[Purpose/Significance] In the digital-intelligent era, artificial intelligence is becoming more and more integrated into children's learning, reading, creativity, and daily lives. Cultivating children's AI literacy has become an important issue related to educational transformation, digital inclusion, and the development of future-oriented public cultural services. Public libraries, as open and trusted social education institutions, have accumulated rich experience in reading promotion, information literacy education, digital literacy services, makerspace activities, and services for minors. However, their current AI literacy practices for children are still mostly exploratory and fragmented. Compared with school-based AI education and general digital literacy studies, insufficient attention has been paid to how public libraries can transform scattered activities into systematic, sustainable, and child-centered AI literacy services. This study therefore focuses on how public libraries can move from fragmented exploration to systematic practice in cultivating children's AI literacy within the Chinese context. [Method/Process] This study adopts a multi-case comparative analysis approach and follows the logic of "international paradigm comparison - domestic case analysis - localized pathway construction". First, it examines representative international practices and identifies two typical paradigms: the deep integration paradigm and the systematic popularization paradigm. The former emphasizes the integration of intelligent spaces and curriculum design, while the latter highlights age-differentiated services, community participation, and broad accessibility. Second, the study investigates seven public libraries in Beijing, Shanghai, Guangzhou, and Shenzhen. Based on publicly released activity information from March 2025 to March 2026, more than forty AI-literacy-related activities were collected and analyzed. These practices were classified into three types: cognitive enlightenment, skills practice, and ecological support. Third, drawing on the idea of the "library as education" and Piaget's theory of cognitive development, the study identifies problems with current practices and proposes a systematic framework consisting of practical pathways and supporting mechanisms. On this basis, the study further develops practical pathways and safeguards to ensure the sustainability and functionality of cultivating children's AI literacy in public libraries. [Results/Conclusions] The study found that, although they offer useful references, international practices still face common challenges. These challenges include the tension between service depth and universal accessibility, dependence on external resources, and the lack of systematic evaluation. Domestic public libraries have developed diverse explorations, such as AI-themed reading promotion, lectures, immersive experiences, programming courses, robotics activities, competitions, AI assistants, parent-oriented reading groups, and librarian training. Nevertheless, these practices remain "bonsai-like": they are vivid as individual projects but have not yet formed a sustainable educational ecosystem. In response, this study proposes three practical pathways. The spatial pathway aims to build a virtual-physical "intelligent interactive space" integrating reading, technology experience, creative expression, discussion, digital resources, and AI learning support. The content pathway proposes an age-differentiated and modular AI literacy resource system, progressing from AI perception and story-based enlightenment to operational understanding, ethical reflection, and creative problem-solving. The collaborative pathway constructs a library-school-family educational community, with public libraries serving as hubs connecting formal education, family learning, and public cultural services. To support implementation, the study further proposes three mechanisms: empowering librarians' AI literacy and pedagogical guidance capacity, developing age-differentiated evaluation tools, and establishing long-term driving mechanisms through diversified investment, policy support, professional standards, and social participation. Future research may include fieldwork, interviews, longitudinal tracking, and validation of the evaluation scale to test and refine this framework.

[Purpose/Significance] The development of artificial intelligence (AI) technology has brought both opportunities and challenges, redefining users' knowledge and capabilities. It is necessary to re-examine and build new capabilities suitable for the era of AI-AI literacy. As the main institution of information literacy education in higher education, university libraries have inherent advantages in carrying out AI literacy education. The research aims to explore the development path of AI literacy education in university libraries and assist them in better cultivating high-quality talent that meetS the needs of the AI era. [Method/Process] The practice of AI literacy education in university libraries is influenced by multiple interwoven factors and is essentially a complex, open and dynamic system. Therefore, introducing collaborative theory into this practical field provides guidance for the collaborative development of AI literacy education in university libraries in China. Through online research on the implementation of AI literacy education in libraries of 42 "Double First-Class" universities in China, a collaborative framework for AI literacy education in university libraries in China from the perspective of collaborative theory was constructed, starting from the educational subjects, educational objects, educational forms and educational contents, and the current collaborative status of each element was analyzed. [Results/Conclusions] The results show that the AI literacy education in domestic university libraries is characterized by the establishment of cooperative relationships by the main body, the emphasis on group mutual learning for the objects, the importance placed on platform complementarity in the form, and the focus on knowledge and skills in the content. Based on the collaborative framework and the research results, the following development paths are proposed: The first is the main body network layer, which requires the establishment of a diversified multi-party collaborative network, the creation of a specialized and composite team, and the improvement of a long-term collaborative management mechanism. The second is the object mutual assistance layer, which should be based on the two-way empowerment role positioning and the establishment of a two-way interactive education platform. The third is the form optimization layer, which should integrate online and offline educational carriers and design a classified and stratified curriculum system. The fourth is the content empowerment layer, which should develop a local framework for AI literacy theory, balance popularizing knowledge and deepening skills, and strengthen thinking cultivation and ethics education. Thus, AI literacy education will evolve from a scattered approach to a systematic one. It will shift from a single-subject focus to an integrated one, and from superficial to profound.

