001) and psychological stress (B = −0.401, p < .001). Digital relatedness also predicted lower anxiety (B = −0.306, p < .001) and stress (B = −0.228, p < .001). These findings suggest that digital competence and supportive peer connections are important protective factors against psychological distress. However, digital autonomy by itself may not reduce stress in highly connected, AI-driven learning environments.
Virtual machine (VM) placement in recent cloud systems remains a challenging and new task due to the challenge of finding the best way to use resources efficiently while also ensuring service reliability. The current study offers an in-depth comparative evaluation of six VM placement algorithms—three established heuristics (Best Fit Decreasing (BFD), First Fit Decreasing (FFD), and Worst Fit (WF)), a baseline RANDOM method, and two metaheuristics (Ant Colony Optimization (ACO) and NSGA-III)—conducted through 2,400 simulation trials. We meticulously evaluate each algorithm across four critical performance metrics: placement success rate, energy consumption, execution time, and service reliability. The statistical analysis confirmed significant differences in performance across all methods using the Kruskal-Wallis and Mann-Whitney tests. Classical heuristics achieve a placement success rate (PSR) of 87.38% with BFD and FFD operating without any service violations. These methods use more energy than other approaches, which is a drawback of their good performance. NSGA-III reduces energy usage by 11% while maintaining acceptable placement performance (71.62%). The execution time (ET) can really vary, from super-fast heuristics to more time-consuming optimization methods. The results indicate that classical heuristics provide the highest reliability. No SLA violations (SLAV) were observed in the scenarios evaluated. Classical heuristics, in particular BFD and FFD, are therefore the most reliable. The study’s findings also show that cloud providers can choose the algorithms that best match their operational priorities. These priorities may include maximizing performance, minimizing energy consumption, or ensuring service stability.
The use of adaptive learning systems with artificial intelligence (AI) in education has changed how educators teach by providing a customized learner experience that matches the particular characteristics of each learner, thus increasing both learner engagement and learning outcomes. This article provides a description of the design, development, and evaluation of a generative AI-based interactive mobile learning system designed to increase learner engagement, increase the ability for the learner to acquire knowledge, and provide improved access to educational resources. By using a mobile-based learning platform, learners have access to real-time tutoring, generated, determined feedback, generated instructional materials, and interactive learning experiences, 24 hours a day, seven days a week. The system architecture consists of user profiling, a learning analytics engine, and a content generation process for creating dynamic course materials using GenAI and providing feedback and making recommendations. These four components of the system work together to create personalized learning paths for individual learners and to enhance the ongoing learning processes of all learners. A thorough assessment of the proposed approach was conducted by utilizing multiple methods with students from a wide variety of educational backgrounds in order to measure the effectiveness of the AI-based adaptive learning system. Students’ performance (effectiveness) was measured using both quantitative metrics and qualitative feedback, including measures of learning success (i.e., test scores), levels of engagement (i.e., how often students engage with the material), completion rate (number of tasks completed vs. total number of tasks assigned), and satisfaction (whether learners felt satisfied with their adaptive learning experience). Based on the results of all evaluations performed, it has been found that AI-based adaptive learning systems provide greater motivation, more successfully comprehend course content, and a higher level of academic performance compared to the use of traditional mobile learning apps.
This is a qualitative systematic review examining the effectiveness of the combination of the extended reality (XR) technologies and biofeedback mechanisms with social-emotional learning (SEL) outcomes in school-aged learners with a specific emphasis on adolescent girls. Based on the 2020–2025 peer-reviewed literature, the review will combine the results of different educational and clinical settings and explore the role of immersive digital environments and physiological feedback systems in emotional regulation, empathy building, and metacognitive development. Instead of using a strict method such as PRISMA, they use the thematic synthesis based on the self-regulated learning (SRL) model, which is composed of the forethought, performance, and self-reflection stages to investigate the effects of technology-based interventions on emotional and cognitive learning processes. The analysis demonstrates that XR environments can support the learner presence, motivation, and perspective-taking in early goal-setting and emotional awareness and regulation in performance by real-time biofeedback. The literature however also identifies gaps especially in long term efficacy, gender specific-design, and scalability of these tools in mainstream classrooms. The review suggests that an Immersive Hybrid Feedback model is a conceptual intermediate between affective computing and educational design as an example of the potential and constraints of adopting embodied technologies into social-emotional learning.
