中文    English
Current Issue
05 February 2025, Volume 37 Issue 2
Analysis of Progress in Data Mining of Scientific Literature Using Large Language Models | Open Access
CAI Yiran, HU Zhengyin, LIU Chunjiang
2025, 37(2):  4-22.  DOI: 10.13998/j.cnki.issn1002-1248.25-0116
Asbtract ( 46 )   HTML ( 6)   PDF (1797KB) ( 343 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] Scientific literature contains rich domain knowledge and scientific data, which can provide high-quality data support for AI-driven scientific research (AI4S). This paper systematically reviews the methods, tools, and applications of arge language models (LLMs) in scientific literature data mining, and discusses their research directions and development trends. It addresses critical shortcomings in interdisciplinary knowledge extraction and provides practical insights to enhance AI4S workflows, thereby aligning AI capabilities with domain-specific scientific needs. [Method/Process] This study employs a systematic literature review and case analysis to formulate a tripartite framework: 1) Methodological dimension: Textual knowledge mining uses dynamic prompts, few-shot learning, and domain-adaptive pre-training (such as MagBERT and MatSciBERT) to improve entity recognition. Scientific data extraction uses chain-of-thought prompting and knowledge graphs (such as ChatExtract and SynAsk) to parse experimental datasets. Chart decoding uses neural networks to extract numerical values and semantic patterns from visual elements. 2) Tool dimension: This explores the core functionalities of notable AI tools, including data mining platforms (such as LitU, SciAIEngine) and knowledge generation systems (such as Agent Laboratory, VirSci), with a focus on multimodal processing and automation. 3) Application dimension: LLMs produce high-quality datasets to tackle the issue of data scarcity. They facilitate tasks such as predicting material properties and diagnosing medical conditions. The scientific credibility of these datasets is ensured through a process of "LLMs + expert validation". [Results/Conclusions] The findings indicate that LLMs significantly improve the automation of scientific literature mining. Methodologically, this research introduces dynamic prompt learning frameworks and domain adaptation fine-tuning technologies to address the shortcomings of traditional rule-driven approaches. In terms of tools, cross-modal parsing tools and interactive analysis platforms have been developed to facilitate end-to-end data mining and knowledge generation. In terms of applications, the study has accelerated the transition of scientific literature from single-modal to multimodal formats, thereby supporting the creation of high-quality scientific datasets, vertical domain-specific models, and knowledge service platforms. However, significant challenges remain, including insufficient depth of domain knowledge embedding, the low efficiency of multimodal data collaboration, and a lack of model interpretability. Future research should focus on developing interpretable LLMs with knowledge graph integration, improving cross-modal alignment techniques, and integrating "human-in-the-loop" systems to enhance reliability. It is also imperative to establish standardized data governance and intellectual property frameworks to promote the ethical utilization of scientific literature data. Such advances will facilitate a shift from efficiency optimization to knowledge generation in AI4S.

Perception and Utilization of Digital Literacy Knowledge by University Library Users from Representational Pers Perspective | Open Access
SUN Zhimei, LIU Yan, LIU Wenyun, LI Ruiqin
2025, 37(2):  23-36.  DOI: 10.13998/j.cnki.issn1002-1248.25-0031
Asbtract ( 44 )   HTML ( 8)   PDF (788KB) ( 29 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] Digital literacy education has become the new educational mission of university libraries. Clarifying the user's perception and utilization mechanism of digital literacy knowledge and optimizing the representation of digital literacy knowledge can promote university libraries to achieve satisfying results in digital literacy education. Based on the frontier of representation theory, this study innovatively puts forward the concept of "sensory digital literacy education", constructs a three-dimensional knowledge perception model including action, image and symbolic representation, and reveals the mechanism of digital literacy knowledge representation and user perception behavior through empirical research. It provides a theoretical anchor for the paradigm shift in library education from tool skills training to cognitive skills training. The "cognition-practice-innovation" teaching system and the "three-in-one" resource construction framework proposed in the study effectively connect the knowledge representation theory with the educational practice scene, and provide a viable way for the three-dimensional implementation of digital literacy education in colleges and universities. [Method/Process] Based on the theories of SOR, TAM and self-efficacy, the theoretical hypothesis model of users' perception and utilization of digital literacy knowledge from the perspective of representation was constructed, and was empirically verified by questionnaire and empirical study. [Results/Conclusions] Action representation, reflexive representation and symbolic representation of digital literacy knowledge all positively affect users' perceived ease of use and perceived usefulness of digital literacy knowledge; perceived ease of use has a positive impact on perceived usefulness; self-efficacy plays a positive moderating role between perceived ease of use, perceived usefulness, and user intention and behavior. Due to the limitations of space and personal energy, the shortcomings of this paper are as follows. First, the methodological level is mainly based on quantitative analysis, and the mining of qualitative dimensions such as details of teacher-student interaction and informal learning scenarios in digital literacy education is insufficient. Secondly, the research object focuses on the groups of teachers and students in colleges and universities, and the issues such as the intergenerational differences of the public's digital literacy and the professional digital literacy needs of professionals have not been covered, and the comparative study of multiple subjects can be expanded in the future. In the future, more research can be done on research methods and research objects. Through the deep coupling of representation theory and educational practice, it is expected to provide a new theoretical mirror for the cultivation of cognitive ability in the digital age, and help to build a three-dimensional educational ecology of "technology empowerment-cognitive development-literacy transfer".

