Abstract This study addresses the critical gap in reliable, scalable assessment of college students’ innovation and entrepreneurship capabilities by developing a digital, data driven evaluation system. Leveraging social media analytics, data mining techniques, and the Analytic Hierarchy Process... This study addresses the critical gap in reliable, scalable assessment of college students’ innovation and entrepreneurship capabilities by developing a digital, data driven evaluation system. Leveraging social media analytics, data mining techniques, and the Analytic Hierarchy Process (AHP), the framework constructs a three tier hierarchical indicator structure encompassing personal qualities (innovative awareness, entrepreneurial motivation, psychological resilience), skill capabilities (professional knowledge, practical innovation, teamwork), and learning experiences (self-directed learning, project participation, entrepreneurial practice). Through crowdsourcing principles and frequency based indicator refinement, the system transforms qualitative competencies into quantifiable metrics organized within a five level classification (A–E). Experimental validation with 1,250 undergraduate students demonstrated that the proposed system exhibited superior psychometric properties relative to conventional approaches, achieving higher reliability (Cronbach’s  = 0.857), operational efficiency (0.853), and structural validity (average factor loading = 0.275). The integrated platform deployed via cloud infrastructure with a WeChat miniprogram interface enables objective benchmarking, precise talent identification, and evidence based pedagogical interventions. By bridging digital youth engagement practices with rigorous multi criteria decision making, this research contributes a scalable, empirically validated instrument for educators and policymakers to systematically nurture innovation ready graduates. The framework’s robust performance supports its application in curriculum design, mentorship allocation, and resource optimization within entrepreneurship education ecosystems, ultimately advancing the development of future ready talent capable of driving socioeconomic transformation. Read More Read Less

Abstract This study presents the design and validation of an AI-integrated neuromuscular assessment system for objective evaluation of physical fitness, addressing the limitations of subjective clinical scales in rehabilitation and athletic training contexts. The proposed framework employs a... This study presents the design and validation of an AI-integrated neuromuscular assessment system for objective evaluation of physical fitness, addressing the limitations of subjective clinical scales in rehabilitation and athletic training contexts. The proposed framework employs a four layer modular architecture: (1) synchronized multimodal data acquisition using high density EMG arrays (8–16 channels), force plates, and inertial measurement units; (2) adaptive feature engineering with EMG Root Mean Square (RMS) as the primary biomarker; (3) a hybrid AI evaluation engine combining regression and classification heads; and (4) an intelligent feedback interface delivering tier based performance grading (A–E). Validation was conducted through a 40 day longitudinal training study with 500 simulation runs. Results demonstrated a strong positive correlation (r = 0.87, p < 0.001) between EMG RMS values and training duration, confirming RMS as a sensitive indicator of neuromuscular adaptation. The AI model achieved exceptional performance metrics, including 96.2% evaluation accuracy, 0.93 test retest reliability, and rapid convergence (MSE < 0.0015 by iteration 9) with minimal overfitting across training, validation, and test datasets. Mechanical parameters peak force (215 ± 28 N), average force (168 ± 21 N), joint angular velocity (3.42 ± 0.38 rad/s), and power output (412 ± 45 W) corroborated EMG derived physiological improvements. The system’s robustness was further validated through tightly bounded error distributions (mean test error: 0.0016 ± 0.0006) across repeated simulations. By transforming raw physiological signals into interpretable fitness scores with clinical grade reliability, this framework establishes a foundation for precision rehabilitation, personalized training optimization, and objective monitoring of neuromuscular function in both clinical and athletic settings. Read More Read Less

Abstract Multilabel multi instance text classification presents unique challenges due to the weak supervision setting where documents (bags) are labeled but constituent sentences (instances) are not, coupled with severe label imbalance where infrequent “tail” labels dominate real world... Multilabel multi instance text classification presents unique challenges due to the weak supervision setting where documents (bags) are labeled but constituent sentences (instances) are not, coupled with severe label imbalance where infrequent “tail” labels dominate real world distributions. Existing approaches typically employ label agnostic aggregation strategies such as max or mean pooling that implicitly assume uniform instance relevance across all labels, an assumption that is frequently violated in social tagging data, where individual sentences often signal specific labels while remaining irrelevant to others. To address this limitation, we propose LA-MIL (Label-Aware Attention Multi Instance Learning), a novel framework that employs dedicated attention heads for each label to enable fine grained, label specific instance selection. This architecture allows different labels to attend to distinct textual evidence within the same document, relaxing the restrictive assumption of uniform instance relevance. Evaluated on the DeliciousMIL benchmark dataset comprising 12,234 web documents annotated with 20 semantic tags, LA-MIL consistently outperforms traditional multi label classifiers, standard MIL models with global pooling, and attention based baselines with shared aggregation mechanisms. Notably, the model achieves significant improvements in macro-F1 scores, demonstrating superior handling of long tailed label distributions. Beyond quantitative gains, LAMIL provides inherent interpretability through learned attention weights that transparently identify labeldiscriminative sentences. Our results establish label aware attention as an essential architectural principle for multi label multi instance learning, particularly in applications requiring both accuracy on imbalanced distributions and human interpretable predictions. Read More Read Less