In this study, the author focused on the digital preservation and cross-cultural communication of Guangzhou Thirteen-Line Garden, a typical heritage site of the Maritime Silk Road. A dual-cycle Building Information Modeling-Convolutional Neural Network fusion framework was constructed to maintain dynamic consistency between geometric accuracy and texture semantics in garden reconstruction. The Multi-View Context Semantic Refinement algorithm was proposed to enhance texture recognition performance for complex components and weathered surfaces in Lingnan garden environments. A Web Extended Reality-knowledge graph multilingual interactive platform was built to support dynamic knowledge organization, real-time semantic reasoning, and cross-cultural narrative delivery. Experimental results demonstrated that the integrated framework significantly improves geometric-texture alignment, reduces system rendering delay, and effectively extends user engagement and learning experience. This approach unifies high-fidelity modeling, semantic knowledge reasoning, and public cultural services.
Aiming at the problem of non-music majors in college failing to convert abstract musical auditory information into stable aesthetic knowledge, this study constructs a framework for music teaching knowledge management. It establishes parametric coding rules to translate six core musical elements into standardized visual prompt knowledge and develops a full-process dynamic evaluation system to measure students' knowledge absorption, discussion participation, and knowledge migration. Taking Schubert's “The Devil” as a typical teaching knowledge carrier, three rounds of prompt iteration experiments optimize the matching between visual knowledge carriers and musical narrative knowledge nodes. Empirical results prove this knowledge governance framework reduces learners' cognitive burden, improves the internalization and long-term retention of music aesthetic knowledge, and clarifies the application boundary of intelligent knowledge tools. This research provides quantitative management standards for the whole-cycle production, iteration, and evaluation of music teaching knowledge.
Wellness tourism demand knowledge is fragmented across online travel agencies, social media, and wearable health applications, creating significant barriers to data-driven decision-making in the digital health economy. This study proposes a multi-source knowledge discovery framework that integrates heterogeneous big data to uncover latent consumer preferences and dynamic market trends. Using a Long Short-Term Memory predictive model with sudden-factor weight adjustment, the authors analyze three-year transaction records and social media sentiment to forecast wellness tourism demand with enhanced accuracy. Cluster analysis reveals distinct knowledge segments including middle-aged chronic-care seekers and young sub-health professionals with unmet service needs. The proposed framework transforms scattered multi-source data into actionable knowledge for product innovation and resource optimization, demonstrating how knowledge management systems can effectively bridge digital health ecosystems and tourism service operations.
This study investigates how digital humans and generative art can enhance inclusive, culturally informed cross-cultural communication. It first notes limitations of traditional AI models—such as cultural bias from skewed training data, poor localization, and rigid feedback mechanisms—that lead to misinterpretations or inauthentic cultural representation. An innovative model integrating multimodal context recognition (capturing language, facial expressions, tone) and dynamic feedback optimization is then proposed, with a multi-loop structure enabling real-time adjustments to cultural cues. Validated across five languages using datasets like MuST-C and real-world case studies, the model outperforms traditional counterparts: it achieves a 90 ms response time, high cultural adaptability, and up to 93.7% user satisfaction. Findings confirm the model's value in facilitating global cultural knowledge transfer and support its potential for educational, heritage, and international collaboration scenarios.
This study proposed a closed-loop knowledge management framework that integrates computer vision with craft rules to extract and redesign traditional carved lacquer relief patterns. A dataset of 120 complete images and 480 local segments was constructed across four motif types. Seven threshold metrics were established to evaluate pattern quality and design applicability. Modular experiments showed that adding boundary enhancement, style encoding, manual verification, and process rewriting progressively improved contour closure (from 0.78 to 0.91), symmetry stability (0.73 to 0.89), and redesign acceptance (74 to 87), while reducing misclassification (0.24 to 0.11). Among three redesign strategies, the balanced translation achieved the highest overall score (84) by preserving cultural authenticity while enabling moderate innovation. The findings demonstrated that a process-rewriting mechanism can effectively externalize tacit craft knowledge and prevent error propagation, offering a replicable knowledge management solution for the digital preservation of intangible cultural heritage.
