10 March 2026, Volume 46 Issue 3
-
Select all|
-
Efficiency Evaluation of Urban Public Library Services in Guangdong-Hong Kong-Macao Greater Bay AreaLIBRARY TRIBUNE. 2026, 46(3): 1-11.Evaluating the efficiency of urban public library services is crucial for improving the effectiveness of public cultural services,accurately identifying resource redundancy or shortages,and optimizing resource allocation. Based on variable returns-to-scale DEA-BCC,SBM super-efficiency,and Tobit models,this paper presents an index system that evaluates the efficiency of urban public library services,using data from 11 cities in the Guangdong-Hong Kong-Macao Greater Bay Area from 2017 to 2023. The study then analyzes the service efficiency of urban public libraries in the region and its influencing factors. The results show that,although cities in the Greater Bay Area have largely ensured equitable access to public library services,the efficiency of these services still have room for improvement in some cities. Some libraries exhibit input-output performance with redundancy and insufficiency;and optimizing resource allocation and promoting resource circulation are key to enhancing public library service efficiency. Accordingly,urban public libraries in the Greater Bay Area should improve their service efficiency through systematic planning according to local conditions,optimize resource utilization to reduce inefficiencies coordinately,and mobilize critical resources to improve the efficacy of resource circulation.
-
LIBRARY TRIBUNE. 2026, 46(3): 12-24.Understanding the key factors of technological leaps is essential for innovators seeking to capitalize on emerging opportunities and accelerate industrial progress. Using patents as discrete units of technology,this paper examines their knowledge foundation through backward citations,and presents measurable indicators across five dimensions：diversity,depth,quality,recency,and time domain. Taking new energy vehicle power battery technology and additive manufacturing technology as examples of emerging and disruptive technologies,respectively,the study explores how the knowledge bases of patented technologies in different fields influence technological leaps by means of binary regression analysis. The results suggest that diversity,depth,and recency facilitate leap transitions in both types of technology,whereas the impact of quality and time domain varies by field.
-
LIBRARY TRIBUNE. 2026, 46(3): 25-34.In recent years,China has preliminarily established a diversified public cultural supply system led by the government and supported by active participation from society and the public. The integration of various types of resources—including basic public cultural services,inclusive non-basic public cultural services,and market-based cultural services—has accelerated. Transmission channels for public cultural resources that are spatially accessible,available online and servable,have continuously expanded,steadily enhancing the capacity to deliver targeted resources to grassroots communities. However,challenges remain,including insufficient coordination among service providers,unclear delineation of responsibilities,an inadequate and structurally imbalanced supply of high-quality public cultural resources,insufficient channels for allocating resources to the community level,low supply-demand matching,and limited primary-level capacity to receive and utilize these resources. Thus,it is imperative to establish a comprehensive operational system for public cultural services,as well as long-term,multidimensional guarantee mechanisms that cover the entire supply,transmission,reception,and feedback chain. Priorities include developing a supply-demand matching mechanism for precise resource allocation,improving the government guarantee mechanism with institutional support,strengthening the social participation mechanism to enhance service momentum,optimizing the resource integration mechanism to boost allocation efficiency,implementing the resource distribution mechanism to facilitate grassroots access,and improving the talent development mechanism to build grassroots capacity.
-
LIBRARY TRIBUNE. 2026, 46(3): 35-45.Adopting a combined approach of literature review and case analysis,this paper explores the intrinsic logic and cultural mechanisms of digital humanities within the context of academic paradigm shifts,focusing on the phenomenon of "ubiquitous digital humanities". Through the theoretical lens of "paradigm revolution",the study presents a conceptual framework for digital humanities encompassing seven dimensions：computation,association,visualization,resources,critique,collaboration,and sharing. Using case studies such as Digital Dunhuang,Imperiia,To See the Tree and the Forest,and Wu Yu Dian,the paper analyzes the practical logic of digital humanities in terms of platform collaboration,knowledge representation,and cultural dissemination. In the era of AGI,the paper argues that digital humanities are becoming increasingly embedded in the knowledge system as a form of "cultural gene",bridging tradition and the future. The study concludes by emphasizing that the future value of digital humanities lies not only in technological integration,but also in its role in fostering new forms of humanistic knowledge characterized by openness,criticality,and generativity. Thus,digital humanities become a key driver for advancing the humanities.
