Tao Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Tao Zhang has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
Lei Jiang
School of Information Management, Heilongjiang University, Harbin, China
4 shared publicationsTaihua Huang
School of Information Management, Heilongjiang University, Harbin, China
3 shared publicationsYing Qi
Heilongjiang University, Harbin, China
3 shared publicationsHaiqun Ma
School of Information Management, Heilongjiang University
2 shared publications- Pengxuan Zhang
School of Information Management, Heilongjiang University
2 shared publications Xiaoli Liu
School of Economics and Management, Northeast Agricultural University, Harbin, China
2 shared publications
Latest Publications
- 30 September 2026
- Published by SAGE Publications in Journal of Information Science
DPBERT: A domain-specific pre-trained model for Chinese data policy texts and its interpretability
Chinese data policy texts are characterised by dense terminology, standardised expressions and complex domain-specific semantics, which pose challenges for general-purpose language models. This study develops Data-Policy BERT (DPBERT), a domain-specific pre-trained model tailored to Chinese data policy texts. Using Chinese-bidirectional encoder representations from transformer-whole-word masking as the backbone, we conduct continued pre-training on 65,498 Chinese data-related policy documents with two masking strategies: masked language modelling and whole-word masking. The resulting models are evaluated on policy text classification, named entity recognition and an interpretability analysis based on gradient-based saliency. Experimental results show that DPBERT-whole-word masking outperforms the baseline models on both downstream tasks, indicating that whole-word masking is better suited to capturing Chinese policy terms and composite concepts. We further construct an interpretability evaluation framework using gradient-based saliency and a feature salient value metric to examine how models attend to core policy tokens. DPBERT-whole-word masking exhibits more concentrated attention on high-saliency policy features. DPBERT and Large Language Model Meta AI are compared under the same Chinese data, fine-tuning procedure and evaluation metrics; the results indicate that DPBERT achieves better performance on policy text classification and named entity recognition within this study’s task scope. This work provides a reference for semantic modelling, entity recognition and interpretable analysis of Chinese data policy texts.
- 30 September 2026
- Published by SAGE Publications in Journal of Information Science
DPBERT: A domain-specific pre-trained model for Chinese data policy texts and its interpretability
Chinese data policy texts are characterised by dense terminology, standardised expressions and complex domain-specific semantics, which pose challenges for general-purpose language models. This study develops Data-Policy BERT (DPBERT), a domain-specific pre-trained model tailored to Chinese data policy texts. Using Chinese-bidirectional encoder representations from transformer-whole-word masking as the backbone, we conduct continued pre-training on 65,498 Chinese data-related policy documents with two masking strategies: masked language modelling and whole-word masking. The resulting models are evaluated on policy text classification, named entity recognition and an interpretability analysis based on gradient-based saliency. Experimental results show that DPBERT-whole-word masking outperforms the baseline models on both downstream tasks, indicating that whole-word masking is better suited to capturing Chinese policy terms and composite concepts. We further construct an interpretability evaluation framework using gradient-based saliency and a feature salient value metric to examine how models attend to core policy tokens. DPBERT-whole-word masking exhibits more concentrated attention on high-saliency policy features. DPBERT and Large Language Model Meta AI are compared under the same Chinese data, fine-tuning procedure and evaluation metrics; the results indicate that DPBERT achieves better performance on policy text classification and named entity recognition within this study’s task scope. This work provides a reference for semantic modelling, entity recognition and interpretable analysis of Chinese data policy texts.
- 4 September 2026
- Published by Emerald Publishing in Reference Services Review
- p. 1-22
A metaverse literacy assessment framework for immersive library information services in the generative AI era
Purpose: As virtual reality (VR) services expand in libraries and increasingly intersect with generative artificial intelligence (AI), questions of user preparedness and competency gaps have become critical to the sustainability of immersive information services. This study aims to develop and empirically test a metaverse literacy assessment framework to evaluate the competencies required for participation in VR- and AI-enabled library information services. Design/methodology/approach: Drawing on information literacy, digital literacy, algorithm literacy and AI literacy, the study constructs a framework comprising four primary dimensions and 13 secondary indicators through literature analysis, the Delphi method and the analytic hierarchy process (AHP). The framework is then validated using survey data from 453 university faculty members and students. Reliability and validity tests, the entropy method and non-parametric tests are employed to examine structural robustness, indicator salience and group differences. Findings: The results show that knowledge, skills and norms play a central role in differentiating competency levels for immersive information environments. Tool knowledge, evaluation capability, ethical literacy and risk-related indicators emerge as particularly salient. Significant differences are found across educational and disciplinary groups, suggesting that competency gaps in VR- and AI-enabled information services are unevenly distributed and require targeted educational interventions. Originality/value: This study repositions metaverse literacy as a service-oriented competency framework for libraries rather than a purely technological construct. It offers a quantitative assessment tool for identifying user preparedness and competency gaps in immersive information services and provides practical guidance for libraries designing onboarding, instruction and risk-mitigation strategies for VR services in the generative AI era.
