Minnan University of Science and Technology

With the rapid development of e-commerce, accurately predicting product sales and achieving dynamic inventory control have become key challenges for enterprises to optimize supply chain management. The traditional Prophet model excels at capturing long-term trends and seasonal characteristics in time series, but has limited modeling capabilities for complex nonlinear relationships. Although LSTM neural networks can effectively learn the dynamic dependencies of sequential data, they tend to overlook explicit temporal patterns. This study constructs an LSTM-Prophet fusion model, first utilizing Prophet to decompose the trends, seasonality, and holiday effects of sales data, and then inputting the residuals into LSTM for nonlinear correction. Finally, the prediction accuracy is improved through weighted fusion. Experiments based on 150,000 daily sales data from an e-commerce platform over three years show that the average absolute error of the fusion model is reduced to 8.7, which is 29.3% and 17.1% lower than that of the single Prophet (12.3) and LSTM (10.5) models, respectively, and the root mean square error decreases by 22.6%. In inventory control simulations, this model drives an 18.4% increase in inventory turnover rate and reduces the out-of-stock rate to 3.2%, effectively balancing prediction accuracy and computational efficiency. However, the model still relies on high-quality historical data, has high computational complexity, and has limited adaptability to new product launches and emergencies.

Purpose This study aims to examine how perceived benefits and risks of technology (GenAI) relate to students’ adoption intentions, which in turn are associated with their development of entrepreneurial skills. The research also uses entrepreneurial opportunity recognition and entrepreneurial competencies as moderator variables. Design/methodology/approach The data were collected from students across 15 nations. The study develops and tests a conceptual model with exogenous and endogenous variables using covariance-based structural equation modeling. The knowledge-based view (KBV) theory supports the theoretical examination across variables. Findings The perceived technological benefits of GenAI are strongly and significantly associated with students’ adoption intentions. Furthermore, the adoption of GenAI as a technology constructively develops students’ entrepreneurial skills. The perceived risks attached to GenAI are not directly linked to students’ GenAI adoption. Entrepreneurial competencies amplify the impact of both perceived benefits and risks of GenAI on students’ adoption intentions, while entrepreneurial opportunity recognition weakens the relationship across perceived GenAI benefits and their intentions to adopt. Practical implications The study highlights the benefits of GenAI technology in academia, directly enhancing students’ entrepreneurial skills. The benefits and risks associated with the technology (GenAI) provide important implications for academic institutions, learners and policymakers. Originality/value This study strengthens the KBV by demonstrating that GenAI not only works as a technological enabler for users but also as a knowledge resource that enhances their entrepreneurial capabilities in the digital entrepreneurial ecosystem.

This study focuses on the design and optimization of consumer behavior prediction models in online e-commerce reviews. To address the issues of slow convergence and insufficient robustness in the traditional Q-learning reinforcement learning algorithm, this article introduces a probabilistic action-selection algorithm. This algorithm employs a multi-step, iterative mechanism that uses instantaneous differencing to increase the likelihood of selecting high-Q actions during model iteration, thereby accelerating the solution process and ensuring robust network optimization. Given the nonlinear and high-noise characteristics of consumer behavior time-series data in e-commerce reviews, we propose a hybrid intelligent prediction model, Q-learning-Artificial Neural Network-Hidden Markov Model (QL-ANN-HMM), that effectively reduces the impact of systematic random errors and significantly improves prediction accuracy. Experimental results demonstrate that the improved Q-learning algorithm achieves 2.71% and 5.96% improvements in mean absolute percentage error (MAPE) and normalized mean squared error (NMSE), respectively, compared to the traditional Q-learning algorithm on the Amazon Reviews 2023 and Flipkart Reviews datasets. Additionally, the QL-ANN-HMM model achieves lower mean absolute error (MAE), MAPE, and NMSE values on both datasets, recorded at 0.0195, 0.019, and 0.0189, respectively. This research not only provides novel theoretical support and technical methods for predicting consumer behavior in online e-commerce reviews but also enables e-commerce platforms to more accurately track market dynamics, optimize resource allocation, and achieve sustainable development by comprehensively analyzing consumer behavioral data.

This study designs and evaluates a mobile-assisted language learning system that integrates knowledge graphs with multimodal interactionincluding touch, voice, and augmented reality -to support deep knowledge comprehension.A controlled experiment compared an experimental group using the proposed system against control groups using either a Bayesian knowledge-tracing tutor or a conventional multimodal application.Results demonstrated that the experimental group achieved significantly higher posttest scores, with a learning gain improvement of nearly 40%.Cognitive load measurements showed a significant reduction, supported by a greatly increased effective interaction ratio.Over 85% of interactions were meaningful, contributing to enhanced knowledge structure formation.These outcomes confirm that tightly coupling knowledge-aware scaffolding with multimodal mobile interaction improves learning efficiency, reduces cognitive burden, and supports integrated knowledge framework development.This research provides theoretical and practical implications for designing intelligent, cognitive-friendly mobile learning systems.

Currently, we have entered the AI era, against which college and university innovation and entrepreneurship education have also ushered in new opportunities for reform. In this regard, this paper conducts an in-depth analysis of AI-empowered innovation and entrepreneurship education in colleges and universities. By dissecting the internal mechanism of AI empowering college and university innovation and entrepreneurship education, it sorts out in detail the dilemmas faced in the previous process of college and university innovation and entrepreneurship education. Finally, it puts forward targeted and effective implementation paths, aiming to provide some references for the reform of college and university innovation and entrepreneurship education.

This study provides an in-depth investigation into the conducted interference mechanisms within variable frequency motor drive systems, aiming to offer theoretical guidance for system design and optimization. Initially, the composition, control strategies, and key design parameters of the variable frequency motor drive system are introduced, laying the groundwork for subsequent analysis. Following this, the conducted interference mechanisms of the Boost PFC circuit are thoroughly explored, including an analysis of interference current conduction paths, proposed suppression methods, and the development of equivalent circuit models for common-mode and differential-mode conducted interference. Building upon this foundation, a comprehensive conducted interference model for the variable frequency motor drive system is constructed, with its interference characteristics validated through experimental analysis. The research reveals that conducted interference primarily originates from the switching actions of power devices in both the front-end Boost PFC and the rear-end three-phase motor drive circuits. Specifically, commonmode interference results from the superposition of noise from both stages, while differential-mode interference is predominantly caused by the front-end MOSFET. These findings provide crucial theoretical support and practical guidance for the suppression of conducted interference and the design of EMI filters.

Algorithms are fundamental to solving complex problems in science and engineering. However, conventional methods often struggle with nonlinear, high-dimensional landscapes. The biologically inspired dung beetle optimization algorithm has offered promising solutions but is still prone to premature convergence and falling into local optima. This article presents a multi-strategy enhanced dung beetle optimization algorithm that integrated tent chaotic mapping for population initialization, a golden sine strategy for position updating, Lévy flights to escape local minima, and dynamic weighting coefficients for adaptive search balancing. These strategies collectively enhanced population diversity and improved the balance between exploration and exploitation. Benchmarking on the CEC2017 test suite and real-world engineering problems demonstrated that the multi-strategy enhanced dung beetle optimization algorithm achieved superior convergence speed, solution accuracy, and robustness when compared with the standard dung beetle optimization algorithm and other state-of-the-art metaheuristics.