-
LIBRARY TRIBUNE. 2026, 46(3): 46-54.As a vital national strategy,the international communication of Chinese public culture must overcome challenges such as the conundrums of voice and reception. One possible solution is participating in the governance of professional international organizations. Through case studies,this article identifies three ways Chinese libraries can get involved in IFLA governance：achieving identity integration through membership and elections,promoting collaboration and interaction through agenda setting,and strengthening symbolic recognition through professional practices. The study recommends that China strengthen its participation in the governance of specialized international organizations in the public cultural sector by bolstering its knowledge reserves,enhancing its responsibility and accountability,engaging in agenda setting,and developing collaborative platforms. These insights could help China improve its effectiveness in future international communication.
-
LIBRARY TRIBUNE. 2026, 46(3): 55-64.Drawing from classic service quality gap theory and the LibQUAL+TM model,this study proposes an index scale to evaluate the quality of intelligent reference services in libraries,and tests its operability. Through an empirical study of Nanjing University Library,the paper systematically assesses the quality of its smart reference services using methods such as questionnaires,factor analysis,and matrix analysis. Based on the research findings,the study recommends that libraries improve their smart reference services by optimizing user interfaces,developing feedback mechanisms,and enhancing service efficiency.
-
LIBRARY TRIBUNE. 2026, 46(3): 65-75.To explore the changing characteristics and trends in public demand for intellectual property （IP） information services,this paper examines public inquiries from the China National Intellectual Property Administration’s website over the past ten years,using a mix-methods approach combining grounded theory and time-series analysis. The results suggest that the theoretical model of public IP information service demand consists of four core categories：IP creation,application,management and protection. Over the past decade,demand related to IP creation has declined,while demand for IP application,management and protection has increased. To improve IP information services,the study proposes expanding service delivery models based on public demand,innovating dissemination and utilization approaches by optimizing service content,and strengthening information service platform development to enhance service efficiency.
-
LIBRARY TRIBUNE. 2026, 46(3): 76-88.Knowledge graphs have emerged as a new form of knowledge organization for public digital cultural resources. Exploring the quality requirements and control strategies for these graphs is of great significance to promoting their high-quality development and service value. This article obtains quality requirement data on public digital cultural resource knowledge graphs from multiple sources,hierarchically analyzes the data based on grounded theory,and develops a quality requirement model. Using the IPA method,it explores differences in users' emphasis and satisfaction with quality requirements to determine the attribute characteristics of each requirement. According to the QFD framework,it establishes a quality house for public digital cultural resource knowledge graphs,thereby identifying quality control measures that meet quality requirements and establishing their priority ranking. Consequently,the study presents a quality requirement model for public digital cultural resource knowledge graphs,comprising five progressive dimensions：characteristic quality,product quality,utility quality,management quality,and cultural quality. The model clarifies the priority sequence of each quality dimension in quality control,and proposes targeted quality control strategies.
-
LIBRARY TRIBUNE. 2026, 46(3): 89-99.Products recorded in local chronicles have a significant impact on the preservation and development of a region's culture and economy. A new study of local chronicle resources within the interpretive framework of digital narratives can provide fresh perspectives on their contemporary value. Using a grounded theory approach,this paper identifies the core formative elements of product knowledge services based on local chronicles within digital narratives,and examines the relationships and logic among these elements. Within the networked narrative structure proposed by Mary-Raul Ryan,the study reveals the progressive cyclical logic chain of "product data - product knowledge - product narrative - product value" hidden within local product knowledge services. Through case studies,the paper proposes countermeasures,such as tracing the origins of local product germplasm resources,shaping cultural memories of local products in multiple dimensions,and promoting the deep integration of regional cultural and tourism industries.
-
LIBRARY TRIBUNE. 2026, 46(3): 100-111.This article presents an index system for evaluating the quality of health rumor refutation information generated by AI,providing the theoretical foundation and practical guidance for AI applications in the field of public health. Through a human-AI collaborative approach,the study designs three types of prompt engineering - simple,detailed,and expert-based - to drive generative AI in developing a quality evaluation index system for debunking health-related misinformation. Using multi-agent collaborative scoring,the study analyzes the strengths and weaknesses of AI-powered rumor refutation;and compares them with traditional methods. The results demonstrate the importance of expert involvement in building evaluation index systems,which ensures their comprehensiveness and professionalism. However,expert participation has a limited effect on improving the overall quality of health rumor refutation information produced by generative AI when assessing the quality of health rumor denials in this system. Content refuting health rumors generated through human-AI collaboration outperforms content produced by AI or experts alone,highlighting the superiority of the dual-subject human-AI approach. Additionally,prompt engineering is essential for balancing quality and efficiency.
