[Purpose/Significance] Against the backdrop of the accelerated development of the digital economy, data elements have become a core production factor, driving the optimization of resource allocation and the transformation and upgrading of of industry in China. A series of national policy documents, including the "Data Element ×" Three-Year Action Plan (2024-2026) and the Digital Agriculture and Rural Development Plan (2019-2025), have established systematic progresses for allocating data elements to the market and integrating them deeply into agricultural production, operations, circulation, and services. With the prominent characteristics of scattered data sources, spatiotemporal heterogeneity and multi-stakeholder coupling, the agricultural sector serves as a typical scenario for observing the whole process of data value evolution. Existing studies mostly focus on single dimensions such as technical empowerment or market transaction mechanisms, and fail to fully reveal the dynamic process of data element value release driven by multi-subject collaboration in specific industrial contexts. This research explores the internal logic and realization process of releasing value of data elements in agricultural scenarios from the perspective of value co-creation, so as to provide theoretical support and practical references for allocating data elements in a market-oriented manner. [Method/Process] Twenty-eight typical cases from the agricultural track of the national "Data Element ×" competition were selected as empirical materials, all of which meet three core screening criteria: a complete business closed loop covering the whole data value chain, participation of two or more types of stakeholders, and quantifiable value release effects. Procedural grounded theory was adopted as the core research method, following the standard three-level coding procedure including open coding, axial coding and selective coding. Double independent coding by two researchers with relevant professional backgrounds was applied to ensure coding reliability, and the inter-coder consistency coefficient reached 0.87. The 28 cases were divided into three groups for initial framework construction, category iteration and theoretical saturation test, respectively, to guarantee the rigor and saturation of the theoretical model. [Results/Conclusions] Through systematic coding analysis, 90 initial concepts, 22 basic categories, 8 main categories and 3 core categories were extracted, and a three-layer "driving-supporting-pathway" theoretical model of data element value release was constructed. The findings show that the value co-creation actor network acts as the driving premise, which breaks the dilemma of "unwilling to share and difficult to share" data through multi-stakeholder collaboration and mutual trust mechanism. The data resource integration and governance mechanism serves as the key support, transforming scattered heterogeneous raw data into high-quality usable data resources through multi-source fusion, full-life-cycle governance and trusted environment construction. The data element value transformation mechanism is a process by which data evolve from a resource to a factor and then to a value through algorithm empowerment, scenario-driven service innovation, and data productization. This study broadens the scope of value co-creation theory in the field of data element research, and provides practical, replicable references for the digital transformation of agriculture. Due to the limited sample selection of excellent cases, the research has a certain survivorship bias. Follow-up studies can include cases with unsatisfactory value release effects and perform cross-industry verification to further improve the model's generalizability.

[Purpose/Significance] Against the backdrop of data becoming a key factor of production, the sharing and utilization of scientific data face significant challenges, including "market failure" and a fragmented policy landscape. Existing academic efforts often analyze policies from an isolated perspective. These efforts lack a holistic framework to understand how policies interact with multiple stakeholders to create value from data. This study aims to address this gap by constructing an integrated analytical framework for scientific data sharing policies. Its primary significance lies in providing a systematic tool to deconstruct policy architecture, dynamically reveal the internal transmission mechanism from policy intervention to value realization, and offer evidence-based insights for optimizing top-level design. This contributes to building a more efficient data governance ecosystem, ultimately enhancing the allocation efficiency of scientific resources and national innovation capacity. [Method/Process] The research employs a mixed-method approach combining theoretical construction and empirical text analysis. Firstly, through a synthesis of literature on policy instruments, stakeholder theory, and data factorisation, a three-dimensional analytical framework encompassing "Policy Instruments, Stakeholders, and Factorisation Stages" was constructed. To animate this static structure, the Stimulus-Organism-Response (SOR) model was introduced as an overarching theoretical lens, formulating a "policy stimulus (S) → stakeholder perception/organism (O) → factorisation response (R)" dynamic mechanism. Secondly, to empirically apply and validate the framework, representative policy documents, including the national "Measures for the Management of Scientific Data" and selected local implementation rules, were