[Purpose/Significance] The deep integration of the latest technological revolution and industrial transformation has created an urgent demand for high-quality multimodal corpora for artificial intelligence-driven scientific discovery (AI4S) and large language models. Traditional coarse-grained knowledge organization methods based on documents have become insufficient for deep knowledge services. This study aims to construct a toolchain for extracting multimodal and multigranular knowledge units from scientific and technological literature, enabling the systematic mining of structured knowledge units from massive literature and enhancing the depth and efficiency of knowledge services. [Method/Process] This study conducted a systematic review of mainstream knowledge extraction tools, both domestical and international, and performed a comparative analysis and screening on dimensions such as technical principles, functional characteristics, application advantages, existing limitations, and processing efficiency. An application demand system was constructed from four levels: identification of research subjects, context tracing, content analysis, and evidence localization. Taking the field of rice breeding as an empirical scenario, a knowledge representation model for multimodal information was constructed based on the physical organizational logic of literature. Documents were divided into four major categories of 22 knowledge units: basic information subjects, structural support, material systems, and academic descriptions. The boundaries between knowledge units are clear, and there are abundant associative relationships. Integrating the extraction needs of various types of scientific and technological literature knowledge units with tool research results, a pipeline-style extraction process framework for multimodal and multigranular knowledge units has been designed. This framework implemented a pipeline-style processing framework for the entire process of document acquisition, physical structure analysis, logical structure reconstruction, multimodal content extraction, and knowledge unit fusion and storage, constructing a cascading processing toolchain from PDF original documents to semi-structured data, and then to structured knowledge. To address three major issues: insufficient accuracy of basic information, chaotic structure of academic statements, and missing information in supporting materials, GROBID domain-adaptive retraining, XML and Markdown fusion parsing, and DeepSeek large model hierarchical extraction instructions were optimized and integrated into a full-chain toolchain. [Results/Conclusions] Preliminary experiments on the toolchain have achieved good extraction of multimodal and multigranular data. In optimization experiments, overall micro-average F1 score of the header model increased by nearly 3 percentage points, significantly enhancing the model's balance and generalization ability when processing documents in diverse formats. The problems of chaotic distribution and weakened structure of academic statement information were successfully solved, achieving robust structured extraction of more than ten types of statement information such as acknowledgements, conflicts of interest, and data availability. The introduction of the large language model DeepSeek enabled deep mining and association of chart titles, formal citation sentences, and related discussion sentences in literature. The model achieved an F1 score greater than 0.99 for extracting chart titles and greater than 0.93 for recognizing formal citation sentences. Verification through the SciWatch platform demonstrates the extraction, presentation, knowledge association, and contextual coherence of charts, supporting deep literature understanding and cross-validation. The multimodal knowledge extraction toolchain for scientific literature constructed in this paper has been able to efficiently and accurately complete the automated extraction and structured application of various knowledge units in scientific literature, covering a complete toolchain, including preprocessing, multimodal information recognition, relation extraction, knowledge fusion, and storage. The research results provide a scalable solution and practical reference for the evolution of domain knowledge mining and knowledge service technology.

