Guangdong Industry Technical College

DeepSeek has emerged as a prominent representative of China’s domestic large language models(LLMs), attracting widespread public attention and generating diverse discussions on social media since its release. To systematically examine public sentiment and thematic concerns surrounding DeepSeek, this study analyzes 86,008 Weibo posts collected between January 19 and March 22, 2025. After data cleaning, 59,679 valid entries were retained. A mixed-method approach was adopted, combining Latent Dirichlet Allocation (LDA) for topic modeling and a fine-tuned BERT-based classifier for sentiment analysis. The BERT model, fine-tuned on 10,000 hand-labeled posts, achieved an accuracy of 82% and was subsequently applied to the remaining unlabeled data. Topic modeling revealed nine major themes in public discourse, including chip innovation and industrial transformation, smart device ecosystem, and China–US technological competition. Sentiment analysis showed that 46.3% of the posts expressed a positive attitude, reflecting public endorsement of DeepSeek’s domestic innovation and application potential, although concerns about information security, technical feasibility, and international political dynamics were also evident. Additionally, major events significantly influenced emotional fluctuations, with peaks in public attention observed on January 27 and March 19. These findings offer empirical insights for policymakers, developers, and the broader public into public acceptance, key concerns, and potential risks associated with emerging AI technologies, and support the formulation of informed strategies for technology governance, user engagement, and product optimization.

IntroductionExploitative leadership, regarded as a manifestation of dark leadership, has recently attracted heightened academic interest because of its harmful consequences for employees. However, existing research has paid relatively little attention to the ways in which such leadership shapes critical employee outcomes, including intentions to leave and procrastination at work. Framed within conservation of resources theory, this study probes the latent pathways through which exploitative leadership erodes employee resources, precipitating both turnover intentions and procrastinatory conduct, while highlighting contextual levers that might temper these deleterious effects.MethodsTo test the proposed hypotheses, we collected two-wave matched data from 296 full-time employees across various industries. Hierarchical regression and bootstrapping analyses were employed to assess the mediation and moderation effects.ResultsThe results unveiled that exploitative leadership precipitates heightened turnover intentions and procrastinatory behavior among employees, with self-control depletion operating as a pivotal mediator. Crucially, perceived organizational support functioned as a buffering mechanism, dampening both the impact of exploitative leadership on self-control depletion and its downstream indirect effects on turnover intention and workplace procrastination.DiscussionThis study contributes to the growing literature on exploitative leadership and its consequences for employee attitudes and behaviors. It extends our understanding of the interactive effects between exploitative leadership and organizational support, and uncovers the critical role of self-control resources in shaping employees’ withdrawal responses. Practical implications for organizational leaders are also discussed.

To solve the problems of insufficient extraction accuracy, disconnection from packaging scene adaptation, and weakened cultural connotation transmission of intangible cultural heritage elements in packaging design, this study carries out the practice of intangible cultural heritage element extraction and algorithm optimization. Research and construct an integrated architecture of “multimodal data layer intangible cultural heritage element extraction layer packaging adaptation optimization layer”, integrating traditional image processing and deep learning techniques: the data layer ensures the adaptation of intangible cultural heritage image quality and packaging printing accuracy through relative total variation denoising and standardization processing; The extraction layer integrates layered extraction logic and attention module to achieve accurate capture of contour, color, and line features; The adaptation layer combines 3D virtual simulation and process validation rules to achieve deep adaptation of elements, packaging shapes, and processes. The experimental results show that the pixel accuracy (PA) of contour extraction under this architecture reaches 96.1 %, and the intersection to union ratio (mIoU) of complex element segmentation reaches 93.1%; The adaptation deviation of three typical packaging forms is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\leqslant 1.5 \text{mm}$</tex>, and the process adaptation qualification rate is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\geqslant 90.8\%$</tex>; The cultural fidelity index (CFI) of intangible cultural heritage elements is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\geqslant 0.91$</tex>, effectively avoiding cultural symbolization collage. This study provides a technical framework for the large-scale and high-quality application of intangible cultural heritage elements in the packaging field, promoting the deep integration of intangible cultural heritage culture and modern packaging design.