chosen as cases. Using qualitative data analysis software NVivo 12, 174 relevant policy clauses were extracted. A rigorous coding process based on the three-dimensional framework was conducted independently by two researchers to ensure reliability. The inter-coder consistency was measured with Cohen's Kappa coefficient, yielding a result of 0.82, which indicates almost perfect agreement. Discrepancies were resolved through discussion and expert consultation. Finally, statistical analysis was performed on the coded data to quantify the distribution of policy attention and identify characteristic patterns. [Results/Conclusions] The study yields three sets of core findings. First, it conceptualizes the factorisation of scientific data as a three-stage transition: "Digitization" (transforming raw information into structured data), "Valorization" (enhancing data into valuable assets through processing), and "Sharization" (releasing multiplied value through circulation and reuse). Second, the quantitative analysis reveals a distinct imbalance in current policy attention allocation. Regarding policy instruments, emphasis is heavily skewed towards "Planning & Organization" (35.63%) and "Sharing & Reuse" (21.84%), while the crucial intermediate stage of "Storage & Publication" is under-supported (10.34%). Concerning stakeholders, "Sharers" (e.g., researchers) are the central focus (43.10%), whereas "Intermediators" (e.g., data centers) are relatively marginalized (23.56%). In terms of factorisation goals, policies overwhelmingly prioritize the final "Sharization" stage (71.84%), overlooking the foundational "Digitization" and "Valorization" stages. Third, the research identifies several synergistic and effective policy pathways, such as "Mandatory Submission + Standard Constraints" and "Data Processing + Talent Incentives". Based on these conclusions, the study proposes that future policy optimization should focus on rebalancing attention towards intermediate processes and intermediary actors, strengthening whole-lifecycle governance, and enhancing the synergy of policy tools. Exploring innovative governance models like data trusts is also recommended to foster a sustainable data-sharing ecosystem. A main limitation of this study is its reliance on textual analysis; future research could employ surveys or interviews to empirically validate the SOR mechanism by measuring stakeholders' actual perceptions and behavioral responses, and test the framework's applicability in other specific data domains.

[Purpose/Significance] The use of short-video algorithmic recommendation services poses risks such as privacy breaches, information cocoons, cognitive manipulation, and degraded content quality. Existing studies mostly focus on risk types and consequences, lacking a systematic explanation of the generative mechanism of users' risk-coping behaviors, especially the internal logical relationships among behavioral factors. This study integrates the risk information seeking and processing (RISP) model and the protective action decision model (PADM) to deconstruct the generative mechanism of short-video users' risk-coping behaviors toward algorithmic recommendation services. The innovation lies in synthesizing cognitive, affective, and behavioral dimensions into a unified model, extending risk protection theories into the algorithmic context, and providing practical guidance for platform algorithm optimization and governance. [Method/Process] Using the critical incident technique (CIT), we collected data through semi-structured interviews and open-ended questionnaires from active short-video users, who had experienced at least one identifiable algorithm-related risk incident. Participants represented diverse demographic backgrounds, spanning different age groups, education levels, and usage frequencies to ensure broad representativeness. Thematic analysis was applied to code the data, examining behavioral elements and their interactions across four sequential stages: attention triggering, risk assessment, emotional response, and behavioral control. Cross-case comparisons were used to map interdependencies among these stages, leading to the development of a comprehensive generative mechanism model. [Results/Conclusions] The study reveals that risk-coping behaviors systematically unfold through four stages: attention triggering, risk assessment, emotional response, and behavioral control. These stages exhibit iterative relationships, with feedback loops also existing between emotions and risk reappraisal. Based on this framework, we propose corresponding management strategies for each stage: during attention triggering, algorithms should be optimized to align with users' cognitive capacities; during risk assessment, bidirectional feedback channels should be enhanced to encourage deeper user engagement; during emotional response, risk perception biases should be mitigated while harnessing the positive potential of emotional responses; and during behavioral control, users' algorithmic rights must be safeguarded, balancing technological power with user responsibility. This study offers an integrative explanation of the coupled dynamics among cognitive, emotional, and behavioral factors within algorithmic risk contexts. Limitations include inherent recall bias in qualitative CIT designs and limited generalizability. Future research should employ quantitative methods to validate the model empirically and test its causal pathways. This will strengthen the model's theoretical and practical foundations for algorithmic governance.