- 4 September 2026
- Published by Emerald Publishing in Reference Services Review
- p. 1-22
A metaverse literacy assessment framework for immersive library information services in the generative AI era
Purpose: As virtual reality (VR) services expand in libraries and increasingly intersect with generative artificial intelligence (AI), questions of user preparedness and competency gaps have become critical to the sustainability of immersive information services. This study aims to develop and empirically test a metaverse literacy assessment framework to evaluate the competencies required for participation in VR- and AI-enabled library information services. Design/methodology/approach: Drawing on information literacy, digital literacy, algorithm literacy and AI literacy, the study constructs a framework comprising four primary dimensions and 13 secondary indicators through literature analysis, the Delphi method and the analytic hierarchy process (AHP). The framework is then validated using survey data from 453 university faculty members and students. Reliability and validity tests, the entropy method and non-parametric tests are employed to examine structural robustness, indicator salience and group differences. Findings: The results show that knowledge, skills and norms play a central role in differentiating competency levels for immersive information environments. Tool knowledge, evaluation capability, ethical literacy and risk-related indicators emerge as particularly salient. Significant differences are found across educational and disciplinary groups, suggesting that competency gaps in VR- and AI-enabled information services are unevenly distributed and require targeted educational interventions. Originality/value: This study repositions metaverse literacy as a service-oriented competency framework for libraries rather than a purely technological construct. It offers a quantitative assessment tool for identifying user preparedness and competency gaps in immersive information services and provides practical guidance for libraries designing onboarding, instruction and risk-mitigation strategies for VR services in the generative AI era.
- 22 April 2026
- Published by Springer Nature in Education and Information Technologies
- Vol. 31  (13) , 1-28
How AI teaching assistant use affects students’ learning outcomes: An empirical study on the differential effects of questioning strategies
- 22 April 2026
- Published by Springer Nature in Education and Information Technologies
- Vol. 31  (13) , 1-28
How AI teaching assistant use affects students’ learning outcomes: An empirical study on the differential effects of questioning strategies
- Open Access
- 26 November 2025
- Published by IMR Press in KNOWLEDGE ORGANIZATION
- Vol. 52  (7)
Tracing the Topic Evolution of AI Literacy Literature: From Knowledge Enlightenment to Knowledge Symbiosis
Artificial intelligence (AI) is profoundly reshaping production modes, lifestyles, and organizational paradigms. Enhancing individual adaptability and creativity in this intelligent era has emerged as an imperative research agenda within public literacy studies. This study investigates the evolution of themes in AI literacy research, with a focus on the evolutionary mechanisms through which the domain’s knowledge architecture transitions from techno-cognitive accumulation to synergistic innovation ecosystems. Employing text-mining methodology, we analyze academic literature spanning 2015–2024. Following standard processing steps—tokenization, stopword removal, and lemmatization—we apply Latent Dirichlet Allocation (LDA) topic modeling to extract and analyze latent topics, tracing paradigm shifts in knowledge production during socio-technical processes of innovation. Evolutionary pathways constructed through topic similarity metrics reveal a progression in AI literacy research from knowledge enlightenment to symbiotic knowledge development. LDA2vec hybrid modeling facilitates deep semantic exploration of stage-specific topics, while subsequent topic intensity evolution and semantic network analyses highlight breakthrough points in AI literacy-driven knowledge innovation. This dual-axis analytical framework—integrating topic evolution with intensity dynamics—provides empirical evidence for predicting disciplinary trajectories while establishing novel theoretical paradigms and methodological approaches to AI literacy. Aligned with knowledge organization principles, these frameworks emphasize systematic structuring, cross-domain classification, and semantic interrelations of AI literacy concepts to enhance knowledge accessibility and interoperability.