The modern translation of traditional mortise-tenon craftsmanship is not merely a technical issue but a knowledge management process of externalizing tacit knowledge into explicit knowledge. This study constructs a dual-loop analytical framework consisting of a “tectonic loop” and an “aesthetic loop” to systematically compare the aesthetic reconstruction pathways of mortise-tenon joints in modern timber design between China and Japan. The results indicate that the Chinese pathway emphasizes joint exposure and transmission of cultural imagery, highlighting cultural recognizability, whereas the Japanese pathway prioritizes proportional control and modular integration, excelling in assembly efficiency and application adaptability. The two pathways achieve a dynamic balance among tectonic authenticity, cultural recognizability, and application adaptability. This research provides a theoretical framework and empirical reference for the knowledge transformation of traditional craftsmanship and cross-cultural design practices.
Classical poetry learning involves converting tacit cultural imagery into explicit understanding. This study develops a mobile blended learning platform that integrates visual narrative and interactive design to support knowledge transfer for Tang and Song poetry related to Anhui. A dual-loop mechanism—combining text-meaning generation and interactive optimization—is implemented, and learner engagement is quantified through behavioral logs (dwell time, annotation use, path navigation). A between-group experiment (N=89) shows the multimodal platform significantly improves knowledge absorption outcomes, including textual comprehension, cultural perception, and learner engagement compared to traditional static pages. Different design elements serve distinct knowledge management functions: visual narrative enhances cultural knowledge construction, annotation access deepens textual comprehension, and path interaction increases revisiting intention. This study provides a verifiable knowledge management framework for digital literary education.
With the digitalization of higher education, university physics teaching requires effective tools for knowledge organization, interaction, and students' understanding of concepts, formulas, and physical laws. However, general-purpose large language models often lack explicit modeling of structured physics knowledge, reducing their reliability in educational settings. To address this limitation, this paper proposes KGLM, a knowledge base–driven large language model framework for university physics courses. KGLM integrates domain-specific subword representations with knowledge graph embeddings to capture relationships among physics concepts, formulas, and physical processes. A bidirectional coupling mechanism injects explicit domain knowledge into answer generation, while a write-back mechanism enables continuous knowledge updating. Experimental results demonstrate that KGLM significantly improves question answering accuracy, equation consistency, and physical-law consistency, providing a more reliable and knowledge-centered solution for physics education.
Under the dual circulation pattern, manufacturing enterprises struggle with global brand building and overseas consumer knowledge accumulation. Cross-border e-commerce serves as an important channel and knowledge management carrier for brand internationalization. This study examines how platform capability, supply chain flexibility, digital marketing, and user data feedback affect manufacturing brand internationalization using multi-source data and quantitative methods including entropy weight and structural equation modeling. Results show that cross-border e-commerce platform capabilities significantly promote brand globalization, with supply chain and digital marketing as key mediators. User data feedback, as a core knowledge management capability, enhances digital marketing efficiency and long-term brand value. This study provides implications for manufacturing firms to achieve brand internationalization through knowledge management and cross-border e-commerce.
Digital protection of traditional culture is shifting from physical archiving to semantic computing. Aiming at the problems of existing data islands and content distortion, this paper puts forward a cooperative mechanism of generative protection and man-machine co-creation which is suitable for local cultural narrative. It relies on the knowledge map to disassemble the semantics of unstructured cultural resources and build a quantitative evaluation model. Combined with topological prior knowledge, the specification of digital twin high-fidelity restoration of cultural relics is formulated. It embeds historical prior knowledge and authenticity penalty factors to avoid cultural deviation caused by algorithm illusion. The national-level drama non-legacy demonstration proves the practicability of high-order graph neural network in analyzing complex semantics, and studies both technical application and humanistic core, providing quantitative technical support for the digital inheritance of traditional culture.