Against the backdrop of the deep integration of mobile technologies and educational digitalization, financial management education has been constrained by insufficient levels of virtualization, limited adaptability to individual learners, and inadequate real-time interactive experiences. Existing educational frameworks have struggled to reconcile the computational constraints of mobile devices with the demands of immersive learning environments. To address these challenges, an intelligent education framework centered on mobile devices and organized around a collaborative three-layer architecture was proposed, with the objective of transcending the limitations of traditional instructional models and enabling deeply virtualized and personalized financial management education tailored to the lightweight and ubiquitous characteristics of mobile technologies. Through the coordinated operation of a mobile interactive terminal layer, an edge intelligence layer, and a cloud-based intelligent hub layer, the proposed framework emphasizes innovations in intelligent mobile interaction design, dynamic allocation of edge–cloud computational resources, and multimodal data-driven personalized adaptation. A closed-loop, collaborative, personalized learning ecosystem is thereby constructed. Prototype system evaluations demonstrate that the proposed framework effectively enhances learning immersion, personalization, and learning efficiency in mobile learning scenarios, providing a novel technological paradigm for the deployment of mobile intelligent education in professional disciplines.
Selecting an effective construction project manager is critical to project success, given the complex, high-risk, and people-intensive nature of construction environments. While existing selection frameworks emphasize technical expertise and managerial experience, behavioral and emotional factors that influence leadership and team performance are often assessed subjectively or overlooked. Moreover, conventional multi-criteria decision-making (MCDM) approaches remain largely static and are not designed to capture emotion-related dynamics during evaluation. To address these limitations, this study proposes an integrated framework that combines a mobile interactive system for affective computing with MCDM techniques for construction project manager selection. Facial expressions elicited during structured interview scenarios are analyzed using deep learning–based emotion recognition, incorporating both categorical emotions and pleasure–arousal–dominance (PAD)–based intensity measures to quantify emotional intelligence objectively. These affective indicators are then integrated with technical, managerial, and communication criteria within an MCDM framework to generate systematic and transparent candidate rankings. The proposed approach advances existing selection methods by operationalizing emotional intelligence as a measurable decision attribute and embedding it within a structured decision-support system. By enabling the joint evaluation of cognitive and affective competencies through a platform-independent framework, this study offers a more holistic, consistent, and context-aware basis for construction project manager selection, with practical implications for improving managerial effectiveness and project outcomes.
In mobile touch interaction environments, a mismatch between marketing information presentation and users’ cognitive load often results in suboptimal interaction experiences and low commercial conversion efficiency, thereby constraining the advancement of mobile marketing optimization. This study proposes an integrated technical framework that combines real-time multimodal cognitive load quantification with reinforcement learning–based adaptive decision-making to dynamically align marketing information presentation with users’ real-time cognitive states. The framework consists of two core modules: a multimodal real-time cognitive load estimation model and a reinforcement learning–driven adaptive information presentation decision engine. The former synchronously collects multimodal data—including touch interaction behaviors, eye-tracking signals, and basic physiological indicators—and constructs a high-discriminability feature system integrated with a temporal multi-head attention fusion network. This design enables precise, millisecond-level cognitive load quantification without reliance on bulky laboratory equipment. The latter module treats cognitive load as the primary state signal, designs a multi-objective reward mechanism that balances shortterm user experience and long-term commercial value, and employs a cloud–edge collaborative deployment architecture to achieve dynamic and adaptive adjustment of marketing information presentation strategies. The two modules are tightly coupled through a real-time data streaming pipeline, effectively addressing the challenges of multimodal synchronization and low-latency decision-making in mobile environments. Experimental results in mobile marketing scenarios demonstrate that the proposed framework accurately captures users’ real-time cognitive load, significantly optimizes information presentation effectiveness, and enhances both interaction experience and commercial conversion efficiency.
The importance of environmental, social, and governance (ESG) in the procurement process of the engineering and construction industry cannot be overemphasized. However, supplier ESG reporting is still done in a traditional manner. This paper introduces a conceptual model that combines mobile-based data capture and analytics of artificial intelligence (AI) to support sustainability in construction and engineering. Digital technologies are the main devices for the real-time capture of ESG data in supplier ecosystems. The model demonstrates how managers, AI supporting them, can obtain ESG dashboards to enhance supplier selection, compliance with the procurement’s sustainability criteria, and rational decision-making. Practical examples from the cement and steel supply chains are provided. The author’s contribution stems from presenting the model in theory and practice, where its use brings ESG integration into the procurement processes in sustainable, transparent, and scalable supply chains.
Traditional physical education (PE) instruction suffers from delayed motion feedback, homogeneous guidance strategies, and the reinforcement of incorrect movement patterns. Existing sensor-based motion recognition systems often struggle to simultaneously achieve real-time responsiveness, personalized instruction, and robust adaptation to complex instructional scenarios. To address these challenges, this study proposes an intelligent PE system that integrates wearable sensors with smartphone-based interaction, forming a full-chain technical framework from multimodal data acquisition to adaptive intervention. The proposed system introduces several key innovations: (1) a multimodal semantic fusion strategy to enhance motion recognition accuracy and contextual understanding; (2) an early event detection mechanism that enables advanced prediction of motion errors and millisecond-level real-time feedback; (3) a personalized adaptive intervention mechanism that dynamically accommodates learners with different skill levels; (4) an edge-cloud collaborative architecture that balances real-time processing on mobile devices with in-depth analytical capabilities; and (5) the construction of a multimodal sports motion dataset and evaluation benchmark for instructional scenarios. This study provides a novel technical paradigm for intelligent mobile PE, and the released dataset and benchmarks offer valuable resources for future research, promoting the digitalization and intelligent transformation of physical education.