- 26 November 2025
- Published by IMR Press in KNOWLEDGE ORGANIZATION
- Vol. 52  (7)
Open AccessTracing the Topic Evolution of AI Literacy Literature: From Knowledge Enlightenment to Knowledge Symbiosis
Artificial intelligence (AI) is profoundly reshaping production modes, lifestyles, and organizational paradigms. Enhancing individual adaptability and creativity in this intelligent era has emerged as an imperative research agenda within public literacy studies. This study investigates the evolution of themes in AI literacy research, with a focus on the evolutionary mechanisms through which the domain’s knowledge architecture transitions from techno-cognitive accumulation to synergistic innovation ecosystems. Employing text-mining methodology, we analyze academic literature spanning 2015–2024. Following standard processing steps—tokenization, stopword removal, and lemmatization—we apply Latent Dirichlet Allocation (LDA) topic modeling to extract and analyze latent topics, tracing paradigm shifts in knowledge production during socio-technical processes of innovation. Evolutionary pathways constructed through topic similarity metrics reveal a progression in AI literacy research from knowledge enlightenment to symbiotic knowledge development. LDA2vec hybrid modeling facilitates deep semantic exploration of stage-specific topics, while subsequent topic intensity evolution and semantic network analyses highlight breakthrough points in AI literacy-driven knowledge innovation. This dual-axis analytical framework—integrating topic evolution with intensity dynamics—provides empirical evidence for predicting disciplinary trajectories while establishing novel theoretical paradigms and methodological approaches to AI literacy. Aligned with knowledge organization principles, these frameworks emphasize systematic structuring, cross-domain classification, and semantic interrelations of AI literacy concepts to enhance knowledge accessibility and interoperability.
- Open Access
- 25 November 2025
- Published by Springer Nature in Humanities and Social Sciences Communications
- Vol. 12  (1)
The interrelationship of key factors affecting the learning outcomes of liberal arts majors: an empirical study
Through the deep integration of digital technologies such as large language models and knowledge graphs into pedagogy, this study explores the impact of technology use on learners’ academic outcomes to inform effective application throughout the teaching process. In this study, 123 undergraduate students majoring in International Chinese Language Education were subjected to pre- and post-tests, with their learning process tracked via a SPOC platform. By integrating Adaptive Structuration Theory (AST) and the Unified Theory of Acceptance and Use of Technology (UTAUT), the analysis of comprehensive learner factors yielded key findings. Learning behavior and self-efficacy positively impacted outcomes for students with low initial skills, yet self-efficacy and pre-learning skill level themselves did not directly determine outcomes. Critically, while the breadth of technology application played a significant mediating role, its overall level of use exerted a negative moderating effect on the relationship between learning behavior and final scores. This indicates that technology acts as a double-edged sword, capable of facilitating learning processes while also potentially undermining the effectiveness of positive learning behaviors when used excessively or without strategic focus.
- 25 November 2025
- Published by Springer Nature in Humanities and Social Sciences Communications
- Vol. 12  (1)
Open AccessThe interrelationship of key factors affecting the learning outcomes of liberal arts majors: an empirical study
Through the deep integration of digital technologies such as large language models and knowledge graphs into pedagogy, this study explores the impact of technology use on learners’ academic outcomes to inform effective application throughout the teaching process. In this study, 123 undergraduate students majoring in International Chinese Language Education were subjected to pre- and post-tests, with their learning process tracked via a SPOC platform. By integrating Adaptive Structuration Theory (AST) and the Unified Theory of Acceptance and Use of Technology (UTAUT), the analysis of comprehensive learner factors yielded key findings. Learning behavior and self-efficacy positively impacted outcomes for students with low initial skills, yet self-efficacy and pre-learning skill level themselves did not directly determine outcomes. Critically, while the breadth of technology application played a significant mediating role, its overall level of use exerted a negative moderating effect on the relationship between learning behavior and final scores. This indicates that technology acts as a double-edged sword, capable of facilitating learning processes while also potentially undermining the effectiveness of positive learning behaviors when used excessively or without strategic focus.