[Purpose/Significance] The governance of health medical data is fundamentally challenged by the "protection-sharing" paradox: the critical need to safeguard sensitive personal information often conflicts with the desire to utilize these data for public benefit. This issue is particularly pressing under China's "Healthy China" initiative, which promotes data sharing while the rapid expansion of medical APPs has led to increasing data misuse incidents. Existing research has extensively explored technological solutions such as blockchain, but a significant gap remains in understanding the dynamic, strategic interactions among the key stakeholders - government regulators, APP operators, and users - who operate with bounded rationality. This study addresses this gap by constructing a tripartite evolutionary game model. Its primary significance lies in dynamically modeling the co-evolution of strategies to identify critical leverage points, thereby providing a theoretical basis for designing effective collaborative governance mechanisms that can reconcile data protection with utilization and ensure the sustainable development of the health data ecosystem. [Method/Process] This study established a three-party evolutionary game model involving government regulators, medical-health APP operators, and users, based on the core assumption of bounded rationality. The model incorporated a comprehensive set of parameters, including direct benefits, various costs (compliance, regulatory), data risks, and network benefits under different regulatory scenarios. Replicator dynamic equations were derived for each party to mathematically describe the evolution of their strategy choices over time. The stability of the system's equilibrium points was rigorously analyzed using Lyapunov's first method to identify key stability thresholds. To validate the theoretical analysis and explore the dynamic evolutionary paths, numerical simulations were conducted using MATLAB. These simulations tested the impact and sensitivity of critical parameters - such as user-perceived data risk under operator self-discipline, user network benefits under dynamic regulation, government compliance rewards, and penalties for overdevelopment - from various initial strategy combinations. [Results/Conclusions] The analysis yielded several critical findings. First, users' authorization decisions are highly sensitive to the operational context, and they are significantly positively influenced by the perceived level of operator self-discipline and the observed intensity of government dynamic regulation. Enhancing user network benefits under effective regulation and reducing perceived data risks are paramount to encouraging authorization. Second, for APP operators, increasing government penalties for overdevelopment acts as a powerful deterrent, rapidly steering operators towards compliance. In contrast, government financial rewards for compliance, while effective, must be carefully balanced against their potential fiscal burden, which can slow the government's own stabilization into a dynamic regulatory role. Third, the system exhibits strong path dependence, capable of converging towards either an inefficient equilibrium (Non-Authorization, Overdevelopment, Passive Regulation) or the optimal Pareto state (Authorization, Self-discipline, Dynamic Regulation), depending heavily on initial conditions. The study concludes that resolving the paradox requires a multi-faceted strategy: advancing and ensuring robust anonymization technologies, implementing intelligent graded supervision that combines incentives and punishments, and firmly establishing institutional safeguards for user data sovereignty to build essential trust. A key limitation is the omission of data leakage risks from government data openness. Future work will integrate empirical data and consider user heterogeneity to refine the model.

[Purpose/Significance] Against the backdrop of the digital wave and the Healthy China initiative, efforts to enhance national health information literacy face challenges, including an insufficient supply of high-quality popular science content and low public enthusiasm for its dissemination. This study aims to explore the internal driving forces, core influencing factors, and transmission paths of the willingness to share online health popular science information. It further intends to provide theoretical support for regulatory authorities and popular science platforms in formulating incentive policies and safeguard mechanisms, thereby promoting the participation of social entities in popular science dissemination, increasing the supply of high-quality popular science resources, and enhancing the health information literacy of the general public. [Method/Process] A three-stage research design of "Grounded Theory - Fuzzy DEMATEL - ISM" was adopted. Firstly, interview data from diverse groups were collected through semi-structured interviews. Grounded Theory was then applied to coding to extract initial influencing factors and construct a multi-dimensional driving force system. Secondly, Fuzzy DEMATEL was used to calculate the centrality and causality degrees, so as to identify key factors. Finally, the interpretive structural modeling (ISM) method was employed to integrate the influencing factors, establish a hierarchical structure, and clarify the transmission logic and action mechanism. This method not only enables the acquisition of the most original influencing factor system from interview materials but also reveals the interaction relationships among these factors, which is in line with the research requirements and trends in the field of information science. [Results/Conclusions] The results of Grounded Theory analysis identified 13 influencing factors, which are categorized into four dimensions. The personal dimension includes four factors: interpersonal interaction traits, perceived utility, health information literacy, and self-efficacy. The information dimension consists of four factors: information quality, information source credibility, information richness, and information clarity. The platform dimension comprises two factors: interaction promotion mechanism and platform technology. The social dimension contains three factors: social economy, social public events, and the clustering effect. Fuzzy DEMATEL analysis indicated that perceived utility, health information literacy, information clarity, and social economy are the key factors. ISM analysis revealed a 4-layer hierarchical structure of influencing factors from the superficial to the deep, with the social economy being the deepest-layer factor. Additionally, four key transmission paths were sorted out. Based on the research conclusions, four suggestions are proposed: Firstly, from the personal dimension, efforts should be made to mobilize the subjective role of users. Secondly, from the information dimension, the information quality and clarity for content creators and sharers should be improved. Thirdly, from the platform dimension, active cooperation with content sharers should be pursued and the interaction mechanism should be optimized. Finally, from the social dimension, the government should promote the development of the health popular science industry. In subsequent studies, empirical tests (such as structural equation modeling and fsQCA) can be incorporated to ensure the reliability and validity of the theory.

