Close×
    2026 Volume 8 Issue 4
    Published: 01 October 2026
      
    • Select all
      |
    • WANG Haixia, XU Zonghuang, SHI Jin
      [Purpose/significance]Studying Finland's experience in critical information literacy education in dealing with AI disinformation can provide a reference for China to prevent related risks, optimize the information literacy education system, and enhance the national ability to identify disinformation. [Method/process]We systematically analyzed Finland's practice and achievements in critical information literacy education in response to AI disinformation, and summarized beneficial experiences. [Result/conclusion]Finland has formed a multi-subject collaborative national education network through a cross-departmental linkage mechanism, established a lifelong learning framework covering all stages of education, adopted an embedded teaching model to integrate critical thinking cultivation into various subject courses, and carried out response skills training combined with practical scenario, which can achieve the simultaneous improvement of cognitive and practical abilities. China should enhance the crucial role of critical information literacy education in countering AI disinformation, attach importance to lifelong literacy cultivation starting from early childhood, promote the in-depth integration of critical thinking into professional curriculum systems, strengthen the training of scenario-based response skills, and build an information literacy education system that meets the needs of the AI era.
    • ZHOU Lanyu
      [Purpose/significance]This study aims to provide an interdisciplinary solution to address the core issues existing in the construction of Chinese corpus, such as structural shortage of language materials, technical adaptation predicament, and insufficient cultural value penetration, thereby promoting the standardized construction of Chinese corpus resources, cultural inheritance, and the development of AI ethics. [Method/process]By integrating theories and methods from library and information science and cognitive linguistics, this study orderly answers the questions of "how Chinese values are carried by language" and "how to organize language materials carrying Chinese values", thereby enhancing the cultural sensitivity and ethical concern of Chinese corpus. Taking the "Chinese Traditional Festival Culture Corpus" as an example, the feasibility and effectiveness of this framework in addressing structural shortage of language materials, enhancing cultural semantic expression, and achieving value alignment are verified. [Result/conclusion]A value framework for the construction of high-quality corpus is systematically established, expanding the disciplinary boundaries of corpus construction and compensating for the shortcomings of semantic superficiality, cultural distortion, or disorganized and difficult-to-maintain language materials under a single disciplinary perspective.
    • GAO Zhi, SUN Gang
      [Purpose/significance]This study explores the evolutionary patterns of interdisciplinary research behavior among Nobel Prize laureates in natural sciences over their careers and its impact on academic performance, aiming to provide theoretical foundations and policy implications for the cultivation of high-level scientific talents and the management of interdisciplinary research. [Method/process] The study focuses on Nobel Prize laureates in natural sciences who began publishing papers since 1960. Using panel data models, it analyzes the authorship patterns, volume, interdisciplinarity, and academic impact of interdisciplinary publications from a career perspective. [Result/conclusion]The interdisciplinary behavior of Nobel Prize laureates in natural sciences exhibits significant stage-based evolution and role differentiation, the peaks for non‑interdisciplinary publications and first‑author interdisciplinary collaborative publications are concentrated in the early stage of academic career, while the peaks for interdisciplinary publications and non-first-author interdisciplinary collaborative publications are concentrated in the late stage of academic career. Among the three authorship patterns of interdisciplinary publications, non-first-author collaborative publications have the largest volume, the highest level of interdisciplinarity, and make the most significant contribution to academic impact. This study reveals the interdisciplinary growth logic of “establishing a foothold first and then crossing disciplinary boundaries” among natural sciences Nobel Prize laureates. It confirms that the value of interdisciplinary research depends on knowledge integration through team collaboration. Recommendations include training and evaluation for young academic talents shall be staged by career phases; interdisciplinary research governance shall move past individual cross-disciplinary incentives toward collaborative team construction; and the evaluation of interdisciplinary work shall shift from identity-based judgment to contribution-based measurement.
    • LI Xiaomin, ZHANG Xiaoqin, WANG Hao, BU Wenru, SHI Bin,
      [Purpose/significance]Establishing literary connections between poets from the perspective of allusion citation is of great significance for exploring information on unknown poets and understanding their works. [Method/process]This study investigates poets' citation patterns from the perspective of allusion citation, integrating both emotional and thematic features. Regarding emotional features, we construct a sentiment recognition model for poetic lines using the Dictionary of Tang Poetry Appreciation, which combines poems with their corresponding critical annotations. The model predicts the sentiment of each poetic line, which is then aligned with the sentiment of the cited allusion to derive citation sentiment. By quantifying the distribution of different citation sentiments for each poet, we identify their citation sentiment types. Regarding thematic features, we employ an author-topic model to extract citation themes, yielding the probability distribution of each poet across themes and generating thematic citation vectors. By combining sentiment types with thematic vectors, we construct poets' citation patterns, upon which community detection is performed based on pattern similarity to form poet groups. Within these groups, information about lesser-known poets can be inferred through the presence of prominent poets. Additionally, we explore the distribution of citation patterns within the same historical period and dominant themes. [Result/conclusion]The findings demonstrate that analyzing allusion citation provides an effective means of uncovering semantic associations among poets and offers auxiliary support for the discovery of works by poets who have not yet received substantial scholarly attention.