- 1 October 2025
- Published by Springer Nature in Scientometrics
- Vol. 130  (10) , 5375-5400
Research on classification prediction method based on gradient saliency text feature extraction: an empirical study on Web of Science categorization
- 1 October 2025
- Published by Springer Nature in Scientometrics
- Vol. 130  (10) , 5375-5400
Research on classification prediction method based on gradient saliency text feature extraction: an empirical study on Web of Science categorization
- Open Access
- 18 June 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15  (12) , 6865
Research on the Evolutionary Pathway of Science–Technology Topic Associations: Discovering Collaborative and Symmetrical Effects
This study employs text mining techniques to conduct a systematic quantitative analysis of cybersecurity-related scientific publications and technological research. It aims to break through the limitations of traditional unidirectional evolutionary research, reveal the knowledge evolution rules between scientific theories and technical practices in this field, and provide valuable references and decision-making support for optimizing the collaborative innovation ecosystem. Firstly, we took academic papers and patent research on cybersecurity from 2005 to 2024 as the research objects and divided them into ten stages according to the time series. Subsequently, we identified scientific and technological topics and formed science–technology topics to assess their similarity. Then, we selected 3040 pairs of collaborative topic pairs and categorized them into three distinct groups: weak, moderate, and strong correlation. Finally, we constructed a science–technology topic association evolution atlas and analyzed the types of evolutionary pathways of topic associations and their mechanisms of action accordingly. The results demonstrate five evolutionary patterns in science–technology topic associations: division, merging, inheritance, co-occurrence, and independent development. Additionally, the science–technology topics demonstrate a high degree of collaboration, exhibiting a collaborative effect of “initial accumulation–fluctuating differentiation–deep collaboration”. Meanwhile, the correlation evolution of strongly related science–technology topics presents a symmetrical effect of “technology–science–technology” and “science–technology/technology–science”.
- 18 June 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15  (12) , 6865
Open AccessResearch on the Evolutionary Pathway of Science–Technology Topic Associations: Discovering Collaborative and Symmetrical Effects
This study employs text mining techniques to conduct a systematic quantitative analysis of cybersecurity-related scientific publications and technological research. It aims to break through the limitations of traditional unidirectional evolutionary research, reveal the knowledge evolution rules between scientific theories and technical practices in this field, and provide valuable references and decision-making support for optimizing the collaborative innovation ecosystem. Firstly, we took academic papers and patent research on cybersecurity from 2005 to 2024 as the research objects and divided them into ten stages according to the time series. Subsequently, we identified scientific and technological topics and formed science–technology topics to assess their similarity. Then, we selected 3040 pairs of collaborative topic pairs and categorized them into three distinct groups: weak, moderate, and strong correlation. Finally, we constructed a science–technology topic association evolution atlas and analyzed the types of evolutionary pathways of topic associations and their mechanisms of action accordingly. The results demonstrate five evolutionary patterns in science–technology topic associations: division, merging, inheritance, co-occurrence, and independent development. Additionally, the science–technology topics demonstrate a high degree of collaboration, exhibiting a collaborative effect of “initial accumulation–fluctuating differentiation–deep collaboration”. Meanwhile, the correlation evolution of strongly related science–technology topics presents a symmetrical effect of “technology–science–technology” and “science–technology/technology–science”.
- 9 January 2025
- Published by SAGE Publications in Journal of Librarianship and Information Science
- Vol. 58  (1) , 346-362
Research on digital literacy literature: Topic mining and evolutionary analysis
The paper employs text analysis techniques to conduct a systematic quantitative analysis of literature in the field of digital literacy, aiming to recognize and organize the research topics and their evolutionary process within this field. The study selected literature data on digital literacy from the Web of Science database between 2004 and 2023 as research subjects, dividing them into 10 stages. Subsequently, the LDA2vec method was used to recognize research topics in the field of digital literacy, and the evolutionary process of these topics was analyzed from two aspects: the evolution of topic content and the evolution of topic intensity. The results indicate that scholars in the field of digital literacy primarily focus on research topics such as education, digital technology, digital practice, and cybersecurity. The research process of these topics can be summarized into three stages: exploration, practice, and expansion. In terms of the evolution of topic content, the identified types of evolution include division, merging, inheritance, disappearance, and generation. In the aspect of topic intensity evolution, it was found that changes in topic intensity are influenced by factors such as policy, research content, and the primary field of study. In the future, the topic of education and digital technology will remain the focus of research in the field of digital literacy. At the same time, the field will concentrate on conducting in-depth research into social topics such as health, aging society, and the challenges and risks associated with digital transformation.