[Purpose/Significance] The new round of technological revolutions and industrial transformations are deeply integrated, and digital technologies, especially those represented by generative artificial intelligence (AI), are advancing rapidly. While empowering high-quality economic and social development, they have also triggered a series of prominent issues related to technological ethics and digital security. Digital security has become an unavoidable core issue in the global digitalization process. In comparison, the "amplification" effect brought by China's over 1.1 billion Internet users has made the impact of digital security more extensive and profound, making it a core element related to national security, economic and social development, and citizens' legitimate rights and interests. The threats to digital security in the era of AI are more complex and diverse. To effectively guide and support residents in facing the challenges of the AI era, it is necessary to grasp the current status of residents' digital security level and the digital divide between different groups. Compared with existing research, this study mainly supplements the research gap in the empirical investigation of residents' perception level of digital security (including AI security). Based on this, conducting a survey of residents' perceptions of digital security and proposing corresponding countermeasures would help us understand the current status of digital security among residents, bridge the new digital divide, and promote classified digital security education measures. It will be important for further strengthening the digital security barrier, safeguarding personal information security, ensuring social stability, and maintaining national security. [Method/Process] Focusing on the research objectives and core themes, this study refers to the "Global Digital Literacy Framework" developed by UNESCO, draws on existing domestic research findings on digital literacy evaluation, integrates its digital security assessment indicators and designs, and combines the characteristics of new digital security risks in the AI era to design a localized digital security awareness survey questionnaire. Through online questionnaire surveys and variance testing methods, a case study was conducted among a total of 2 000 residents from the eastern, central, and western regions of China. The research data cover the demographic attributes of the respondents, basic digital security capabilities, digital device protection, and privacy protection indicators, while also incorporating multi-dimensional assessment indicators for digital health protection and AI security, to empirically assess the level of residents' digital security and the current status of digital divide among different groups. [Results/Conclusions] The research results indicate that residents' perception of digital security is not high overall. The basic digital security capability dimension performed relatively well. The digital health protection and artificial intelligence security dimensions performed better than the digital device protection and privacy protection dimensions, which scored the lowest. Residents recognize the importance of digital security in the era of AI and possess a certain level of digital security knowledge, but there is still considerable room for improvement in practical application of digital devices and personal privacy protection capabilities. Influenced by factors such as age, registered residence, education level, occupation, income, and device usage duration, there are significant differences in residents' digital security perception levels, especially among rural residents and the elderly, who are in urgent need of improvement. Based on the research results, the following suggestions are proposed. We should improve the digital literacy education system to ensure lifelong learning of residents' digital security capabilities. Efforts should be paid attention to the digital security rights and interests of the elderly population and enhance digital security inclusiveness. We should promote extensive collaboration among diverse social entities to enhance the digital security perception level of rural residents, and further promote monthly improvement activities to build a solid digital security barrier. This study focuses on empirical investigation and analysis in the field of residents' digital security, and proposes targeted coping strategies based on the research results, demonstrating a certain degree of innovation. However, this study has certain limitations due to its reliance on literature analysis and online surveys, as well as its relatively limited sample selection. Future studies could more comprehensively and deeply explore residents' digital security capabilities by expanding the survey sample, combining offline field research, or adopting a multidimensional stratified sampling framework. This would provide more universally applicable practical references for enhancing residents' digital security capabilities and further bridging the digital divide.