    • CHENG Tiejun, WANG Tong, WEI Jianxiang, FENG Lanping
      [Purpose/significance]Under the influence of cross-social media interactions, online public opinion can ferment simultaneously across different media platforms, deeply interact, and resonate multiple times, which is more likely to trigger a chain reaction of risks or even new types of risks. Therefore, it is significant to analyze the interactive influence of risks based on cross-social media public opinion data to promote the management of online public opinion towards preemptive prevention and early warning. [Method/process]Firstly, based on the analysis of the risk characteristics of cross-social media public opinion, a risk early warning indicator system for cross-social media online public opinion is constructed, and the Analytic Hierarchy Process(AHP) is adopted to determine the indicator weights. Secondly, the optimal segmentation method is used to identify the risk early warning thresholds for Weibo and Bilibili, as well as for aggregated cross-platform online public opinion. Finally, a dynamic early warning model for cross-social media public opinion risk based on evidence theory is developed. [Result/conclusion]Through case studys on three typical events including the Henan Rainstorm Incident, the results show that the proposed model is effective and practical in dividing online public opinion warning thresholds and providing dynamic real-time risk warnings for cross-social media public opinion. The findings can provide theoretical references for the early warning and management of online public opinion on cross-social media platforms.
    • ZHOU Haicheng, XU Shuo, MAO Jin, CHEN Liang, HE Guoxiu⁵, WANG Yuzhuo⁶, ZHANG Chengzhi⁷
      [Purpose/significance]In September 2025, the 8th Chengdu Conference on Scientometrics & Evaluation, hosted by the National Science Library (Chengdu), Chinese Academy of Sciences, and undertaken by the Scientometrics & Evaluation Research Center (SERC), was held in Chengdu. This article reviews the Academic Salon on Full-Text Bibliometric Analysis at the forum, summarizing methodological advances, application expansion, and paradigm shifts in full-text bibliometrics in the era of large language models (LLMs), and providing references for future research. [Method/process]Based on invited talks and on-site discussions, this article synthesizes the research paradigms and application directions of LLM enabled full-text bibliometrics. It highlights key viewpoints and practical approaches across topics including intelligent data annotation, fine-grained knowledge understanding, knowledge relation inference, knowledge tracing and flow analysis, scientific document understanding and automated review generation, and scientific research evaluation. [Result/conclusion]LLMs are driving full-text bibliometrics from shallow semantic analysis centered on topic terms and citation structures toward knowledge intelligence grounded in fine-grained knowledge understanding and knowledge reasoning. Information science research is undergoing notable changes in data utilization, interpretation of knowledge units, reasoning practices, explanation of bibliometric outcomes, and human-AI collaboration. Nevertheless, evaluating the reliability of LLM outputs, achieving a prospective and critical comprehension of scientific literature, and verifying the validity of bibliometric metrics are still the main challenges.
    • LIU Yinuo, LU Can, YANG Jianlin,
      [Purpose/significance]Against the backdrop of the platform economy, traditional misinformation detection tools are insufficient to identify "halt-true and halt-false" health information, which poses challenges to the public's health information literacy. Thus, systematic deconstruction of the narrative techniques and dissemination mechanisms of health marketing information is of critical academic and practical significance. [Method/process]From a data-driven perspective, this study constructs an integrated analytical framework characterized by the tri-layered interaction of "narrative construction, resource organization, and cognitive impact". Drawing on information ecology theory and framing theory, it further employs deep learning, semantic similarity analysis, and other methods to empirically deconstruct the data narratives of health marketing information in the context of milk tea consumption. [Result/conclusion]Research shows that health marketing narratives influence consumers' cognitive frameworks by constructing dialogic contexts and setting systematic expressing biases, thereby shifting their decision-making logic toward "health cognition". This study can provide theoretical support and practical guidance for enhancing the public's ability to discern health information and improving the health information ecosystem.
    • CONG Tianshi, ZHANG Wei
      [Purpose/significance]In response to the problems of non-standard expression in academic abstracts and lack of standardized methods for writing innovative highlights, this study explores an automatic generation method for innovative highlights based on the large language model. The aim is to provide researchers with diverse expression references and optimization suggestions, assist them in more accurately extracting and presenting research innovation points, improve the accuracy of academic innovation expression, and serve innovation content evaluation. [Method/process]Firstly, using the theory of ternary function in systemic functional linguistics, this paper analyzes the innovation highlights from multiple dimensions, such as innovation triggers, core terms, tense voice, sentence structure and abstract complementarity. Secondly, summarize the language expression characteristics of innovative highlights, and form a standardized language expression structure. Finally, the reasoning output of the large language model is optimized by designing the thinking chain according to the language characteristics to realize the automatic generation of innovative highlights. [Result/conclusion]Experimental results demonstrate that the chain-of-thought prompts designed in this study can effectively constrain the quality of text generated by large language models, with the generated research highlights performing well in terms of content richness, logical clarity, and expression accuracy. This research provides an automated solution to assist scholars in efficiently extracting key innovations and supports large-scale innovative highlight generation for publishing institutions.