- 9 January 2025
- Published by SAGE Publications in Journal of Librarianship and Information Science
- Vol. 58  (1) , 346-362
Research on digital literacy literature: Topic mining and evolutionary analysis
The paper employs text analysis techniques to conduct a systematic quantitative analysis of literature in the field of digital literacy, aiming to recognize and organize the research topics and their evolutionary process within this field. The study selected literature data on digital literacy from the Web of Science database between 2004 and 2023 as research subjects, dividing them into 10 stages. Subsequently, the LDA2vec method was used to recognize research topics in the field of digital literacy, and the evolutionary process of these topics was analyzed from two aspects: the evolution of topic content and the evolution of topic intensity. The results indicate that scholars in the field of digital literacy primarily focus on research topics such as education, digital technology, digital practice, and cybersecurity. The research process of these topics can be summarized into three stages: exploration, practice, and expansion. In terms of the evolution of topic content, the identified types of evolution include division, merging, inheritance, disappearance, and generation. In the aspect of topic intensity evolution, it was found that changes in topic intensity are influenced by factors such as policy, research content, and the primary field of study. In the future, the topic of education and digital technology will remain the focus of research in the field of digital literacy. At the same time, the field will concentrate on conducting in-depth research into social topics such as health, aging society, and the challenges and risks associated with digital transformation.
- RetractionOpen Access
- 13 August 2024
- Published by Springer Nature in Journal of The Institution of Engineers (India): Series B
- Vol. 106  (4) , 1349
Retraction Note: Research on Hot Topic Recognition and Its Evolution Analysis Method Based on LDA
- 13 August 2024
- Published by Springer Nature in Journal of The Institution of Engineers (India): Series B
- Vol. 106  (4) , 1349
RetractionOpen AccessRetraction Note: Research on Hot Topic Recognition and Its Evolution Analysis Method Based on LDA
- 17 January 2024
- Published by SAGE Publications in Journal of Librarianship and Information Science
- Vol. 57  (2) , 390-401
How to measure the interdisciplinarity of library and information science courses? Based on empirical research at the University of North Carolina at Chapel Hill
The article proposes interdisciplinary indicators suitable for measuring and evaluating the interdisciplinarity of Library and Information Science (LIS) curriculum at the University of North Carolina at Chapel Hill, with the aim of addressing the problems that arise in interdisciplinary education through suggestions. The interdisciplinarity of the curriculum is simulated on the Coursera platform and the interdisciplinary indicators are adopted to measure the interdisciplinarity of the disciplines and the curriculum. The results of the interdisciplinary simulations show a wide range of disciplinary knowledge and a high degree of interdisciplinarity. The interdisciplinary process of the curriculum shows that certain courses have a significant impact on the number of discipline categories. The comprehensive evaluation results for the three measures are consistent, indicating that the impact of the interdisciplinary measures varies from course to course. This study aims to provide constructive ideas for interdisciplinary curriculum design and talent cultivation in the field of Library and Information Science. In addition, the measurement and evaluation method can provide a reference for other disciplines.
- 17 January 2024
- Published by SAGE Publications in Journal of Librarianship and Information Science
- Vol. 57  (2) , 390-401
How to measure the interdisciplinarity of library and information science courses? Based on empirical research at the University of North Carolina at Chapel Hill
The article proposes interdisciplinary indicators suitable for measuring and evaluating the interdisciplinarity of Library and Information Science (LIS) curriculum at the University of North Carolina at Chapel Hill, with the aim of addressing the problems that arise in interdisciplinary education through suggestions. The interdisciplinarity of the curriculum is simulated on the Coursera platform and the interdisciplinary indicators are adopted to measure the interdisciplinarity of the disciplines and the curriculum. The results of the interdisciplinary simulations show a wide range of disciplinary knowledge and a high degree of interdisciplinarity. The interdisciplinary process of the curriculum shows that certain courses have a significant impact on the number of discipline categories. The comprehensive evaluation results for the three measures are consistent, indicating that the impact of the interdisciplinary measures varies from course to course. This study aims to provide constructive ideas for interdisciplinary curriculum design and talent cultivation in the field of Library and Information Science. In addition, the measurement and evaluation method can provide a reference for other disciplines.