Risks and Regulations for Application of the LLaMA Model in University Future Learning Centers | Open Access
QIAO Jinhua, MA Xueyun
2025, 37(2):  37-48.  DOI: 10.13998/j.cnki.issn1002-1248.25-0139
Asbtract ( 29 )   HTML ( 5)   PDF (895KB) ( 303 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] The rapid advancement of artificial intelligence (AI) technology is transforming various sectors, particularly in higher education. The LLaMA (Large Language Model Meta AI) represents a significant innovation in this arena, making its application within university future learning centers increasingly important. As institutions of higher education strive to create environments conducive to learning and growth, understanding the construction requirements of future learning centers becomes paramount. This study delves into the integration of LLaMA core technologies in these learning spaces and emphasizes the importance of evolving libraries into intelligent learning support systems. [Method/Process] The methodology employed in this research combines technical deconstruction and scenes for validation, allowing for a comprehensive analysis of the legal risks associated with embedding advanced technologies in educational frameworks. By systematically examining these potential risks, the study aims to establish a well-rounded perspective on the implications of AI deployment in educational settings. [Results/Conclusions] The study identifies three principal challenges encountered in the application of the LLaMA within university learning centers. The first challenge arises from reliability risks linked to content generated by the AI, which may be affected by biases present in the training data. Such biases can lead to the dissemination of inaccurate or misleading information, undermining the trustworthiness of educational resources. Secondly, there are privacy leakage risks, particularly associated with the retention of user behavioral data. As AI systems analyze user interactions, there is a potential for sensitive information to be exposed or misused, raising concerns about student privacy and data security. The third challenge involves ownership determination dilemmas regarding the content generated through AI-driven creative processes. These dilemmas are intricately tied to existing copyright law frameworks, which may not adequately address the complexities introduced by human-machine collaboration in content creation. In response to these challenges, the study proposes several pathways for governance aimed at effectively navigating the landscape of AI in education. It suggests the implementation of dynamic data cleansing mechanisms to address reliability risks and inaccuracies. Additionally, establishing tiered privacy protection systems can help safeguard against user data breaches. Legal frameworks also need refinement to ensure clear ownership distribution for outputs of human-machine collaboration. Ultimately, optimizing the application of the LLaMA model in university future learning centers necessitates a careful balance between technological innovation and legal regulation. By focusing on technical refinement, risk control, and relevant regulatory measures, the development and application of AI can be advanced, facilitating a more integrated evolution of artificial intelligence and educational practices.

Influencing Mechanism of the Social Role of Government Digital Human on Public Adoption Behavior | Open Access
WEI Tianyu, LIU Zhongyi, ZHANG Ning
2025, 37(2):  49-60.  DOI: 10.13998/j.cnki.issn1002-1248.25-0142
Asbtract ( 49 )   HTML ( 11)   PDF (755KB) ( 98 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] Under the background of digital government construction, as a new type of service subject of human-machine collaborative governance, the influence mechanism of the social role positioning of government digital humans on public adoption behavior urgently needs theoretical exploration. Most existing studies have focused on the technical level. This study, based on the perspective of social role theory, explores the influencing mechanism of different role positioning of government digital humans in government service scenarios on public adoption behavior, which is of great significance for optimizing government services and improving the intelligent level of government services. [Method/Process] An experimental research method was adopted to construct a two-factor inter-group experimental design of "social role-business type", and a simulation experiment of government service scenarios was carried out through random grouping. Based on previous studies, we defined the role positioning of "advisors" and "decision-makers" for government digital humans, and constructed experimental scenarios by combining two service scenarios of consultation and approval. The subjects were randomly grouped to complete the role cognition test and human-computer interaction tasks. Data were collected by using the research path combining situation simulation and questionnaire survey. The psychological mechanism and decision-making logic of the public's adoption behavior were analyzed through the data analysis results. [Results/Conclusions] The research findings are as follows: 1) There is a significant interaction effect between the social roles and business types of government digital humans. In approval service scenarios, the decision-maker role is more capable of promoting public adoption behavior than the advisor role; 2) Human-computer trust perception plays a crucial mediating role in the influence path of social roles on the public's adoption behavior, revealing the core value of the trust mechanism in human-computer interaction; 3) The synergy effect between role authority and task fit constitutes an important mechanism influencing public cognition. This study expands the explanatory boundary of the social role theory in the field of intelligent government services and provides theoretical support for the construction of smart government services. However, there are still certain limitations. The service scenario simulation in the experimental design is difficult to fully restore the complexity of real government services. Future research can extend the multi-dimensional role classification system and deepen the mechanism exploration by combining the mixed research method. We have verified the applicability of the theoretical model in real government service scenarios and expand the existing conclusions. In addition, exploration on the dynamic impact of long-term interaction between government digital humans and the public on behavioral evolution is also a potential research direction.