-
LIBRARY TRIBUNE. 2026, 46(3): 112-123.Long document processing is a crucial capability for applying large language models （LLMs） in knowledge-intensive domains,yet systematic,empirical evaluations of their performance and core bottlenecks remain insufficient. This paper presents a multi-level,cross-genre framework for evaluating the performance of five leading LLMs on two types of corpora：popular reading materials and academic literature. The models are tested across five dimensions through three progressively difficult tasks at the paragraph,chapter,and full-text levels：content extraction accuracy,information coverage,citation existence,citation correctness,and citation localization precision. The study reveals significant asymmetry in the models’ ability to process long documents. While most LLMs can accurately extract local information,their performance sharply declines in tasks requiring full-text integration. Gemini 2.5 Pro and Claude 3.7 Sonnet demonstrate superior overall performance and robustness. Furthermore,normalized text structure enhances model stability when processing local information but significantly impacts citation traceability,and has limited effectiveness in improving holistic integration capacity.
-
LIBRARY TRIBUNE. 2026, 46(3): 124-135.This paper explores methods for fine-tuning large language models （LLMs） to generate SQL queries from natural language descriptions. This enables the retrieval of library database reports and overcomes the limitations of existing business systems’ rigid query functions,which struggle to meet various operational needs. Based on reporting analysis and decision-making needs within various library divisions,the study creates specialized Text-to-SQL training and test datasets tailored to library operational databases and the Chinese Library Classification System. LoRA fine-tuning was performed on the training dataset using medium-sized LLMs with six to seven billion parameters from domestic and international sources. The study validates the effectiveness of SQL query generation using the test dataset;and deploys the optimal model via FastChat. The model was then integrated into a visual interactive system,whose performance was analyzed in real-world scenarios. Through fine-tuning,the Chatglm3-6B model achieved an execution accuracy of 0.9426. After examining the effects of adopting fine-tuned medium-sized LLMs and the ultra-large Qwen-Max model in real operational interactions,the study confirms the effectiveness of Text-to-SQL fine-tuning for LLMs when processing natural language queries on business data from libraries.
-
LIBRARY TRIBUNE. 2026, 46(3): 136-147.To address limitations in “stranglehold” technology research,such as the absence of identification mechanisms for alternative technologies and imprecise analysis of technological elements,this paper proposes a method that integrates prompt engineering with BERT-LSTM models to identify “stranglehold” technologies. First,ECCN items are analyzed based on the Commercial Control List （CCL）,and key core patents along the main technical path are identified using the SPC algorithm through patent searches. Then,large language model-based prompt engineering is employed to extract “problem-solution pairs”,and examine the technical efficacy of the approach. The function-oriented search （FOS） tool is used to preliminarily identify patents with alternative technical effects. Next,the BERT-LSTM model is employed to accurately identify sample patents with technical substitution capabilities based on the binary classification of patent texts. Prompt engineering enables the extraction of “solution-category pairs” and the systematic identification of alternative technical solutions. Finally,a scientific-industrial evaluation system is established to grade the potential of alternative technologies. Taking lithography technology as an example,this study outlines the application process of this method,and systematically identifies five alternative technologies for EUV technology and their substitution potential.
-
LIBRARY TRIBUNE. 2026, 46(3): 148-160.To address issues such as semantic ambiguity,information fragmentation,and poor adaptability in traditional knowledge extraction,this paper presents a method for creating tourism knowledge graphs based on large language models and prompt engineering （TTKE-LLM）. The method aims to enhance the performance and generalization capabilities of knowledge extraction,thus advancing semantic modeling for smart tourism. First,the study introduces a tourism knowledge representation schema that covers core categories such as attractions,transportation,and cuisine. Then,in response to the complex nature of travel texts,it proposes a step-by-step prompting strategy incorporating few-shot examples,chain-of-thought reasoning,and constrained prompts. This strategy extracts high-quality triplets from unstructured texts by guiding the DeepSeek large language model to perform instruction fine-tuning with minimal manual annotation. Next,multi-source data were integrated to build a Harbin tourism knowledge graph,enabling semantic fusion and visualization. The effectiveness of the proposed method was validated through comparative and ablation experiments. Experimental results show that TTKE-LLM outperforms comparative models in entity recognition and relation extraction tasks;and that the stepwise prompting strategy significantly improves extraction performance. The framework demonstrates strong adaptability in recommendation and Q&A scenarios,and provides a technical approach that is both replicable and scalable for smart tourism.