The inheritance of intangible cultural heritage (ICH) dances relies on oral transmission and demonstrations, which fail to preserve tacit movement knowledge. Video cannot capture joint angles, center-of-gravity shifts, or embedded semantics. This paper designs a motion capture and visualization system integrating multi-source acquisition, semantic mapping, and feedback writeback. Inertial nodes, depth cameras, and pressure plates enable synchronous calibration. Skeleton trajectories, rhythm deviations, and center-of-gravity shifts are decomposed into visual layers. Inheritor annotations and learner records are stored as writeback records for continuous updating. Results show skeleton fidelity exceeds 89 points across four dances. With 16 concurrent viewers, latency reaches 106 ms, exceeding the completeness threshold and indicating deployment limits. The capture + semantic scheme raises the median reproduction score to approximately 92 points. The core value lies in integrating capture accuracy, semantic interpretation, and knowledge management into a sustainable inheritance chain.
This study constructs a dual-loop digital translation model for intangible cultural heritage dyeing and weaving technology. It decomposes craft features into color, pattern, texture, and semantics to support modern clothing design. Data include 120 craft images, 18 inheritor interviews, 36 design schemes, 18 expert assessments, and 228 user questionnaires. Four design strategies are compared, namely direct transplantation, local reconstruction, structural translation, and color recoding. Results show that direct transplantation maintains high cultural authenticity but low wearability. Structural translation and local reconstruction balance cultural expression and modern adaptability. Statistical tests confirm significant differences among strategies in cultural recognition, aesthetic acceptance, and purchase intention. The dual-loop model ensures craft authenticity and design adaptation. This research provides a knowledge transformation path for heritage inheritance and market-oriented design application.
Financial digitalization (FinDig) is reconceptualized as an enhancement of organizational knowledge management capabilities, specifically the systematic codification, sharing, and application of financial knowledge. By reducing internal knowledge transfer barriers (management friction) and external decision noise (information asymmetry), FinDig can improve total factor productivity (TFP). Using panel data of Chinese listed firms from 2013 to 2023, this study constructs a FinDig index and estimate TFP via the Levinsohn–Petrin method. Results reveal an inverted U-shaped relationship between FinDig and TFP, with 96.5% of firms remaining on the left side of the turning point still enjoying positive returns. Mechanism tests confirm that reducing investment-inefficiency, particularly curbing over-investment, is the core channel through which FinDig affects TFP. This study offers a knowledge management-based micro-explanation for the Solow paradox in digital transformation.
Intangible cultural heritage dance recognition faces challenges such as high action similarity, small sample size, strong rhythm coupling, and insufficient semantic modeling. To address these issues, this paper proposes a knowledge-guided deep learning framework for skeleton-based intangible cultural heritage dance movement recognition. The method integrates spatio-temporal graph convolution, transfer learning, and multimodal semantic alignment, using cultural knowledge priors to enhance fine-grained feature discrimination and reduce sequence alignment errors. A multi-dimensional evaluation system including classification accuracy and dynamic time warping distance is constructed for comprehensive verification. Experiments on a self-built multi-genre dance skeleton dataset show that the proposed method significantly improves recognition accuracy, generalization ability, and temporal alignment stability compared with traditional models. This study provides an effective knowledge-driven solution for intelligent analysis and digital protection of intangible cultural heritage dances.
This study proposes a knowledge-based strategic cost management model integrating goal programming (GP) and accounting indicators to improve managerial decision-making in complex and competitive business environments. As global competition intensifies and cost structures become more intricate, organizations require decision support tools that treat cost not only as a control variable but also as a strategic performance driver. The model combines accounting indicators, including cost variances, activity-based costing, overhead allocation, and efficiency ratios, with GP to address multiple and conflicting cost-related objectives under operational and financial constraints. By integrating quantitative accounting data with qualitative managerial knowledge, organizational priorities, and experiential insights, the framework supports transparent and informed decision-making. Experimental findings indicate cost reductions of up to 18%, resource utilization exceeding 90%, and significantly lower deviations compared to traditional methods.
