[Purpose/Significance] Currently, large language models (LLMs) and agents have emerged as core technical paradigms in artificial intelligence, with their integration into scientific research scenarios holding profound significance for transforming research paradigms. Traditional scientific research is facing an increasing number of challenges such as inefficient literature searches, the processing of massive amounts of data, repetitive experimental tasks, and barriers to collaborative innovation. Agents, empowered by LLMs, offer a promising solution to these bottlenecks by enabling intelligent automation and adaptive collaboration across research workflows. Beyond basic task assistance, they play a pivotal role in facilitating knowledge fusion, accelerating breakthroughs in frontier areas, and reshaping traditional research models. This study aims to clarify the core techniques and applications of agents in scientific research, highlighting their transition from auxiliary tools to integral innovation partners, which is crucial for accelerating knowledge discovery, enhancing research efficiency, and promoting the shift toward intelligent and collaborative research models. [Method/Process] Employing an objective, inductive approach, this study thoroughly explains the core technical modules of agents including planning, perception, action, and memory, as well as the operational mechanisms of multi-agent collaboration. It also integrates an analysis of agent applications throughout the entire scientific research lifecycle. This analysis covers key scenarios including literature review and idea formulation, experimental planning and design, data processing and execution, result analysis and knowledge discovery, and research report composition. By analyzing the application value and existing limitations of agents, this study proposes prospects and recommendations for the application and development of agents in scientific research scenarios. [Results/Conclusions The findings reveal that LLM-driven agents are evolving from basic task executors to active participants in scientific discovery, demonstrating significant transformative potential throughout the entire research workflow. They facilitate more efficient information processing, smarter experimental design, and deeper knowledge integration, thereby redefining traditional research patterns. However, several challenges persist, including limitations in long-range reasoning capabilities, and underdeveloped ecosystem support. There are also ethical and security concerns, such as data privacy and academic integrity. To address these, future efforts should focus on strengthening intelligent computing infrastructure for scientific data, deepening collaborative development of domain-specific agents, establishing a unified open collaboration framework with standardized interfaces, and building "human-in-the-loop" hybrid systems and multiple evaluation mechanisms. These measures will enable agents to become core partners in scientific innovation, driving the transition of research paradigms toward greater intelligence and collaboration.

[Purpose/Significance] In the era of cloud computing, the Library Services Platform (LSP) failed to become a unified solution for libraries it promised to be. Now, it faces new development bottlenecks in the era of smart libraries. Its relatively rigid architecture, isolated data models, and limited intelligence level make it difficult to meet modern users' urgent demands for access to new resource ecosystems and proactive services. This limitation stems from the fact that existing LSPs are rooted in a resource management design philosophy. They lack native support for intelligence, personalization, and ecosystem integration, which hinders their ability to serve as a core component in the construction of smart libraries. [Method/Process] The rapid development of large language model (LLM) technology is promoting libraries to transition from digital intelligent phases into a new era of intelligent services. As AI agents are increasingly emerge as a core strategy for LLM applications, this paper proposes a next-generation LSP architecture called A-LSP, which is agent-oriented. The core of A-LSP consists of a three-layer logical model. 1) Layer 1: Compatibility & Tools - MCP Marketplace, serving as the foundation of the platform, this layer bridges the agent ecosystem with the external world. It transforms existing heterogeneous library systems (including legacy LSPs) and external tools into invocable "capability units" for agents through standardized protocols. 2) Layer 2: Orchestration & Intelligence－Agent Middleware. Functioning as the platform's "operating system" and "brain," this layer handles agent lifecycle management, task planning and decomposition, state and memory maintenance, and most crucially, the coordination of multi-agent collaboration. 3) Layer 3: Application & Ecosystem - Agent Marketplace. This functional layer serves users and developers, where various reusable agents encapsulating specific business logic are published, discovered, combined, and invoked, creating a rich application ecosystem. This architecture enables the implementation of new platform strategies without replacing legacy systems, establishing a modern technological platform with endogenous intelligence, inclusive compatibility, and an open ecosystem. This agent-based library service platform can be seen as a significant upgrade to existing LSPs, it drives their transformation from resource management-centric to agent service-centric, establishing itself as the library service platform for the AI era. [Results/Conclusions] Moreover, this paper puts forward a "Five Centers" construction demand framework for future libraries, namely, the Smart Resource Center, Smart Service Center, Smart Learning Center, Smart Scholarly Communication Center, and Smart Cultural Heritage Center, to build a blueprint for the integration of library technology and business. For each center, it delineates a representative complex application scenario and analyzes the underlying multi-agent collaboration processes, thereby clearly demonstrating A-LSP's deep integration with each center's operational logic and illuminating its profound impact on future library service models.