[Purpose/Significance] Against the global surge of generative artificial intelligence (GenAI) and large language models (LLMs), academic libraries are undergoing a critical paradigm shift in their reference services. While "AI Virtual Librarians" (AIVL) are increasingly adopted to enhance efficiency, cross-national evidence regarding how they are configured alongside traditional "Human Live Reference" (HLR) remains scarce. This study aims to reveal the structural differences in human-AI configurations between Chinese and international top-tier university libraries. It seeks to identify the divergence between "technology-driven" and "human-centric" service models and proposes a governance-oriented hybrid pathway to inform the digital transformation of academic libraries. [Method/Process] The study established two high-resource samples: 42 libraries from China's "Double First-Class" universities and 94 libraries from the U.S. News Top 100 World Universities. A systematic website investigation and standardized interaction tests were conducted to collect data on service availability and deployment models. The study not only quantified the deployment of HLR and AIVL (classified into rule-based and LLM-based) but also qualitatively evaluated the "Core Service Contents" and "Linkage Mechanisms" (e.g., traceability, boundaries, and human fallback). Chi-square tests were employed for statistical analysis, and robustness checks were performed using both broad and strict counting rules to ensure validity. [Results/Conclusions] Results indicate that while the overall service coverage is similar across groups (approx. 74%), the service structure diverges significantly. International libraries predominantly rely on the "Human-only" mode (66.0%), prioritizing deep research support, academic integrity, and privacy protection. In contrast, Chinese libraries show a significantly higher adoption of AIVL (57.1% vs. 8.5%) and LLMs (26.2% vs. 1.1%), with 52.4% operating in an "AI-only" mode. Content analysis reveals that Chinese AIVLs focus on transactional efficiency and 24/7 accessibility, whereas international counterparts focus on distinct research guides and governance. The study identifies a critical trade-off: China's aggressive AI adoption enhances accessibility but faces challenges regarding answer hallucinations and the lack of human fallback mechanisms. To address these challenges, the paper recommends a "Human-AI Collaborative Loop" model. Key strategies include: 1) Implementing risk-tiered routing, where low-risk transactional queries are handled by AI and high-risk research inquiries are directed to humans; 2) Optimizing AI reliability through Retrieval-Augmented Generation (RAG) and controlled knowledge bases to ensure traceability; 3) Establishing clear governance boundaries and stratified implementation paths for libraries with different resource levels, ensuring a balance between technological innovation and service ethics.

[Purpose/Significance] Generative artificial intelligence, big data, knowledge graphs, and other smart digital technologies are being rapidly integrated into university library resource development, subject services, information literacy education, and smart consultation, shifting service models from a purely resource-supply approach to one focused on knowledge discovery, data support, and intelligent collaboration. Librarian skill development is also moving from general knowledge updates to continuous capability building centered on job tasks, service scenarios, technology applications, and organizational collaboration. In reality, some university libraries still face issues such as training content being disconnected from positions, a gap between technology learning and service application, unclear job roles, insufficient cross-department collaboration, and underutilization of communities. Based on this, this study examines how organizational support systems-through rules, tools, communities, and coordinated division of labor-promote the development of librarians' smart digital skills and their application in services. The research topic stems from the practical tension between "rapid technology iteration" and "relatively lagging librarian skill updates" during digital transformation. Unlike previous studies that focus on individual learning motivation, skill components, or training models, this paper places librarian learning within the organizational activity system, revealing the interactions between job tasks, technical tools, institutional rules, learning communities, and project division. It proposes a skill-building path of "job task guidance-community collaboration-tool adaptation-rule feedback," aiming to shift related research from static description to dynamic mechanism analysis, and provides a basis for university libraries to respond to technological substitution, role redefinition, and service upgrading. [Method/Process] This study uses activity theory as the analytical framework and regards librarians as activity subjects, intelligent platforms, digital resources, and training tools as mediating tools, training policies, job norms, and performance evaluations as rules, and librarian teams, users, teaching staff, and technical departments as communities. It also uses job division to analyze the process of turning learning outcomes into service practice. Using purposive sampling, three different types of university libraries were selected, and semi-structured interviews were conducted with 18 librarians responsible for resource development, subject services, information literacy education, technical support, and overall management. Organizational documents such as training policies, project records, and job descriptions were also collected for triangulation. Data analysis followed