    • LIU Xiao, WU Lei, SONG Yanfei, HUANG Xintong
      [Purpose/significance]Policies serve as a critical safeguard for guiding cutting-edge technological innovation and fostering a modern industrial system. Systematically evaluating the future industrial policy system and exploring its optimization paths are of great significance for enhancing the scientificity of policies and strengthening industrial competitive advantages. [Method/process]Based on a three-dimensional of framework "theme-tool-effectiveness", this study conducts a systematic quantitative assessment of 19 provincial-level future industry policies in China using a combination of content analysis, LDA-based topic modeling, and the PMC index approach. [Result/conclusion]The research indicate that while the thematic coverage is comprehensive, coordination is fragmented. Stronger links are needed across different government departments and regions. The policy tools show the structural imbalance of "strong supply side, weak environmental side, insufficient demand side". As a result, relevant innovation-chain and industry-chain integration is difficult. The overall effectiveness of policy implementation is good. But the policy system faces the problems of timeliness, evaluation precision, and iterative responsiveness. Consequently, this study proposes three corresponding optimization paths:establishing the system of cross-sector and cross-regional collaborative governance, making supply tools, environment tools and demand tools in dynamical balance, and optimizing policy by incorporating regulatory sandbox and third-party evaluation mechanisms. The research provides theoretical and practical references for resolving the structural contradiction of "high policy effectiveness-low dynamic adaptability". It also facilitates a shift in future industrial policy from isolated breakthroughs to systemic empowerment.
    • LAI Liubin, XIE Hua, PENG Chen, CHEN Lin
      [Purpose/significance]Collaborative innovation is an important means of breaking through one's own resource bottlenecks and achieving technological upgrading. Studying the impact of patent alliance relationships on collaborative innovation performance among alliance partners is of significant importance, and at the same time, it provides guiding value for enterprises in effectively utilizing patent alliances to promote open innovation. [Method/process]Based on the panel data of more than 5 000 pairs of MPEG LA members, this paper focuses on the alliance network of patent pools.A fixed effects negative binomial regression model was used to analyze the influence mechanism of alliances within patent pools on collaborative innovation performance between partners, and the moderating effects of network centrality, competitive learning and technological similarity. [Result/conclusion]The results show that alliances within patent pools have a positive impact on collaborative innovation performance.Network centrality and competitive learning significantly weaken the association between patent alliances and collaborative innovation performance, while technological similarity cannot significantly moderate the association between alliances and collaborative innovation.
    • TIAN Hongkun
      [Purpose/significance]In response to the frequent issuance of U.S. artificial intelligence (AI) policies, pronounced textual heterogeneity, and strengthening cross-agency linkages during the early period of Trump's second term, this study identifies the agenda structure, stage-specific shifts, and patterns of cross-agency coordination in U.S. AI policy, thereby providing a basis for science and technology security assessment in China. [Method/process]Using a corpus of 26 national-level and agency-level policy documents issued in 2025, this study constructs an ACE-LDA framework that builds on LDA document-topic distributions by incorporating cross-period topic alignment, a conservation mechanism for topic transitions, and a verifiable evidence chain. The framework is used to conduct topic identification, quarterly comparison, and agency-topic aggregation. [Result/conclusion]U.S. AI policy is structured around four themes: national-level AI action and competitiveness, agency-level AI strategies and capability building, federal AI governance and procurement rules, and agency compliance and risk management. Together, these themes exhibit a chain-like pattern of cross-agency coordination encompassing narrative framing, rule translation, use-case deployment, and implementation calibration. The policy focus gradually shifts from national mobilization toward translation into institutional arrangements, risk governance, and rule consolidation. The underlying governance logic is characterized by centralized goal setting, layered use of policy instruments, and feedback-based accountability. Particular attention should be paid to the spillover effects and associated risks arising from interactions among standards, procurement, compliance, and cross-border rules.
    • ZHAO Qiyu, MAO Taitian, LIU Yulan, LEI Yuanzhuo
      [Purpose/significance]Non‑standard procedures, insufficient multi‑stakeholder collaboration, and non‑compliant factor allocation are critical issues that urgently need to be addressed in government data quality management. [Method/process]Employing the meta-ethnography approach, this study integrates and analyzes 11 relevant primary documents to sort out and identify existing management challenges and explore improvement directions. [Result/conclusion]Research finds that government data quality management respectively faces the problem of disconnection, imbalance, and deficiency at the levels of processes, subjects, and factors, respectively. Accordingly, based on data governance theory, this paper proposes targeted solutions from three dimensions:establishing the concept of data value, balancing the demands of three types of stakeholders, and optimizing the allocation of capital, talent, and technological factors.