[Purpose/Significance] Through interdisciplinary comparison, the core connotation, common core and field differentiation of the Agent concept are revealed, the Agent-related concepts and theories contained in library science are revealed, and the innovative value of AI Agent driven by large language models to the core services of libraries is analyzed to promote the transformation of knowledge services to intelligent and collaborative paradigms. Understanding the interdisciplinary nature of Agents will help library science, information science and other related disciplines to better design and apply AI technologies and achieve the core mission of connecting humans and knowledge in a more efficient, accurate and humane way. It will also enable library science to more accurately integrate the essence of the six major disciplines and transform the traditional three-subject relationship of readers, librarians and systems into a new collaborative paradigm. [Method/Process] Using interdisciplinary literature analysis, the definition, theoretical evolution and application characteristics of Agent in six major disciplines of philosophy, economics, law, biology, sociology and computer science are sorted out, the concepts and theories related to Agent contained in library science are explored, and the commonalities and differentiation of Agent in the five-dimensional characteristics of autonomy, perception, purpose, adaptability and interactivity of each discipline are compared. The theoretical essence of the six major disciplines is mapped to the three-dimensional subject in the practical field of library science, and the Agent role coordination mechanism of readers, librarians and systems is analyzed. Readers are entities with intentions and autonomous consciousness, and they will actively initiate information search behaviors based on information needs such as learning knowledge and solving problems and will also adjust their strategies according to environmental changes such as technical tools and social culture, reflecting agent-like adaptability. Librarians serve as service intermediaries and information gatekeepers, connecting readers with resources through technical services such as classification and cataloging, and helping users clarify their information needs and improve their information literacy through reader services such as reference consultation. The library's information systems will also simulate human agent capabilities through algorithms or technologies. Automated search engines or crawlers will collect data according to preset rules, and personalized information recommendations will be made based on user behavior, driving the library's management and services towards automation and intelligence. [Results/Conclusions] The commonality of interdisciplinary agents revolves around the realization of goals by autonomous actors in the environment. The five-dimensional characteristics constitute an interdisciplinary consensus, and the differences are due to the core issues of the disciplines. The essence of a library is a multi-agent system. The reader agent integrates philosophical intentionality and economic game strategy, reflecting demand-driven adaptive retrieval. The librarian agent inherits the legal agency rights and responsibilities and the sociological structural initiative, becoming a professional intermediary between resources and users. The system agent draws on the biological evolutionary adaptation and computer perception closed loop, and advances to an intelligent base for autonomous optimization. AI Agent is a technical enhancement of the inherent agent characteristics of library science. It realizes automation, personalization and intelligent service upgrades through large language models, realizes intention understanding, tool calling, and multi-agent collaboration, and drives the three-element subject from passive response to active collaboration. The three-element agent framework for library science is created, which clarifies the collaborative agent roles of readers, librarians, and systems, and reveals the deep logic of AI Agent driven by large language models and library science. An interdisciplinary comparative study of the Agent concepts reveals that its essence is a practical vehicle for achieving autonomous decision-making in a specific environment. Philosophy gives it depth of consciousness, economics models strategic games, law defines the boundaries of rights and responsibilities, biology reveals evolutionary logic, sociology anchors structural interactions, and computer science ultimately achieves a closed-loop technology. Library science constructs a ternary collaborative intelligent ecosystem that transforms abstract autonomy into a practical paradigm of knowledge connection through the dynamic collaboration of readers, librarians, and systems.