[Purpose/Significance] In response to the urgent need for deep digital transformation and the release of data value in smart libraries, this study is based on the research perspective of integrated application scenarios and data collaborative governance. It targets commonly encountered in smart libraries, such as fragmented data governance, insufficient collaboration among entities, weak scene adaptation, and the disconnection between theory and practice. This study breaks through the limitation of existing research that only regards scenarios as external application conditions. Taking scenario demands as the core driving force and internal logic for constructing governance mechanisms, it systematically explores the internal mechanisms and implementation paths of application scenarios that drive data collaborative governance. It further improves the theoretical system of smart library data governance and fills the academic gap in the research on integrating scenario-driven and collaborative governance. It also provides innovative ideas and theoretical support for enhancing the efficiency of allocating data resources, strengthening the enabling effect of smart services, and supporting the high-quality development of public cultural services. [Method/Process] With scenario theory and collaborative governance theory as the core theoretical basis, and by comprehensively applying methods such as theoretical deduction, framework construction, case empirical research, and normative research, this study analyzed the connotation and operational characteristics of application scenarios that drive data collaborative governance. It scientifically classified smart library applications into three types: core basic, value-added innovative, and emergency response. It also constructed a governance mechanism with five interlinked and differentiated subjects, objects, platforms, technologies, and systems adapted to different scenarios. It selected the Jiaxing City Library as an example to empirically verify, extracting and forming operational, replicable, and promotional governance strategies and implementation paths. [Results/Conclusions] The research indicates that there is a significant dynamic coupling, bidirectional iteration, and closed-loop evolution relationship between application scenarios, data collaborative governance mechanisms, and the development vision of smart libraries. The targeted allocation of governance elements is driven by scenario demands, and the iterative upgrade of scenarios is driven by governance effectiveness feedback. Efficient implementation is achieved through scenario design optimization, element resource allocation, data integration applications, and effectiveness evaluation feedback. The research verifies the scientific basis and practical feasibility of the theoretical framework. It can alaso provide valuable insights for enhancing the efficiency of smart library data governance and maximizing data value. This study is limited because as it only uses single-case empirical research. Further research can be carried out in the future in areas such as multi-case comparisons, cross-regional library collaborations, the deep integration of digital and intelligent technologies, and long-term governance mechanisms.

[Purpose/Significance] Against the backdrop of intensifying global technological competition and the drive for scientific and technological progress under national innovation strategies, generative artificial intelligence (AI) technology, as an emerging disruptive technology, has had a profound impact on the economy and society through its widespread application. However, the diffusion of this technology in the market still faces numerous challenges. This paper aims to delve into the micro-level decision-making factors influencing enterprises' research and development (R&D) of generative AI technology, as well as the specific impact of user group interactions on the effectiveness of technology diffusion, by constructing a complex network evolutionary game model. The research seeks to uncover the inherent laws governing technology diffusion, providing a scientific basis for policymakers and corporate practitioners to promote the healthy development and effective diffusion of generative AI technology, thereby fostering comprehensive socio-economic progress. [Method/Process] This paper adopts the complex network evolutionary game model as the primary research method, integrating complex network theory, technological innovation diffusion theory, and social influence theory to construct a game model for corporate decision-making regarding generative AI technology. By incorporating the structural characteristics of complex networks and the dynamic mechanisms of evolutionary games, the study simulates the R&D decision-making processes of enterprises under varying conditions of user adoption rates, government subsidy levels, differences in technology benefits and costs, and technology spillover effects. Simultaneously, numerical simulation analysis is employed to explore the specific impacts of changes in these factors on the diffusion effectiveness of generative AI technology decisions, thereby thoroughly revealing the micro-mechanisms underlying technology diffusion. [Results/Conclusions] The research results indicate that an increase in user adoption rates significantly and positively drives the diffusion of generative AI technology, with moderate user dependency behaviors further accelerating this process. Government subsidies play a particularly prominent role in promoting technology diffusion when user adoption rates and the initial proportion of enterprises choosing R&D strategies in the network are low. However, as these proportions rise, the marginal effect of subsidies gradually diminishes. The difference in benefits between enterprises that develop generative AI technology and those that do not has a marked impact on technology diffusion, whereas the impact of cost differences is relatively minor. Furthermore, the spillover effects of generative AI technology may induce free-rider behaviors among other enterprises, hindering technology diffusion. Additionally, when the maturity level of generative AI technology is low, it reduces user trust in the technology, thereby inhibiting its widespread dissemination. Based on these conclusions, this paper proposes policy recommendations such as encouraging user participation, flexibly adjusting subsidy policies, enhancing technology maturity, and establishing intellectual property laws and regulations to facilitate the effective diffusion of generative AI technology.