Substitution and Overlap: The Relationship Between User Digitization and Service Accessibility in Public Cultural Participation | Open Access
WANG Yuanming, WANG Xueli
2025, 37(2):  61-74.  DOI: 10.13998/j.cnki.issn1002-1248.25-0065
Asbtract ( 62 )   HTML ( 14)   PDF (629KB) ( 115 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] The ongoing digital transformation has led to significant changes in public cultural services, particularly in content generation, communication channels, and modes of public participation. "Accessibility," a key indicator of the extent to which citizens' cultural rights are realized, is typically assessed along four dimensions: availability, acceptability, accessibility, and adaptability. Previous research has focused primarily on the supply side of accessibility, examining how factors such as the distribution of cultural resources, infrastructure development, and policy support affect user engagement. However, with the widespread adoption of digital technologies, individuals' ability and willingness to access information, utilize services, and provide feedback - collectively referred to as "digital literacy" - has become an increasingly important variable influencing cultural participation. Consequently, this study seeks to explore the relationship between users' digital literacy and the accessibility of public cultural services from a demand-side perspective. It aims to provide a more systematic theoretical framework and practical approach to optimizing the effectiveness of public cultural services. [Methods/Process] This study assesses users' digital literacy by examining their level of digital access, Internet usage, and service availability based on data collected from the Beijing-Tianjin-Hebei region. A structured questionnaire yielded 892 valid responses. To analyze the relationship between users' digital literacy and the accessibility of public cultural services, the study applies a generalized ordered logit model. A generalized ordered logit model is employed to analyze the substitution and overlap effects between users' digital literacy and the various dimensions of service accessibility. [Results/Conclusions] There is currently a digital divide exists between different demographic groups. A significant substitution effect is observed between traditional public cultural accessibility and users' digital literacy, with limited overlap between the two. Digitization has driven the modernization of public cultural resources and services, particularly in terms of technology and service delivery. However, there remains a time lag between the users' digital literacy of users and the digital transformation of the public cultural supply side. This lag suggests that the digital needs of users and the availability of digital cultural services are not fully aligned, which negatively impacts the effectiveness of public cultural services. Therefore, enhancing users' digital literacy, especially improving their ability to adapt to digital cultural resources, is a crucial factor in transitioning public cultural services from "accessibility" to "enjoyment". In promoting the digital upgrading of public cultural services, greater emphasis should be placed on developing users' capabilities and anticipating their needs.

