10 May 2026, Volume 46 Issue 5
-
Select all|
-
LIBRARY TRIBUNE. 2026, 46(5): 1-12.Researchers’ digital competency is a key element in promoting open science in the digital age. Based on fundamental concepts,primary activities,and the digital competencies required of researchers in the context of open science,this paper proposes 7 research hypotheses and 36 sub-propositions,which are subsequently validated using a questionnaire survey. The empirical study indicates that researchers’ digital competencies for open science encompass open science fundamentals and digital literacy,as well as digital competencies in open access,open data,open-source code and software,open peer review,open innovation,and open collaboration. Thus,the study presents a “Researchers’ Digital Competence Framework for Open Science”,consisting of competency domains and items. The framework establishes a competency benchmark for researchers engaging in open science practices,and guides them in continuously enhancing their digital competencies. This advances theoretical research and practical development in the field.
-
LIBRARY TRIBUNE. 2026, 46(5): 13-26.In-depth research on the diffusion and thematic evolution of personal information protection policies in China can help identify their basic characteristics and diffusion patterns,thereby enhancing the effectiveness of policy implementation. This article uses social network analysis,time series analysis,and thematic analysis to examine 377 personal information protection policies. By developing a reference relationship network,the study explores the characteristics of policy diffusion across three dimensions：time,space,and theme. Temporally,the diffusion of personal information protection policies in China exhibits a typical S-shaped curve. Spatially,they display uneven regional distribution and clustering effects,primarily following a top-down diffusion model supplemented by parallel diffusion,with certain ladder-like progressive features. Thematically,guided by the central government’s directives,the policies have become more detailed and specialized over time,with obvious regional variations while maintaining a balance of continuity and innovation.
-
LIBRARY TRIBUNE. 2026, 46(5): 27-38.The efficient operation of trusted data spaces relies on deep synergy among multiple elements. However,current construction practices demonstrate deficiencies in element synergy,highlighting the urgent need for theoretical construction and practical reflection on synergy. Through the cross-validation of policy texts,international experiences,and domestic practices,this paper identifies five core elements of trusted data spaces：institution,technology,actor,resource,and scenario. Following a static-to-dynamic analytical logic,the study develops a theoretical framework and mechanisms for multi-element synergy in trusted data spaces from the perspectives of embedding and coupling. To verify the explanatory power of this synergy theory,Shenzhen,Jiangsu,and Wenzhou were selected as case studies. The study suggests that trusted data space synergy follows the core mechanism of "embedding constrains coupling,and coupling shapes the path",manifesting as three differentiated models：the "platform-intermediary" model dominated by institution-technology coupling,the "regionally integrative" model characterized by balanced advancement of three major mechanisms,and the "pilot-breakthrough" model driven by scenario-value coupling. These models stem from distinctly preferential coupling pathways constrained by local embedding contexts,leading to differentiated trade-offs among market agility,system integration,and scenario-specific value.
-
LIBRARY TRIBUNE. 2026, 46(5): 39-52.Quantitative analysis of bibliometric differences and similarities -- such as manuscript length,number of authors,number of references,and research topics -- between uncited and cited papers in STEM disciplines can guide researchers in writing high-impact manuscripts and publishers in screening them,as well as in evaluating journal papers. Based on the citation data from Web of Science between 1997 and 2016 across four science and engineering disciplines,this paper compares the differences in page length,author numbers,and reference counts between zero-cited and cited papers within a five-year citation window after publication. Additionally,it examines variations in zero-citation and citation rates across papers with different page lengths,author numbers,and reference counts,and quantitatively analyzes thematic differences in uncited papers. The results suggest that the zero-citation rate is inversely proportional to page lengths,author numbers and reference counts. As these metrics increase,the non-citation rate shows a declining trend,eventually stabilizing at a low level. The range of eigenvalues at which the non-citation rate declines to a lower stable level varies by science and engineering discipline. There is no significant difference in the trend of non-citation rates increasing with literature eigenvalues between science and engineering disciplines,or between the periods of 1997-2001 and 2012-2016. In both science and engineering disciplines,and in both the early （1997-2001） and new （2012-2016） periods,authors tend to cite longer papers with more references. Furthermore,the non-citation rate of specific papers may be influenced by research topics,topic popularity,and time differences in thematic development.
