Artificial intelligence in digital media, humanities, and information science: a multidimensional analysis of research trends and user perceptions

Abstract

The integration of Artificial Intelligence (AI) in digital media, humanities, and information science has rapidly transformed scholarly communication, digital knowledge management, and academic information retrieval. Despite significant advances, limited research explores AI’s potential in academic libraries, digital repositories, and scholarly publishing. This study examines research trends and user perceptions of AI in knowledge dissemination and management, addressing a gap in understanding its impact. This study maps global research trajectories and explores scholars’ views on the importance of AI in Library and Information Science (LIS) through a mixed-method approach combining bibliometric and web-based sentiment analysis. The key finding highlights that AI most significantly impacts automated metadata indexing, citation analysis, and AI-driven recommendation systems. The study calls for creating interdisciplinary collaboration, bettering AI transparency, and finding solutions to the ethical issues of bias and privacy protection. This research helps form the responsible integration of AI in LIS and digital knowledge systems.

Introduction

Artificial intelligence (AI) technology has progressed rapidly in the past few years and it has received attention worldwide. AI started to evolve from theoretical research to practical applications that took over many fields such as Machine Learning (Sivapalaratnam, 2019; Deo, 2024), Natural Language Processing (Mehta and Devarakonda, 2018; Hassan et al., 2021), Computer Vision (Ade, 1988; Kutlugun and Eyupoglu, 2020), Speech Recognition (Duan et al., 2021) and Automated Decision making (Ning et al., 2022). The improvement of these technologies not only changed the way various industries function but also influenced library science, digital archives, knowledge management, and scholarly communication. Innovations like automated metadata indexing, AI-driven recommendation systems, intelligent data retrieval in digital repositories, and AI-supported research evaluation (scientometrics and bibliometrics) are transforming how information is organized, accessed, and disseminated within the academic ecosystem. Given these advancements, it is essential to consider their implications for the management of digital knowledge in Library and Information Science (LIS).

As the usage of digital platforms for access to information continues to grow, there is a need to explore how AI can enhance the management of digital knowledge in libraries, archives, and academic information systems. Past studies have systematically investigated the application of AI in areas such as commercial digital media and social media (Xu et al., 2024; Lu and Nam, 2021), yet there is a lack of study on the part of AI’s opportunities in transforming digital libraries, academic repositories, and scholarly publishing. Specifically, the following areas have not yet been fully explored by these studies:

• The lack of holistic, mixed-method studies combining quantitative bibliometric data and qualitative user sentiment.

• The underrepresentation of academic repositories and digital libraries in AI discourse.

• The need for updated perspectives on how AI impacts knowledge dissemination processes from both systemic and user-centered viewpoints.

Therefore, this study aims to address the research gap by combining dynamic and static methods, integrating AI with academic repositories and digital libraries, and updating our understanding of AI’s impact on knowledge dissemination.

The majority of previous research on AI research has typically relied on bibliometric analysis (Di Vaio et al., 2020; Mostafa et al., 2023), an approach that reveals research trends and hotspots by counting and analyzing existing academic literature (Goodell et al., 2021). Bibliometric research offers essential quantitative tools to the academic community. It reveals research trends and directions, assesses academic influence, analyzes interdisciplinary collaboration, supports decision-making and resource allocation, and systematically examines literature structure. (Donthu et al., 2021). It enhances the understanding and management of scientific activities while fostering interdisciplinary cooperation and knowledge dissemination. Bibliometric analyses are conducted based on structured academic data, which does not necessarily encompass the most recent knowledge evolution nor emerging concerns related to AI in the field of knowledge management. To fill this gap, web-based text analysis and sentiment analysis have been adopted by researchers to offer a more dynamic and qualitative perception of the impact of AI on the academic community, researchers, and information professionals (Lin et al., 2022; Yang et al., 2024). Web-based text including social media content, blogs, and forum discussions (Patil and Kulkarni 2018). Furthermore, with the aid of natural language processing technology, text content can be sentiment-analyzed (Zhang et al., 2018), revealing public emotional tendencies and attitudes toward specific topics, uncovering hidden associations and patterns in the textual data, and assisting researchers in discovering new research areas or hot topics (Ravi and Ravi, 2015).