With the continuous expansion of accounting data scale, the traditional single machine processing mode faces the dual challenges of low efficiency and accuracy guarantee. General parallel computing technology is difficult to meet the requirements of accounting data structuring, decimal accuracy, and compliance due to the lack of adaptation to accounting business characteristics. This study aims to solve the problem of efficient and compliant processing of accounting data, and constructs a three-layer parallel processing framework of “data algorithm application”. Three types of accounting specific parallel algorithms are designed: dynamic sharding algorithm based on “account timestamp composite hash” to ensure data correlation and sharding balance, BCD code parallel conversion algorithm to achieve accuracy lossless and efficiency improvement through SIMD instructions and verification pipeline, and double entry bookkeeping parallel verification algorithm relying on incremental tagging and two-level verification to balance compliance and processing speed. Based on Spark and HDFS technology, a parallel processing platform was built, and experimental verification showed that the algorithm's shard balance remained stable at 0.90 or above, the BCD code conversion accuracy reached 100%, and the complex verification time was reduced by more than 70% compared to full processing; The batch processing efficiency of the platform is 3-5 times higher than that of traditional standalone machines, and the real-time verification response time is controlled within 2.3 seconds. The research results indicate that the proposed algorithm and platform can effectively address the adaptation gap of general parallel technology in accounting scenarios, providing technical support and practical solutions for the upgrading of enterprise accounting informatization.

This study constructs a collaborative innovation network using joint patent data from 100 R&D institutions in the electronics technology industry from 2021 to 2023, and employs Stochastic Actor-Oriented Models to analyze the network dynamics of the collaborative innovation network and the diffusion of technological innovations within the network over the three-year period with the example of 3D printing technology. The results show that there are apparent transitivity and institutional-type heterogeneity effects in the evolution of the collaborative innovation network. Institutions with strong R&D capabilities have a tendency to establish more connections in the evolution, and there is the phenomenon of strong alliances. In terms of the diffusion of technological innovation, the R&D atmosphere in the network has no significant effect on the diffusion of 3D printing technology. The R&D accumulation of R&D institutions has a significant positive effect on their application of new technologies for innovation.

Against the backdrop of the digital economy era, tremendous changes have taken place in the working methods, concepts, and requirements of enterprise management. Managers are not only required to possess necessary management knowledge and skills but also a certain level of digital literacy and the ability to apply digital technologies. This poses more stringent requirements for the cultivation of management talents in higher vocational colleges. In response, higher vocational colleges should improve the talent cultivation system, constantly update educational concepts, and innovate educational means to cultivate students' comprehensive management capabilities and enable them to better meet the needs of society and the industry.

This paper presents a computational research framework to systematically map and analyze the complex dynamics of High-Quality Agricultural Development (HQAD) in China, overcoming the limitations of traditional review methods.Our methodology integrates two core computer technologies: (1) Bibliometric Analysis, utilizing tools like CiteSpace to conduct co-word and cluster analysis on 2,871 academic publications from the China National Knowledge Infrastructure (CNKI) database (2021-2025), revealing the core research themes and their evolution. (2) Big Data Analytics, for which we propose and detail a workflow involving web crawling, text mining, and sentiment analysis to process unstructured data from policy documents, e-commerce platforms, and social media. This workflow is specifically designed to analyze emergent concepts like the "digital village" and "Agricultural New Quality Productive Forces" (ANQPF) with high granularity and in real-time. The results of our bibliometric analysis identify a significant shift towards technology-driven topics, while our proposed big data framework offers a robust and scalable methodology to empirically validate influencing factors and track policy impacts. By foregrounding these computer technologies, this paper provides not only a review of the HQAD field but also a replicable computational blueprint for future research, contributing to the optimization of data-driven policymaking for sustainable agriculture.