[Purpose/Significance] This paper investigates how generative artificial intelligence (GenAI) is reshaping the Searching as Learning (SAL) paradigm, focusing on its implications, challenges, and prospects in Library and Information Science (LIS). Traditional SAL emphasizes the cognitive and metacognitive processes by which users acquire and construct knowledge through information retrieval. However, the advent of GenAI - especially large language models (LLMs) - introduces a transformative shift from keyword-based querying to dynamic, dialogic, and multimodal interactions. This study aims to clarify the conceptual and practical significance of GenAI-driven SAL, explore its technical trajectories, and evaluate its impact on learners' behavior, learning strategies, and information literacy. It also highlights the emerging ethical and epistemological challenges posed by GenAI systems in learning-oriented search contexts. [Method/Process] Using the PRISMA-ScR framework, this study conducted a scoping review of academic and gray literature published between January 2023 and May 2025. A total of 1 681 records were retrieved from major academic databases and preprint repositories. After screening titles, abstracts, and full texts, 22 studies were selected for in-depth qualitative analysis. Thematic coding and synthesis were conducted to extract recurring patterns and theoretical insights across three key dimensions: GenAI-enhanced search technologies, evolving user behaviors in SAL contexts, and normative concerns associated with credibility, agency, and transparency. The analysis was grounded in LIS theories, including information behavior, metacognitive models of learning, and digital/information literacy frameworks. [Results/Conclusions] The results reveal that GenAI is fundamentally reshaping SAL in three key areas. First, in terms of technology, GenAI systems (e.g., GPT-based chat interfaces) provide conversational, context-aware, and multimodal assistance, transforming SAL from reactive searching to proactive co-learning. These systems scaffold learning through adaptive query reformulation, real-time content summarization, and source triangulations supporting iterative reflection and cognitive engagement. Such affordances mirror the functions traditionally associated with human tutors, thereby expanding learners' capacity for critical inquiry and self-directed exploration. Second, user behaviors in SAL are undergoing a paradigm shift. Learners increasingly engage in human-AI co-construction of knowledge, participating in iterative query-dialogue loops that facilitate concept clarification and knowledge synthesis. While this enhances engagement, personalization, and perceived learning efficiency, it also raises concerns. Over-reliance on AI-generated content may undermine learners' critical thinking, reduce information discernment, and promote passive consumption. The study identifies a dual effect. While GenAI augments higher-order thinking and strategic learning, it can also lead to superficial comprehension when learners lack the skills to critically evaluate AI output. Third, the review underscores the urgency of addressing ethical and pedagogical challenges. Issues such as AI hallucination, algorithmic opacity, and biased content threaten the credibility of GenAI-enhanced learning environments. From an LIS perspective, this necessitates a reconfiguration of information literacy education to include AI literacy. Students must be equipped not only to retrieve and evaluate information, but also to interrogate algorithmic sources, verify provenance, and triangulate AI outputs with authoritative references. GenAI should be positioned as a cognitive assistant, not a definitive knowledge authority. GenAI holds substantial promise in enhancing SAL through greater interactivity, personalization, and cognitive scaffolding. However, these benefits must be balanced with informed practices that mitigate risks to learner autonomy, critical reasoning, and information ethics. This work establishes an analytical foundation for future research and practices at the intersection of AI, learning, and information behavior.

[Purpose/Significance] As generative artificial intelligence (AI) transforms library services, existing evaluation systems fail to capture dynamic characteristics of AI-driven resource discovery. This study develops a dynamic evaluation framework for public libraries' AI-enhanced services, addressing the gap between technological innovation and service assessment. [Method/Process] The research employed a mixed-methods approach to develop and verify a multi-dimensional evaluation framework based on Knowledge Organization Systems (KOS) theory. The framework comprises five primary dimensions: physical environment, technical architecture, content organization, user interaction, and innovation capability-operationalized through fifteen secondary indicators. Each indicator was carefully designed to capture AI-specific capabilities, including cognitive guidance efficiency, multimodal interaction precision, semantic network depth, and generation-enhanced utilization rate. A sophisticated hybrid weighting methodology was implemented, integrating subjective and objective approaches. For subjective weights, the Analytic Hierarchy Process was employed with 30 domain experts constructing pairwise comparison matrices using standardized scaling methods. Geometric mean aggregation was applied to synthesize individual judgments, with consistency ratios maintained below the threshold to ensure logical coherence. For objective weights, the entropy method analyzed actual evaluation data variance, with greater variance indicating higher discriminatory power. The final weights were derived through multiplicative synthesis combining both approaches. The empirical validation study involved collecting 492 valid questionnaires from 14 strategically selected public libraries representing different stages of AI implementation between September and November 2024: one municipal library with comprehensive AI deployment, 11 district libraries with partial implementation, and 2 county libraries in early adoption phases. The questionnaire utilized a five-point Likert scale to assess real-time service performance across multiple scenarios. Statistical analysis employed fuzzy comprehensive evaluation to handle uncertainty in subjective assessments, structural equation modeling to validate construct relationships, and latent class analysis to identify distinct user interaction patterns. The framework demonstrated high reliability with Cronbach's alpha reaching 0.845 and strong construct validity with KMO value of 0.873. [Results/Conclusions] Content organization emerged as the most critical dimension with a combined weight of 0.302 2, while semantic network depth, cognitive guidance efficiency, and cross-media consistency ranked as top secondary indicators with weights of 0.090 3, 0.086 1, and 0.084 7 respectively. Performance evaluation revealed content organization scoring 74.873 points versus user interaction at 68.040 points, highlighting the gap between technical capabilities and user experience. Significant differences existed across library levels, with municipal libraries outperforming county libraries by over one point in technical architecture and semantic network depth. Four distinct user patterns emerged: technology-oriented, content-immersive, efficiency-focused, and assistance-dependent. Each requires a tailored service approach. The study proposes the following optimization strategies: multimodal interaction frameworks, adaptive user profiling, hierarchical collaboration mechanisms, and knowledge graph-based content reorganization.