This study employs a combination of bibliometric analysis and an examination of how AI standards and practices are currently recorded, taught, and applied in the design and implementation of new technologies. Through bibliometric techniques, we map research trends, collaboration networks, and innovative developments driven by AI in both AI research and the applications within publishing and LIS. This approach draws on the framework established by Donthu et al. (2021). Instead, web-based text analysis examines the introduction of AI into scholarly communication, digital preservation, and automated research processes among LIS professionals, researchers, and users of digital knowledge as a source of emerging concerns and areas of interest (Lin et al., 2022). Further, the sentiment analysis shows how scholars, librarians, and researchers understand AI’s contribution and challenge to knowledge management and digital information retrieval (Zhang et al., 2018).

This study combines quantitative bibliometric methods in digital knowledge management, scholarly information retrieval, and academic communication with qualitative text analysis to give a complete image of the evolving role of AI technology. Combining bibliometrics with sentiment analysis allows us to gain deeper insights into research trends and attitudes in academic fields, highlighting researchers’ perspectives on specific topics and their impact. This approach offers a more comprehensive perspective, surpassing the insights provided by any single method, and illustrating the complexity of research dynamics. The results intend to inform debates on the ethical questions in the use of AI-related innovations for LIS such as transparency, bias, and privacy protection in AI-powered knowledge dissemination systems. Since AI is now going to mold the future of academic libraries, digital repositories, and research networks, understanding the effect of AI in the information science and humanities disciplines is important to chart and integrate it.

Materials and methods

Data collection

To analyze the evolution of the current research of AI in the digital media, humanities, and information science, we performed bibliometric analysis using the Web of Science (WOS) Core Collection which is considered to be a widely known scholarly database featuring high-quality academic publications (Tao et al., 2020; Bircan and Salah, 2022). An approach to using this system was developed to allow us to systematically analyze research trends; knowledge structure and scholarly contribution for the impact of AI on digital libraries, academic publishing, and applications of AI in LIS.

A literature search was conducted in WOS using the keywords TS = (“artificial intelligence” AND “media” AND “information”). This initial search yielded 1477 articles. To ensure the integrity and relevance of our dataset, we applied the following inclusion criteria:

• Publication Years: The search was conducted on August 7, 2024, and the results showed that the earliest publication dates back to 1995, thus defining the publication time range from 1995 to 2024.

• Quick Filters (Exclude): Early Access, Enriched Cited References, Open publisher-invited reviews

• Document Types (Exclude): Early Access, Editorial Material, Retracted Publication, Letter, Book Review, Meeting Abstract

• Language Filtering: Only English-language publications were retained to facilitate accessibility and standardization in analysis.

  After applying the above filters, we identified 978 data entries. Guided by the core theme of “AI-driven transformation of the information ecology, as manifested in LIS, scholarly communication, digital knowledge management, and related media fields,” we systematically screened the initial 978 records. This approach allowed us to investigate how AI reshapes information flows, knowledge production, and communication patterns across these interconnected domains.

• Inclusion Criteria: The core discussion of the literature should ideally address changes prompted by AI technology, involving the evolution of information ecosystems, the restructuring of knowledge activities, or the intelligent transformation of interactions among people, information, and environment.

• Exclusion Criteria: Studies whose contributions were primarily limited to pure technical improvements in AI algorithms or models, without substantive exploration of their implications within the aforementioned information ecology or social practices; Works that mentioned AI or related domains only peripherally, without placing their interplay or AI’s driving role at the center of the discussion.

To finalize the literature selection, we jointly reviewed all screened articles to verify their relevance and ensure methodological consistency with our research objectives.