[Purpose/Significance] As digital technologies such as 5G and generative AI become more prevalent in higher education, university libraries have evolved from traditional collections of books to ecosystems of cross-modal and multi-source resources, encompassing core collection resources, open-access resources, and user-generated content. However, the "resource silo" issue caused by heterogeneous resources and the mismatch between passive services and dynamic user scenarios in research and teaching remain unresolved. Existing studies lack integrated closed-loop mechanisms linking resources, scenarios, and users. This study aims to address these gaps by promoting libraries' transformation from "resource storage centers" to "proactive knowledge service centers." Its key innovation lies in constructing a scenario-driven three-dimensional collaborative model, which bridges the disconnect between resource integration and scenario adaptation, providing theoretical and practical support for intelligent library development. [Method/Process] Guided by ERG demand theory and context-aware computing, this study adopts a mixed-methods approach combining literature research, technical design, and case validation. A three-dimensional collaborative model of "Resource Integration - Scenario Adaptation - Smart Services" was proposed. For resource integration, a "three-dimensional integration + four-step fusion" framework was developed: standardized access via unified DCAT-AP/RDA metadata and multi-protocol gateways, associative reorganization through cross-modal semantic matching and knowledge graph aggregation, and hierarchical storage (hot/warm/cold tiers). The four-step fusion includes data preprocessing, modality conversion (ViT, Whisper-large, YOLOv8 models), feature fusion (attention mechanism + Transformer encoder), and knowledge generation (knowledge graphs, rule bases). An innovative five-dimensional dynamic scenario model (S=f(P,R,S,T,C)) quantifies user profiles, resource attributes, spatial locations, temporal contexts, and social connections for precise scenario identification. Technically, a "cloud-edge-device" architecture provides support, while a hierarchical service pathway (instant/in-depth/customized services) and a multi-dimensional evaluation system (resource/service/user dimensions) ensure closed-loop optimization. [Results/Conclusions] The model effectively achieves in-depth integration of multi-source cross-modal resources and precise scenario adaptation. Validated through typical applications - full-cycle research support and immersive teaching (VR ancient book restoration, MR anatomy demonstration) - it significantly enhances resource utilization efficiency and user experience, resolving the core pain point of resource-scenario disconnection. The model strongly supports libraries' transformation from passive resource supply to proactive knowledge services. Limitations include limited application of cross-modal technologies to virtual reality resources, insufficient coverage of management and social service scenarios, and the need for long-term validation of the evaluation system. Future research will deepen large-model-aided cross-modal fusion, expand scenario coverage, improve the evaluation system with third-party participation, and promote inter-university resource sharing to better support higher education development.

[Purpose/Significance] With the accelerated convergence of artificial intelligence and the metaverse, smart library information services are undergoing a profound transformation from tool-oriented functional optimization toward holistic cognitive support. Traditional information retrieval and service models increasingly struggle to explain and support complex cognitive activities involving multi-agent collaboration, contextual awareness, and continuous knowledge construction. From the perspective of human-machine-environment collaborative cognition, this study aims to explore the paradigm shift of smart library information services in intelligent digital environments and to establish an integrated theoretical framework that coordinates technological systems, cognitive processes, and contextual factors, thereby providing a systematic theoretical foundation for service model innovation and capability enhancement in smart libraries. [Method/Process] This study first reviews the evolutionary trajectory of information search paradigms - from symbolic computation and semantic understanding to social perception - through systematic literature analysis. We proposed Ecological Search as an emerging paradigm. Drawing on distributed cognition, embodied cognition, and information ecology theories, a human-machine-environment cognitive symbiosis search architecture was constructed, driven by a dual core of social multi-agent communities and contextualized metaverse environments. The architecture operates through an inner-outer dual-loop mechanism consisting of environmental perception and intention emergence, federated retrieval and knowledge fusion, collaborative generation and narrative construction, and cognitive evolution and ecological calibration. Furthermore, an "interaction-knowledge-context" three-dimensional analytical model was developed to decompose key service capabilities and derive differentiated integration pathways under diverse service objectives. [Results/Conclusions] The study proposed three smart library information service models: interaction-enhanced integration, knowledge-reconstructive integration, and context-immersive integration, and clarified how a unified cognitive architecture can be flexibly configured for different user groups and service scenarios. The findings indicate that the ecological search paradigm transcends system-centered instrumental rationality and reconceptualizes information search as a human-machine-environment collaborative process supporting continuous cognitive construction. By integrating multi-agent systems and contextualized environments, this paradigm provides essential mechanisms for smart libraries to move beyond information provision toward advanced cognitive support. The study offers theoretical insights and practical implications for achieving an ecological transformation of smart library information services while balancing technological innovation and human-centered values.