-
LIBRARY TRIBUNE. 2026, 46(5): 53-64.In the era of digital intelligence,digitizing ancient books has become an important means of promoting the activation of traditional classical knowledge and the innovation of humanistic research paradigms. As comprehensive repositories of ancient Chinese knowledge,LeiShu （encyclopedias in ancient China） face challenges in digitization due to the heterogeneity of their knowledge systems,caused by differences in text storage,retrieval techniques,and compilation styles. Reconstructing the knowledge organization systems of these encyclopedias based on specific ontologies and developing unified semantic models can enable cross-encyclopedia knowledge aggregation,intelligent compilation and restoration. Using the Yongle Encyclopedia,Pianzhi,and Complete Records of All Flowers as datasets,this paper presents a LeiShu ontology comprising 101 classes,93 object properties,and 44 data properties. Core classes include "Leishu" “Compilation Style” and "Content Structure". A four-tier conceptual mapping mechanism was developed to standardize directory hierarchies across different encyclopedias. Based on the thematic categories of “Heaven,Earth,Humans,Affairs,and Objects,” content from different encyclopedias was aggregated to form an interconnected knowledge network. By integrating diverse and heterogeneous knowledge systems from the encyclopedias through the ontology,the study establishes a semantic association network spanning eras and themes,thus effectively addressing challenges such as the difficult alignment of knowledge units and weak theme associations. This approach provides a technical solution for discovering and reconstructing lost content,and offers insights into the intelligent preservation of ancient books and the innovative utilization of cultural heritage.
-
LIBRARY TRIBUNE. 2026, 46(5): 65-72.Digital humanities （DH） infrastructure is a crucial foundation for digital humanities research. In the digital intelligence era,driven by technologies such as generative AI,examining the needs of humanities scholars regarding DH infrastructure can shed light on functional services and the strategic development of infrastructure geared toward humanities disciplines and intelligent upgrades. This study identifies the elements of DH infrastructure needs based on in-depth interviews and grounded theory. Using the Kano model and the better-worse satisfaction coefficient,the paper categorizes these elements,and conducts a comparative analysis of the distinct needs and requirements of scholars in history,literature,and the arts. The findings reveal that the needs of humanities scholars encompass three dimensions—data,systems,and services—involving 26 specific requirements. These needs can be categorized as must-have,one-dimensional,attractive,or indifferent. While the overall needs and requirements of scholars across these disciplines are generally consistent,distinct disciplinary emphases persist. Strategically,systematic development of data,functions,services,and facilities must be strengthened to advance the evolution of DH infrastructure toward intelligence,integration,and normalization.
-
LIBRARY TRIBUNE. 2026, 46(5): 73-84.Identifying named entities in classical texts—the vital repository of Chinese civilization—holds great significance for advancing digital humanities,digitizing ancient manuscripts,and building knowledge graphs of ancient books. To address challenges in model architecture and transfer learning capability for named entity recognition （NER） in classical texts,this paper proposes AT-LSFT：a novel framework based on the collaboration of large and small models,as well as system fine-tuning strategies. Specifically,the large language model,LLaMA3-8B,provides semantic guidance and entity pre-localization,while the smaller BiLSTM-CRF model handles feature decoupling and fine-grained entity recognition. Through three phases of system fine-tuning involving prompt engineering,knowledge distillation,and dynamic feedback,the two models effectively balance semantic comprehension and reasoning efficiency. Extensive experiments on 19 baseline models,including traditional,deep learning-based,pre-trained,and large language models,show that AT-LSFT achieves superior performance with an F1 score of 84.8%. The framework consistently outperforms existing mainstream methods in recognizing various types of classical text entities,such as persons,locations,official titles,and organizations,demonstrating stronger semantic parsing capabilities and generalization performance. It also excels at recognizing low-frequency entities and resolving semantically ambiguous samples.