familiarization, open coding, theme generation, theme review, and theoretical integration. This method is suitable for revealing interactive mechanisms and action logic within organizational contexts. This study is exploratory and qualitative in nature, without performing related statistical analysis, significance testing, or general causal inference. [Results/Conclusions] The study found that librarians' digital intelligence is not just about personal willingness or one-time training; it is shaped by job tasks, organizational rules, digital tools, learning communities, and project division together. Job tasks can enhance the sense of learning goals; adapting to rules and tools can boost learning motivation and efficiency; communities help calibrate learning direction through feedback, resource sharing, and peer evaluation; proper division of labor helps leverage the complementary strengths of technical and service-oriented librarians, easing the mismatch between tech supply and service demand; conflicts between old rules and new tasks drive institutional adjustments and updates to the skills system. It is recommended to set up tiered training aligned with job tasks, build cross-departmental learning communities, improve trial and feedback mechanisms for digital tools, and include knowledge sharing, project outcomes, and service innovation in performance evaluations. Limitations of the study include a restricted sample and regional coverage, mainly relying on interviews and organizational documents. In the future, the sample could be expanded, combining surveys, behavioral data, and longitudinal cases to further explore AI ethics, librarian professional identity, human-machine collaboration boundaries, and digital intelligence assessment standards.

[Purpose/Significance] Against the backdrop of China's aging population and advancing digitalization, the digital divide has created disparities in how efficiently older adults can access health information. This has emerged as a critical barrier to their health management. Existing studies have explored factors influencing older adults' health management but rarely focused on the digital divide within this group or deeply integrated individual traits, environmental factors, and health behaviors. Grounded in the Triadic Reciprocal Determinism Theory, this study investigates how exposure to online health information affects older adults' health management behaviors and uncovers how the digital divide manifests in this process. Its innovation lies in addressing gaps in previous research by systematically examining the heterogeneous impacts of online health information exposure on health management across different subgroups (e.g., urban vs. rural residents, those with varying digital skills). This enriches the theoretical system of older adults' health information behavior and health management, while offering practical guidance for tackling the key issue of enhancing older adults' health management and bridging the digital divide amid aging and digitalization. [Method/Process] Guided by the Triadic Reciprocal Determinism Theory, the study formulates hypotheses through literature review. It utilizes cross-sectional data from the 2020 Chinese Longitudinal Aging Social Survey (CLASS), with a valid sample of 3 118 adults aged 60 and above. Health management is influenced by older adults' attitudes toward handling minor illnesses and the frequency of their physical exercise. Online health information exposure is measured by three indicators: the convenience of accessing online health services, engagement in online health management, and ownership of smartwatches or bracelets. Control variables include gender, age, education, and household registration type. Multiple regression models are constructed to test hypotheses, supplemented by robustness tests (adding control variables and replacing dependent variables), heterogeneity analysis (stratified by smartphone proficiency), and mediating effect analysis (comparing urban and rural groups) to ensure result reliability. [Results/Conclusions] Key findings reveal that the convenience of accessing health services via the Internet and engagement in online health management both positively boost older adults' initiative to seek medical care when dealing with minor illnesses. Smart Wearable devices exert a "double-edged sword" effect: they have a negative impact on medical-seeking behavior for minor illnesses yet correlate positively with more frequent physical exercise. Older adults proficient in smartphone use show greater activity in health management. Significant urban-rural differences exist in reliance on social network information-rural older adults are more likely to enhance their health management initiative through family and friend networks, while their urban counterparts tend to obtain health information directly via the Internet. It is recommended that health information communication strategies be formulated in consideration of urban-rural differences, with targeted measures to improve older adults' digital health literacy and bridge the digital divide. Limitations of the study include data constraints, as CLASS data lacks in-depth exploration of older adults' motivations and resistance toward health information exposure, and incomplete measurement of the digital divide, excluding variables such as network speed and regional infrastructure. Future research should adopt mixed methods, integrating questionnaires and semi-structured interviews to supplement qualitative data, expand the measurement dimensions of health management behaviors, and explore the dynamic relationship between technology adoption and health behaviors from the perspective of older adults, providing a more solid basis for targeted interventions.