College students' psychological crises are concealed and diverse, making traditional questionnaire and interview screenings limited by low frequency and time lag. This study constructs a big data-driven closed-loop early warning system that fuses online learning behavior, social semantics, and wearable physiological signals. To address cross-modal heterogeneity, the authors use a multi-task deep learning architecture with Text-CNN, Bi-LSTM, and GCN to capture semantic, emotional, and temporal features. Dynamic risk classification and reinforcement learning scheduling enable real-time interpretable early warning, while differential privacy and federated learning ensure data security. Two-semester real-world deployment shows improved sensitivity and reduced false positives. The system integrates with campus platforms to provide customized intervention strategies, and experiments confirm low latency and energy consumption compared with baselines. This research provides a scalable paradigm for digital campus mental health governance.
While knowledge management systems (KMS) are widely deployed in vocational education, how they facilitate value-oriented knowledge internalization remains unclear. Drawing on the socialization, externalization, combination, internalization (SECI) model, this study examines how smart technology usage (STU) associates with ideological outcomes (IO) in vocational colleges, and via what mediating and moderating mechanisms. Analyzing platform log and survey data from 274 students using structural equation modeling (SEM), results show that STU does not directly improve IO. Instead, its effects are mediated by classroom participation (CP) and learning engagement (LE). Among frequency, timeliness, personalization, and visualization, feedback timeliness and process visualization are the strongest predictors of CP and LE. Technology self-efficacy (TSE) moderates the STU-LE relationship. These findings supplement the KM literature from a vocational education context, highlighting how technological resources interact with individual capabilities to shape value-oriented learning outcomes.
The increasing emphasis on STEAM education has expanded expectations for kindergarten teachers to design integrated and developmentally appropriate learning activities. However, the essential pedagogical knowledge needed for effective curriculum design is often experiential, tacit, and confined to individual classrooms. This study contends that teacher-led communities of ed-practice provide a practical mechanism for managing this knowledge. This examines a teacher-led community through a knowledge management lens to illustrate how pedagogical knowledge is shared, co-constructed, and transferred in kindergarten STEAM education. Using an exploratory qualitative approach, the study analyzes community interactions, reflective practices, and teachers’ curriculum artifacts. Analysis revealed three interrelated knowledge processes—externalization, co-construction, and adaptive transfer—through which pedagogical knowledge is circulated within the community. These results offer practical implications for institutions seeking systematic, collaborative approaches to curriculum design.
Against the backdrop of new quality productivity, this study addresses the link breakage and inefficient knowledge reuse in the industry-education-research integration of Jiangxi’s intangible cultural heritage (ICH) applied courses. The authors constructed a dual-cycle integration mechanism and a quantitative evaluation model, and conducted empirical analysis based on a three-dimensional dataset covering 8 ICH courses, 64 projects and multi-stakeholder feedback. Results reveal that new quality productivity factors exert differentiated empowering effects on curriculum effectiveness, with the research and application chains being the prominent weaknesses of the whole integration link. Curriculum integration level positively boosts industry adoption, while scene adaptability and feedback efficiency are key determinants of practical landing effects. The authors further propose targeted strategies to smooth the knowledge transfer and reuse loop, providing a knowledge management framework for ICH course reform and industry-education integration.
Based on digital twinning technology, this manuscript builds a life-cycle knowledge management system. This system covers many modules such as coding specification, version control, dynamic update, event-driven governance mechanism, and closed-loop evaluation. Verified by several typical scenarios, the integrity of knowledge association in most core tasks can reach more than 90%, and the training-related scenarios are close to 90%. The system establishes a two-way correlation between fault tracing and experience reuse and can provide risk prediction support in the work order allocation stage. The core value of the system lies in bringing knowledge update, verification, reuse, and elimination into the standardized governance framework, and providing a traceable and dynamic iterative platform for aviation maintenance knowledge management.