-
LIBRARY TRIBUNE. 2026, 46(5): 85-96.As key agents in the transmission of cultural heritage,book collectors’ spatial and temporal distribution throughout Chinese history provides valuable insights into how knowledge preservation,shifts in cultural centers,and political-economic structures interact over time. From a digital humanities perspective,this study employs geographic information system concepts to analyze the native-places and dynasties of 1,443 book collectors from the pre-Qin period to the Republic of China era,revealing the spatiotemporal patterns in the distribution of their places of origin. The findings demonstrate a steady increase in the annual average number of book collectors. Spatially,a gradient decrease pattern emerged from the coastal core regions （Jiangsu and Zhejiang） toward the inland peripheral areas. The geographic distribution of book collectors differed between the northern and southern regions：in the north,they concentrated in provincial capitals,in the south,they dispersed along the Yangtze River basin and the southeast coast. Temporally,the origins of book collectors underwent clear regional transitions across dynasties. Initially centered in the north from the Qin through the Sui-Tang period,the collectors’ native places gradually shifted southward from the Five Dynasties and Ten Kingdoms period to the Song-Yuan era;finally,they concentrated in the Jiangnan and southeastern regions during the Ming and Qing dynasties and the Republic of China period. This ongoing southward shift reflects the developmental trajectory shaped by the interplay of political,economic,technological,and cultural factors throughout Chinese history.
-
LIBRARY TRIBUNE. 2026, 46(5): 97-106.With the application of artificial intelligence （AI） in the library sector,AI virtual librarians have become a focus for public libraries exploring intelligent service transformation. Based on a survey of public libraries in China’s 31 provinces and 333 cities,this study indicates that 155 of them have deployed a total of 170 AI virtual librarians. This paper outlines the characteristics of AI virtual librarians in terms of deployment,virtual image,interaction methods,and technical support;and examines their cultural expression,image design,service accessibility,and technological governance. The study recommends integrating virtual librarians more closely with local cultures and library collections,promoting diverse and professional visual designs,improving accessibility,and enhancing technological transparency and ethical controllability to promote the high-quality development of smart library services.
-
LIBRARY TRIBUNE. 2026, 46(5): 107-118.Measuring sentiment in various aspects of peer review comments and analyzing reviewer disagreements provides journal editors and conference chairs with ideas for optimizing the review process and improving evaluation criteria. It also helps authors refine controversial aspects of their manuscripts,thereby increasing the likelihood of acceptance. A corpus of 37,847 publicly available peer reviews was created using documents collected from the Nature Communications website between April 2016 and November 2024. Aspects of peer review were identified,and the performance of large language models in sentiment classification across these aspects was compared. Then,the optimal model was used to screen the widely reviewed aspects and process the entire corpus. Finally,reviewer disagreement was measured using cosine distance and standardized entropy. Experiments indicate that DeepSeek-V3 achieves optimal performance in sentiment classification of review aspects at a temperature parameter of 0.6. Reviewers generally evaluate five key aspects：motivation/impact,clarity,soundness,substance,and originality/novelty. The degree of disagreement in peer review comments varies by disciplinary field. The highest average divergence is observed in the physical sciences,followed by the sciences and the social sciences. Additionally,the degree of disagreement varies among the different review aspects. Reviewers tend to reach a consensus more easily on motivation/impact,but are more likely to disagree about soundness and substance.