[Purpose/Significance] The rapid advancement of digital and artificial intelligence (AI) technology is profoundly transforming the functional positioning and service models of university libraries, shifting them from traditional resource management institutions to smart academic service hubs. This study aims to analyze the opportunities and challenges posed by digital and AI technology to university libraries, exploring how they can uphold their educational mission in the transition toward digitalization, ecological sustainability, and academic excellence, achieving the unity of technological empowerment and value preservation to promote sustainable library development. [Method/Process] The research followed the logical framework of "technological empowerment-risk assessment-value orientation," and conducted progressive and dialectical analysis. In the dimension of technological empowerment, it focused on three core transformation directions - resources, space, and platforms - to explore the ecological restructuring pathways of university libraries driven by digital and AI technologies. In the dimension of risk impact, it analyzed potential risks and value deviations arising from the application of digital and AI technologies across four aspects: reading cognition, service essence, resource development, and ethical privacy. In the dimension of value adherence, based on the core mission of university libraries, it proposed five value adherence dimensions: "people-centered, content-based, education-oriented, fairness-guided, and staff-focused". [Results/Conclusions] Digital technology has given university libraries new impetus in terms of resource integration, spatial reconstruction, and service upgrades. However, it has also brought risks and challenges, such as shallow reading, the instrumentalization of services, the homogenization of resources, and ethical and privacy issues. In the process of digitization, there is a close internal logic between technological empowerment, risk challenges, and value preservation in university libraries. Technological empowerment is the driving force for transformation, providing tools and paths for transformation, but its application requires value preservation as a prerequisite. Risk challenges are inevitable accompanying problems in the process of technological application, and they are a concrete manifestation of the contradiction between technological empowerment and value adherence. They need to rely on value adherence to guide technological direction and achieve dynamic balance. Value adherence is the key to maintaining the essence of a library, providing guidance for both technological empowerment and risk challenges, and ensuring that the library adheres to its original intention of "serving education and knowledge dissemination". University libraries need to find a balance among the three, using technology as a means, value as a guide, talent as support, and fairness as the bottom line, to construct an academic service model of "technology ecology value" collaborative development. This will enhance service efficiency and innovation potential while demonstrating the value of libraries in the high-quality development of higher education and the cultivation of talented professionals.

[Purpose/Significance] In the context of the "Healthy China 2030" strategy and the rapid advancement of medical technology, the innovative capacity of medical postgraduates is pivotal in driving clinical breakthroughs and translating medical research into practice. This study focuses on medical postgraduates, exploring the complex causal relationships between various dimensions of information literacy and innovation ability from an information science perspective. The goal is to provide empirical evidence that can enhance the innovation capacity of medical postgraduates. [Method/Process] A mixed-methods research design was employed. First, a questionnaire survey was conducted with 368 medical postgraduates from a university in Hainan, China, measuring four dimensions of information literacy (information awareness, knowledge, capability, and ethics) and innovation ability. Fuzzy-set qualitative comparative analysis (fsQCA) was then used to identify the pathways through which combinations of these dimensions lead to high or low levels of innovation ability. To supplement and validate the quantitative findings, semi-structured interviews were conducted with 20 postgraduates, representing varying levels of innovation ability, to provide deeper insights. [Results/Conclusions] The fsQCA revealed a complex causal structure, identifying four configurations leading to high innovation ability (overall consistency: 0.835; coverage 0.702) and three configurations associated with low innovation ability (overall consistency: 0.913; coverage 0.658). The results show that no single dimension is a necessary condition for high innovation. Instead, multiple equally effective pathways exist. Information capability emerged as a core condition in three of the four high-level configurations, with a necessity consistency of 0.802, highlighting its foundational role as an "approximate necessary condition." The