Upon thorough review, 460 articles were deemed suitable. (see Fig. 1), which was exported into a Microsoft Excel (Appendix 1.xlsx) file for further bibliometric processing. Citation data extraction including publication counts, citation networks, and author collaborations, were collected from WOS for analysis.

[See PDF for image]

Fig. 1

Data collection and processing.

To complement the bibliometric insights, we conducted web-based text analysis on scholarly discussions and professional discourse related to AI’s role in digital knowledge management, information retrieval, and academic research workflows. We searched Zhihu (https://www.zhihu.com) and CSDN (https://www.csdn.net) using the keywords “AI and digital information science” in August, 2024, as these platforms have significant influence in China by uniting technical developers and scholars, while their user-reviewed content ensures reliability and offers grassroots expert opinions alongside the experiences of technical practitioners, thereby enriching the research perspective of bibliometric data.

Using Python’s requests package, we made requests to the target website while employing the time package to control the crawling speed. We utilized regular expressions from the re module to extract the relevant parts of the Uniform Resource Locator (URL) and applied urllib.parse.unquote to decode the URL strings. Ultimately, we stored the crawled content in a Microsoft Excel file using the pandas package, successfully retrieving 30 relevant text transcripts from discussions on AI’s role in academic publishing, library automation, and LIS professional networks.

Bibliometric and data processing

Bibliometric analysis and visualization were conducted using VOSviewer and Scimago Graphica software. These tools were selected for their user-friendly interfaces and robust capabilities in visualizing complex bibliometric data. VOSviewer excels in mapping and analyzing bibliometric networks and is particularly effective in handling large datasets (Donthu et al., 2021), while Scimago Graphica is well-suited for mapping global research trends and providing a clear, intuitive visual representation of scholarly relationships (Hassan-Montero et al., 2022). It allowed us to identify key areas of research, groups of interest, and emerging topics of AI-driven Knowledge Management, Knowledge Communication, and Knowledge Retrieval.

To contribute to the bibliometric findings, we then analyzed 30 online discussions and professional discussions of AI in LIS, in academic publishing and digital archives (see Fig. 1). In the visual analysis of text keywords, this study first performed text preprocessing using Python’s NLTK library (Panigrahi and Asha 2019), which included sentence segmentation, word tokenization, and the removal of stop words by loading a stopwords.txt file to eliminate meaningless words such as “we,” “will,” and “one.” Next, we clarified the crawled content by employing regular expressions to remove punctuation marks, spaces, brackets, and other extraneous elements. We then utilized the Jieba package for additional word segmentation. Finally, we employed Python’s WordCloud tool to graphically represent the core vocabulary within the corpus, configuring the parameters of the WordCloud package, calling the generate function, and generating a word cloud using the matplotlib module. This step guaranteed that AI in LIS, knowledge dissemination, and academic metadata management terms will frequently occur and capture the key themes related to AI in LIS, knowledge dissemination, and academic metadata management.

After performing text cleaning, we used Python’s Panda library to calculate word frequency statistics to quantify the most common terminologies and thematic trends in scholarly discourse on AI and digital knowledge ecosystems.

To investigate sentiment dynamics in academic discussions, we utilized VADER sentiment analysis (a lexicon-based measure designed for social and textual data) to measure (score) 310 discussion sentences on a scale from 1 to -1. Then classed to broaden general sentiment into five categories.

• Extreme Positive (≥0.5)—Strongly favorable perspectives on AI’s role in LIS and scholarly communication.

• Moderate Positive (0.5 > score ≥ 0.1)—Generally optimistic but with some cautious considerations.

• Balanced conversations with both the positive and negative aspects (0.1 > score ≥ -0.1).

• The score is Moderate Negative (−0.1 > score ≥ −0.5): Concerns about what AI can do in knowledge management.

• Extreme Negative (< −0.5)—Directed at ethical risks, bias, and automation challenges.