[Purpose/Significance] The rapid advancement of Generative Artificial Intelligence (GAI) is fundamentally reshaping the landscape of knowledge production and dissemination, propelling information literacy education into a critical phase of paradigm reconstruction. However, contemporary university-level information literacy programs are often constrained by structural impediments, including pedagogical homogenization that fails to address individual learner differences, fragmented multimodal resources that hinder holistic cognitive development, and the "suspension" of ethical education, where abstract moral principles are difficult to internalize into concrete practice. These challenges severely restrict the transition of information literacy education from static skill transmission to dynamic, value-oriented cultivation. Therefore, exploring a novel educational model deeply empowered by GAI is theoretically significant for reconstructing the theoretical framework of "Cognition-Competence-Value" synergy framework and is imperative for cultivating responsible digital citizens who can think critically and make ethical decisions in the intelligent era. [Method/Process] To address these challenges, this study synthesizes three core theoretical pillars: Multimodal Cognitive Construction, Human-Computer Collaborative Evolution, and Value-Sensitive Design. This model, called the Three-Dimensional Spiral Model (3DSM), is centered on "Cognition-Competence-Value." This model posits a dynamic, mutually reinforcing mechanism in which these three dimensions spiral upward through continuous interaction. To empirically validate the model's efficacy, a rigorous 8-week quasi-experiment was conducted at China University of Petroleum (East China). The study involved 120 participants who were randomly assigned to experimental and control groups. The experimental group participated in an intervention based on the 3DSM that utilized advanced GAI technologies, including an improved CLIP model for multimodal alignment, a dynamic knowledge graph for personalized path planning, and a "value sandbox" for ethical simulations. The teaching design followed a spiral curriculum, progressing from "Multimodal Information Deconstruction" to "Human-Computer Collaborative Innovation," and finally to "Ethical Internalization." In contrast, the control group followed a traditional "lecture plus practice" model. A mixed-methods approach was employed for the evaluation. This approach combined quantitative metrics, such as retrieval efficiency logs and Jaccard similarity coefficients for accuracy, with the CTIC standardized test, which measures information awareness, tool application, and ethical cognition. The evaluation also included a qualitative analysis of learning artifacts and behavioral trajectories. [Results/Conclusions] The empirical findings demonstrate that the 3DSM significantly enhances learners' comprehensive information literacy. Statistically, the experimental group exhibited a 53% improvement in information retrieval efficiency compared to the control group, with a retrieval accuracy (Jaccard similarity) increase from 0.68 to 0.89. Furthermore, the accuracy rate of technical ethical decision-making reached 89.2%, and the effect size was substantial (Cohen's d=1.37), indicating a large practical impact. Mechanism analysis revealed three key drivers of this success. First, the improved cross-modal alignment optimized cognitive efficiency by enabling accurate deconstruction of heterogeneous resources. Second, the dynamic knowledge graph facilitated capability evolution through personalized, adaptive learning paths. Third, the "Ethical Pre-regulation" mechanism, where ethical constraints are applied at the onset of cognitive tasks, effectively resolved the "ethical suspension" problem by calibrating cognitive paths and preventing algorithmic bias. This research contributes to the field by providing a systematic, theoretical framework for the synergistic development of cognition, competence, and value in the GAI era. It offers libraries and educational institutions a replicable, evidence-based implementation pathway for deeply integrating GAI into their curricula, thereby transforming information literacy education into a dynamic ecosystem of human-machine symbiosis and value co-creation.