A systematic mapping of artificial intelligence (AI)-based mobile driver coaching for behavior change and road safety in land transportation has been conducted following PRISMA, resulting in nine empirical studies (n = 9). The evidence is organized through a taxonomy covering intervention goals, AI techniques, data sources, deployment patterns, outcome and engagement metrics, and privacy/dependability safeguards. Across the mapped interventions, a substantial share targets specific risky behaviors, particularly phone-related distraction, reporting relative reductions of up to 21% versus controls in controlled evaluations. Deployment designs are predominantly hybrid, combining on-device inference for timely detection with server-side components for aggregation and program features. Based on the extracted evidence, a system-oriented design framework and practical guidelines are proposed to support researchers and mobile system designers in selecting architectures and evaluation metrics for interactive driver coaching. Key limitations include heterogeneous outcome definitions and predominantly short- to medium-term follow-up, which constrain comparability and generalization across settings.
Traditional mobile learning systems for college English commonly suffer from high interaction latency, coarse-grained feedback, poor contextual adaptability, and insufficient utilization of on-device computing resources, which significantly constrain the learning experience and instructional effectiveness. To address these challenges, this study integrates mobile heterogeneous computing, lightweight on-device models, augmented reality (AR) visual interaction, and context-awareness technologies to construct an interactive intelligent mobile system aimed at improving teaching effectiveness in college English. The research focuses on optimizing mobile-side technical implementation by designing a heterogeneous computing scheduling engine tailored for mobile terminals to enable efficient collaborative execution of multiple on-device tasks. A lightweight on-device speech assessment and prosody correction algorithm is proposed to reduce reliance on cloud computing and improve feedback accuracy. In addition, a low-power, highly robust AR-based interactive corrective feedback mechanism is developed to address pronunciation organ movement correction in mobile learning scenarios. Furthermore, a context-aware collaborative filtering recommendation strategy is introduced to seamlessly adapt instructional content to fragmented mobile learning contexts. Multi-dimensional experimental results demonstrate that the proposed system effectively reduces interaction latency, optimizes mobile power consumption and runtime stability, and significantly enhances the effectiveness of college English speaking instruction. This study provides a novel solution for deep integration of technology and pedagogy in mobile education and offers technical references for related research and engineering applications.
Academic English writing on mobile platforms has been constrained by limited screen-based interaction, insufficient adaptation to fragmented learning contexts, and feedback mechanisms that lack pedagogical depth and individual specificity. To address these challenges, a mobile-native interactive English writing guidance system integrating natural language processing (NLP) was designed. Three core innovations were introduced. First, a touchoptimized interaction paradigm was constructed to enable direct interaction between text and feedback. Second, an ontology-based explainable feedback mechanism was proposed to enhance feedback precision and instructional value. Third, a dynamic assessment-driven progressive feedback generation algorithm was developed to adaptively support learners at varying proficiency levels. The system with a multi-agent collaborative architecture integrates a lightweight academic writing ontology knowledge base with mobile-adapted NLP model optimization strategies. Experimental results demonstrated that the proposed system significantly outperformed mainstream baseline approaches in interaction efficiency, feedback quality, and writing performance improvement. Task completion time was reduced by 21.6%–24.1% compared with the control group, feedback precision reached 89.7%, and a 2.1-point improvement was observed in three-dimensional writing quality scores. Correlation analysis revealed a significant negative relationship between interaction efficiency and user dissatisfaction, while a significant positive relationship was identified between feedback explainability and error correction rates. Ablation experiments further confirmed the critical contribution of the three proposed modules to overall system performance. This study establishes a novel paradigm for intelligent writing guidance in mobile contexts and advances the deep integration of digital education and mobile artificial intelligence (AI).