[Purpose/Significance] As digital government accelerates, the artificial intelligence (AI) literacy of grassroots civil servants has become critical to promoting smart government management. Grassroots-level civil servants who possess high levels of digital and AI literacy are indispensable drivers in establishing a digital and smart government. However, significant differences among grassroots civil servants in AI literacy and digital skills adaptation make it difficult for them to fully adapt to the requirements of smart government management. To effectively apply AI technologies in grassroots governance, it is essential for the effective application of AI technologies in grassroots governance to systematically identify its influencing factors and propose targeted cultivation paths, thereby improving public service quality and governance efficiency. [Method/Process] This study integrates the Technology Acceptance Model (TAM) and Innovation Diffusion Theory (IDT) to construct a TAM-IDT analytical framework. Based on empirical research identifying the AI literacy deficiencies of current grassroots civil servants, the TAM-IDT analytical framework systematically examines the impact mechanisms of key variables, perceived usefulness, perceived ease of use, and behavioral attitude, on AI literacy. The framework also proposes stage-based and group-specific cultivation strategies. The study uses local government civil servants as its research sample. It collects data through questionnaires and interviews, and employs structural equation modeling and mediation effect analysis for empirical validation. [Results/Conclusions] The findings reveal that behavioral attitude has a significant positive impact on AI literacy. Perceived usefulness notably enhances behavioral intention, while perceived ease of use has a negative effect on behavioral attitude, suggesting that individuals who perceive greater difficulty may be more motivated to learn. However, one of the highlights of this study is that civil servants who are proficient in AI technology or have used it in their work have a lower desire to learn more about it. Further analysis shows that perceived ease of use positively influences behavioral attitude indirectly through perceived usefulness. Additionally, both cognitive variables indirectly affect AI literacy via behavioral attitude, forming a "cognition-intention-behavior" influence chain. Based on these results and the classification of stages and types of technology adoption using Innovation Diffusion Theory (IDT), a three-dimensional, differentiated AI literacy cultivation strategy called "perception diffusion collaboration" was proposed. This strategy is based on the five elements, stages, and the groups of people involved in innovation diffusion. It offers a theoretical foundation and practical path for improving AI literacy among grassroots civil servants and advancing the modernization of grassroots governance.

[Purpose/Significance] Amid the global wave of digital transformation in education, artificial intelligence (AI) has emerged as a driving force behind Japanese educational reform, propelling the country's education system toward an "AI+" model. The "Approved Program for Mathematics,Data science and AI Smart Higher Education" (MDASH), led by the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT), outlines a comprehensive framework for designing and implementing AI literacy (AIL) education in Japanese universities. MDASH not only reflects the Japanese strategic response to the AI-driven future, but also provides valuable theoretical references and practical guidance for enhancing AIL education in China. This study provides a detailed analysis of the "MDASH literacy-level" (MDASHL) curriculum model design, paying a particular attention to the model's modules and the mechanisms of interaction between them. It also examines the theoretical references from MDASHL review system to the AIL framework studies. The study proposes innovative implementation strategies for AIL education from unique perspectives, especially the "industry-academia integration" aspect. [Method/Process] Using internet research and literature analysis, starting with an exploration of Japanese national AI policy landscape, the study traces the evolution of Japanese AI policies and the contextual origins of the MDASH. It describes the objectives and philosophy of Japanese AIL education and delves into the theoretical underpinnings of the MDASHL curriculum model based on the mapping relationship between indicators of AIL frameworks and the components of the MDASHL review system. We select Hokuriku University, Wakayama University, Chiba University, and Kansai Univerisity as samples because they were approved by MDASHL and demonstrated exemplary effects. We introduce their subject curriculum design and specific teaching initiatives, identify the commonalities and unique characteristics of their AIL education, and further elaborate on their specific educational implementation pathways. [Results/Conclusions] The findings indicate that the Japanese MDASHL curriculum model is deeply rooted in the AIL frameworks. It summarizes five educational directions for Japanese AI literacy education: recognition, realization, comprehension, ethics, and practical operation. By comparing the current status of AIL education in China and Japan, the study found that Japanese AIL education has achieved rapid responsiveness and systematic development under the unified coordination of MEXT. It suggests that Japanese AI literacy education strategies have localized value, from which beneficial insights can be drawn in three areas: strategic planning, curriculum design, and industry-academia integration. These strategies provide innovative solutions for developing AIL education systems in Chinese universities. However, this study acknowledges limitations in the sample size. To comprehensively capture the full landscape of Japanese AIL education development, future research should expand the sample size, summarize its patterns and characteristics more thoroughly, and enhance the persuasiveness and generalizability of the findings.