Evolutionary Game Research on Stakeholder Decision-Making in Preprint Platform | Open Access
LI Huimiao, NIU Xiaohong, MA Zhuo, GUO Mohan
2025, 37(2):  75-87.  DOI: 10.13998/j.cnki.issn1002-1248.25-0088
Asbtract ( 24 )   HTML ( 5)   PDF (1674KB) ( 25 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] With the globalization of knowledge sharing and the vigorous development of preprint at home and abroad, the role of preprint platform in academic exchange has been recognized and appreciated by academic community. This paper introduces the evolutionary simulation method for the first time from the previous research on government and enterprises to the research on preprint platform, and takes the three main stakeholders in the construction of preprint platform, that is government, researcher and the public as the main players of the game. Different from the existing research, this paper uses system dynamics theory and software to fill the gap in quantitative analysis of SD model, and combines qualitative and quantitative research to further enrich the research content of the preprint platform through game model construction and simulation analysis. This research aims to guide stakeholders to actively participate in the construction of preprint platform, improve the utilization rate of domestic preprint platform by users, and promote the construction of preprint platform in China. [Method/Process] This study established a tripartite evolutionary game model of "government, researcher, and the public" to analyze the strategic stability of the three stakeholders. Vensim PLE software was used to simulate and analyze the the SD model, focusing on the influence of mixed strategies and external sensitivity variables on stakeholders' decision behavior. [Results/Conclusions] In the construction of preprint platform, the willingness of government supervision is mainly influenced by the supervision cost and credibility Within a reasonable range, the higher the scientific research funding for researchers or the more severe the penalty for their passive participation, the greater the willingness of researchers to participate actively. The public's willingness to cooperate is influenced by the costs of participation and the social dividends. In the future, the construction of the preprint platform can be continuously promoted from three perspectives: formulating the framework of the underlying reward and penalty mechanism of the preprint platform, establishing the reputation evaluation mechanism of researchers, and accelerating the construction of the government's open scientific innovation service. However, due to the limitation of the author's professional ability, the cognition of the preprint platform and the consideration of the relevant policy establishment process are relatively limited. There are many stakeholders involved in the construction of preprint platform, and there are also many factors that can affect the decision-making behavior in the external environment and system. In this study, three stakeholders from three main aspects are selected to model and study the external influence. In the future, we can select stakeholders from different angles and increase the influencing factors to expand the research on preprint platform.

A Study of the Factors Influencing Information Literacy on Farmers' Willingness to Participate in E-Commerce: the Mediating Effects of Perceived Benefits and Perceived Risks, and the Moderating Effect of Government Support | Open Access
MA Ning, ZHANG Xiaoyi, LIU Qixuan
2025, 37(2):  88-99.  DOI: 10.13998/j.cnki.issn1002-1248.25-0119
Asbtract ( 30 )   HTML ( 6)   PDF (762KB) ( 36 )  
Figures and Tables | References | Related Articles | Metrics

[Purpose/Significance] Promoting the digital transformation of agricultural product circulation through e-commerce has become a crucial way for rural revitalization in China. For three consecutive years, China's No. 1 Central Document has listed the high-quality development of agricultural e-commerce as a priority for upgrading the level of rural industrial development. However, persistent disparities in information literacy and imbalance in risk-benefit perceptions among farmer groups constrain the effective popularization of e-commerce platforms for agricultural products. To address this issue, this study integrates the Theory of Planned Behavior (TPB) and the perceived benefit-risk theory to construct a conceptual framework. It explores the relationship pathways among information literacy, perceived risks, perceived benefits, government support, and farmers' willingness to participate in e-commerce, aiming to provide theoretical insights for governments and enterprises to deepen the high-quality development of agricultural e-commerce business and rural revitalization. [Method/Process] Based on the above background, this paper integrates and proposes a conceptual model that includes the relationship of five potential variables: information literacy, perceived benefits, perceived risks, government support and engagement intention, based on the theory of planned behaviour and the theory of perceived benefits-perceived risks. In order to ensure the appropriateness of the sample distribution as well as the convenience, authenticity and reliability of the data collection, this study used a combination of online (WeChat group of village committees) and offline (recruiting home-based university students for field survey) to collect questionnaires from farmers across the country, and a total of 730 valid farmers' sample data were collected. Finally, based on the above data, the direct paths of perceived benefits, perceived risks and farmers' information literacy on farmers' willingness to participate in agricultural e-commerce were analyzed using Partial Least Squares Structural Equation Modelling (PLS-SEM). We focus our analysis on verifying the mediating roles of perceived benefits and perceived risks, as well as the moderating role of government support, in enhancing farmers' willingness to participate in agricultural e-commerce. [Results/Conclusions] The findings reveal that increased information literacy strengthens farmers' willingness to engage in agricultural e-commerce. Most farmers prefer participation scenarios with high perceived benefits and low perceived risks, where government support plays a key role in endorsing and leading trust. In this regard, local governments should establish tiered training systems and risk-hedging mechanisms (e.g., agricultural insurance, logistics subsidies) to address age-specific demand for information literacy improvement in rural areas and mitigate operational risks. We suggest actively publicizing national high-quality rural e-commerce demonstration cases and improving the perception of benefits to motivate farmers to participate, so as to achieve the high-quality development of agricultural e-commerce in a multi-initiative way. In addition, future research should pay more attention to the breadth of the sample coverage and the depth of the sample research process, and consider using all offline field research to further examine the impact of regional differences and the differences in the digital characteristics of the new farmers (Generation Z) on their willingness to participate in e-commerce. This will provide empirical evidence and guidance for rural revitalization and high-quality development of agricultural e-commerce.