-
LIBRARY TRIBUNE. 2026, 46(5): 119-126.As an important innovation in blockchain technology,the non-fungible token （NFT） has attracted widespread attention. However,discussions within the library community mostly focus on technological applications,and lack an in-depth theoretical review from a legal perspective. This paper examines disputes over the legal nature of NFT,focusing on theories of property rights,creditor rights and intellectual property rights;and argues that NFT is a special type of property interest related to creditor rights,characterized by relativity and performance obligations. Based on this analysis,the study presents a framework of creditor rights and an implementation path for NFT-enabled libraries. At the management level,it strengthens information storage security and intellectual property governance by reconstructing multi-party service contracts and clarifying the nature of claims. At the service level,it leverages the divisibility and programmability of creditor rights to achieve the differentiated supply of digital cultural resources and dynamic incentives for community governance. The article proposes standardized guidelines from the perspectives of system development and risk management：advocating for coordinating fundamental rights reviews,personal information protection and public cultural positioning at the level of system compliance;and establishing a comprehensive governance system that encompasses intelligent contract auditing,the synergy of legal and technical rules,and a performance emergency mechanism at the level of risk prevention and control. Thus,it ensures the legitimacy and robustness of the digital transformation of libraries.
-
LIBRARY TRIBUNE. 2026, 46(5): 127-140.The development of emerging technologies is making the interaction between technology and society increasingly complex. Exploring the generation and evolution mechanisms of technology’s social impacts from a dynamic relational perspective is not only an inherent requirement for refining technology governance theories but also provides critical decision-making support for optimizing industrial policies. Drawing on socio-technical systems theory,this study uses chats about “autonomous driving” from the Weibo platform as its data source. Causal event pairs were extracted using the DeepSeek-V3 Large Language Model,and event semantic generalization was achieved through the BERTopic model. The paper develops an event evolutionary graph of public opinions on science and technology,revealing the dynamic patterns in “technology-society” interactions by integrating multi-dimensional attributes such as sentiment analysis and influence metrics. The results show that the social impacts of autonomous driving technology exhibit orderly transmission characteristics,revealing the nonlinear adaptation process through which technologies become embedded in society. Public sentiment exerts a reverse shaping effect on technological research and development （R&D） directions and institutional rule formation,where persistent negative feedback on specific issues drives the emergence and discussion of governance topics. High-influence events accelerate the system’s evolution. Meanwhile,mid-to-high influence events exhibit high synchrony with fluctuations in overall public opinion,and low-influence negative opinions have a cumulative risk effect. By integrating multiple attributes into an event evolutionary graph,this study systematically reveals the multidimensional social impacts triggered by emerging technologies and their evolutionary mechanisms. Thus,it provides technology companies with data-driven decision support for optimizing R&D strategies and helps research management departments improve policies.
-
LIBRARY TRIBUNE. 2026, 46(5): 141-149.With the growing popularity of generative AI,privacy risks have expanded from the information layer to the data layer. The lying-flat attitude toward data privacy poses new challenges for privacy protection. Exploring the factors that influence privacy lying-flat behavior can foster rational comprehension of this phenomenon and provide a localized perspective for explaining the privacy paradox. Through a grounded theory-based coding analysis of the transcripts from thirty interviewees,this paper examines the factors influencing lying-flat behavior regarding personal data privacy within the framework of the Theory of Planned Behavior. The study indicates that lying-flat behavior occurs due to factors such as affective and instrumental attitudes,internal and external control capabilities,and subjective norms. These factors collectively influence the intention to engage in lying-flat behavior through privacy attitudes,subjective norms,and perceived behavioral control,thereby affecting the behavior itself.
-
LIBRARY TRIBUNE. 2026, 46(5): 150-160.With The rapid development of generative artificial intelligence （GAI）,the paradigm of scientific research and innovation is undergoing profound changes. In the context of human-AI collaboration,graduate students encounter both opportunities and challenges in strengthening their knowledge innovation capabilities. Through in-depth interviews and qualitative content coding analysis,this paper identifies the elements that constitute these capabilities,including knowledge accumulation and acquisition,knowledge representation and interaction,knowledge criticism and discernment,knowledge transfer and integration,knowledge security and risk control,and knowledge innovation tool use. The study proposes the following pathways to enhance graduate students’ knowledge innovation capabilities：dual-channel knowledge accumulation,iterative prompt optimization,critical thinking and knowledge value assessment,knowledge transfer and integration,ethical awareness and risk governance,and tool-enabled collaborative optimization.