configurations also revealed significant synergistic effects. For instance, one pathway (Configuration 1: high information awareness and capability compensating for low knowledge) demonstrates that strong awareness and practical skills can offset gaps in theoretical knowledge, often facilitated by AI tools. Interview data reinforced these findings: high-innovation postgraduates emphasized the importance of information capability in efficiently synthesizing evidence, while those with low innovation identified weak information awareness (e.g., insensitivity to research frontiers) and limited information capability as primary barriers. The study also identified three distinct pathways to low innovation, characterized by the absence of key dimensions, such as awareness, knowledge, and capability, occasionally compounded by ethical lapses. The study concludes that fostering innovation among medical postgraduates requires shifting from a one-size-fits-all approach to a configuration-oriented support system. By combining quantitative pathways with qualitative insights, universities can develop tailored, multi-layered information literacy programs. This study is limited by its single-institution sample and cross-sectional design. The identified pathways may be context-specific and do not capture the dynamic evolution of information literacy configurations over time. Future research should expand to multi-center studies across diverse institutional contexts. We will use longitudinal designs to examine how configurations change over time. Additionally, we will explore how factors such as supervisory style and resource availability influence these pathways.

[Purpose/Significance] Artificial intelligence is profoundly transforming library services worldwide, making it essential to understand the factors influencing librarians' AI adoption intentions for promoting smart library development. Unlike previous studies that primarily focused on general technology adoption in commercial settings, this research specifically targets academic librarians in Chinese universities who face unique professional challenges and institutional constraints in the digital transformation era. This study aims to identify key determinants of AI adoption intention among academic librarians, providing theoretical foundations and practical guidance for optimizing library AI service systems. The research contributes to the existing literature by constructing an integrated framework that combines individual-level perceptions with organizational and environmental factors to explain technology adoption behavior in the academic library context. [Method/Process] Drawing upon the Technology Acceptance Model (TAM) and Technology-Organization-Environment (TOE) framework, this study constructs an integrated theoretical model encompassing three dimensions: individual perceptions (perceived usefulness and perceived ease of use), organizational factors (organizational readiness and management support), and environmental factors (external environment). Six research hypotheses were proposed based on the theoretical framework. A questionnaire survey was conducted among academic librarians from various types of higher education institutions across 12 provinces in China during November 2025. A total of 177 valid responses were collected from research universities, teaching-oriented universities, and vocational colleges, with an effective response rate of 84.7%. The measurement items were adapted from validated scales and underwent rigorous translation and back-translation procedures. Partial Least Squares Structural Equation Modeling (PLS-SEM) was employed for measurement model assessment, structural model evaluation, hypothesis testing, and mediation effect analysis. [Results/Conclusions] The findings reveal that perceived usefulness exerts the strongest influence on AI adoption intention (β=0.447, p<0.001), followed by the external environment (β=0.354, p<0.001) and perceived ease of use (β=0.234, p<0.001). Contrary to theoretical expectations, organizational readiness (β=-0.099, p=0.075) and management support (β=0.034, p=0.593) showed no significant effects on adoption intention. The mediation analysis demonstrates that perceived ease of use influences adoption intention both directly (β=0.234) and indirectly through perceived usefulness (β=0.205, t=5.450, p<0.001), with the indirect effect accounting for 46.7% of the total effect. The integrated model explains 62.6% of variance in adoption intention, demonstrating substantial explanatory power. These findings suggest that during the early stage of AI technology adoption in Chinese academic libraries, individual-level perceptions and external competitive pressures predominate over organizational factors. This pattern may be attributed to the fact that librarians currently access AI technologies primarily through personal exploration using publicly available tools rather than through organizational deployment. The study provides differentiated practical recommendations for universities at different development stages and suggests future research directions including longitudinal designs and cross-cultural comparisons.