The lack of information literacy among learners in open education has become a core bottleneck hindering global access to high-quality education. Existing mobile learning interventions suffer from unclear objectives and fragmented processes, with disputes over the fragmented value of mobile learning remaining unaddressed. Traditional learning analytics frameworks are also inadequate to meet the real-time and contextual needs of ubiquitous mobile learning environments. This study aims to develop a mobile-specific information literacy enhancement framework, ML-ILMDF, based on a contextualized literacy development model. The framework specifies its technical implementation details and establishes dynamic coupling strategies for context and literacy to provide precise interventions. The study systematically validates the educational effectiveness, core mechanism rationality, and engineering feasibility of the framework. A quasi-experimental study with 320 global open university learners, lasting 16 weeks, integrates multi-source mobile data collection, five-dimensional information literacy assessments, and sub-studies using dynamic hypergraphs and collaborative filtering recommendation algorithms for comparison. Simultaneously, the technical performance evaluation of the framework is conducted. The innovation of this study lies in proposing a contextualized literacy development model that achieves dynamic coupling between micro-contexts and macro-literacy, clarifying the mobile-specific technical architecture and implementation details of ML-ILMDF to overcome the limitations of traditional frameworks. It empirically addresses the fragmented nature of mobile learning and offers a practical, actionable paradigm for cultivating information literacy in open education through mobile learning technologies.
Contemporary mobile music applications are largely limited to interface-level control mechanisms, making it difficult to achieve deep understanding of users’ creative intentions and truly collaborative interaction. Moreover, the trade-off among resource constraints, multimodal signal fusion efficiency, and real-time generation quality constitutes a fundamental bottleneck for creative AI applications on mobile platforms. To address these challenges, this paper proposes and implements a multimodal sensor fusion-based mobile system for real-time music generation, enabling synergistic interaction among gesture, posture, and acoustic environment cues. The system adopts a hierarchically decoupled edge-intelligent music interaction architecture, consisting of a multimodal music semantic understanding network and a resource-adaptive generation engine. The former employs a lightweight cross-modal attention mechanism to directly map heterogeneous sensor signals—captured from gestures, body posture, and acoustic environments—into a structured music semantic space, effectively translating low-level perceptual signals into high-level creative intent. The latter dynamically switches generation modes according to real-time device states, achieving an optimal balance between computational resource consumption and generation quality. To support personalization while preserving user privacy, we design an edge-personalized learning framework that combines cloud-based federated meta-learning pretraining with on-device incremental fine-tuning. Experimental results demonstrate that the proposed system achieves an end-to-end P99 latency below 45 ms on mainstream mobile devices, while significantly reducing power consumption and memory footprint. The generated music exhibits superior performance in terms of melodic fluency and harmonic consistency. This work pushes the deployment boundary of creative AI under resource-constrained environments and establishes a technical paradigm for multimodal interaction and intelligent music generation on mobile platforms, providing key technological support for next-generation interactive artistic experiences.
This study looked at how mobile-driven strategies affect organizational agility and strength, with a focus on mobile technology capability, digital transformation, organizational resilience, and operational innovation capability. Researchers used a quantitative approach and collected data from 225 respondents via a structured questionnaire based on validated scales. The analysis used Partial Least Squares Structural Equation Modeling (PLS-SEM) to evaluate the measurement model and the relationships between the main concepts. The results showed that mobile technology capability and digital transformation both had strong positive effects on organizational agility. This agility then helped improve organizational resilience, which in turn boosted operational innovation capability. Mediation analysis also confirmed that organizational agility played a key role in linking mobile technology capability and digital transformation to organizational resilience. For managers, the study suggests that mobile and digital transformation efforts should go hand in hand with practices that build agility to strengthen resilience and drive innovation. What makes this study unique is that it brings these ideas together in one framework and shows how organizational agility helps turn digital capabilities into organizational strength.
The digital transformation of financial ecosystems has accelerated in recent years, with mobile payment systems emerging as a cornerstone of modern commerce. While much of the existing research has emphasized consumer adoption of mobile payments, the determinants that drive businesses to adopt and integrate these systems remain comparatively underexplored. Businesses, as critical stakeholders in the payment ecosystem, not only influence consumer usage but also shape the broader trajectory of financial inclusion and digital economy growth. Against this backdrop, this study applies the perceived e-readiness model (PERM) to investigate the factors influencing business adoption of mobile payment solutions. The model emphasizes organizational readiness for the adoption of mobile payments and studies both internal and external readiness factors that impact the adoption of mobile payment technology. The internal readiness is accessed by perceived organizational e-readiness (POER) constructs like awareness, human resource, business resource etc., and the external readiness is accessed by the readiness of the Government, market forces, and the readiness of the competitors. The study was conducted on 144 businesses spanning multiple sectors. The data was collected through a survey using a questionnaire on a Likert scale. The data was analysed by employing partial least squares structural equation modelling. The findings reveal that awareness, human resources, and strategic alignment are the most critical enablers of mobile payment adoption among businesses. While external ecosystem factors (government, market forces, and support industries) matter, the readiness and commitment of the businesses themselves through informed leadership, skilled employees, and strategic planning play the dominant role. The study contributes to both theory and practice by extending the application of PERM to the context of digital financial services and offering actionable recommendations for policymakers, technology service providers, and business managers.