For visualizing the sentiment distribution and thematic clustering, we used Matplotlib charts which allowed to understand in an intuitive way how academic community perceives AI’s role in digital library, research repository, and automated scholarly indexing. In this way, these visualizations showed additional contextual information regarding how the academic polemic surrounding AI is seen in LIS and digital humanities.

Results

Trends in the number of published articles based on bibliometric analysis

In the past couple of years, developments in AI technology and the importance of digital knowledge management, scholarly communication, and academic information retrieval have increased exponentially AI-related research publications in LIS, digital humanities, and scholarly publishing. This trend has been further accelerated by the diversification of academic channel of dissemination, e.g., digital repositories, open access platforms and index system driven by AI technologies.

We then grouped the publication trends in three definable stages based on their evolution of AI in LIS and connected fields to examine how AI has developed over time (Fig. 2):

[See PDF for image]

Fig. 2

Trends in AI-Related LIS Research Publications (1995–2024): Annual and Cumulative Analysis.

1. First Stage (1995-2009): Initial Exploration and Slow Growth

   • Only two articles were published in 1995, marking the early theoretical phase of AI applications in LIS.

   • From 1995 to 2009, publication rates remained low and sporadic, with some years seeing no publications at all.

   • This phase reflects the limited adoption of AI in library sciences, archives, and scholarly communication systems before significant advancements in machine learning and natural language processing (NLP).

2. Second Stage (2010–2014): Emergence and Increased Interest

   • A notable surge in AI-related LIS publications occurred in 2010, accounting for 58.73% of publications in this phase.

   • By 2014, the number of published articles reached 49, comprising 10.65% of total indexed publications in the dataset.

   • This period corresponds to the adoption of AI-powered tools in digital libraries, knowledge organization systems, and citation analysis frameworks, reflecting the growing intersection between AI and LIS research.

3. Third Stage (2015–2024): Rapid Expansion and Maturity

   • A significant increase in research output has been observed, with 73.91% of the total indexed publications falling within this period.

   • The peak in 2019-2022 aligns with the rise of deep learning models, AI-driven recommendation systems for libraries, and automated metadata indexing.

   • However, a slight decline in publication growth was noted in 2023, possibly indicating a shift towards applied AI solutions over foundational research.

   • As of August 7, 2024, 34 articles have been published within the first 8 months, suggesting that AI’s role in LIS research remains an active and evolving area.

This bibliometric analysis highlights the increasing importance of AI in LIS, digital knowledge management, and academic publishing, demonstrating how AI technologies are reshaping library automation, research impact assessment, and digital archives.

Hot spots and trends in the literature based on bibliometric analysis

The analysis of key research themes in AI applications in LIS, digital media, and scholarly communication provides insights into the evolution of research priorities over time. Authors often emphasize the core focus of their studies through keywords, enabling the tracking of emerging research trends and thematic shifts (Liu et al., 2022). The top 20 most frequently occurring keywords across the three stages of AI in LIS research are presented in Table 1, reflecting changing research interests and technological advancements.

Table 1. Top 20 most frequently used authors’ keywords: a global and temporal analysis.

NO.	First stage (1995–2010)	F	Second stage (2011–2014)	F	Third stage (2015–2024)	F
1	virtual reality	3	classification	3	artificial intelligence	97
2	interactive narrative	2	data mining	3	machine learning	25
3	planning	2	feature selection	2	new media	20
4	advanced man-machine interfaces	1	privacy	2	social media	15
5	artistic installations	1	clustering	1	deep learning	14
6	authoring tools	1	data publication	1	big data	12
7	behavioral systems	1	data publishing	1	covid-19	8
8	camera control	1	density-based	1	new media art	8
9	character animation	1	ensemble methods	1	virtual reality	8
10	digital technologies	1	location-based service	1	human-machine communication	7
11	fully immersive viewing systems	1	membership function	1	China	6
12	grammars	1	multi-label learning	1	algorithms	5
13	immersive entertainment	1	random forest	1	deepfakes	5
14	immersive environments	1	sample reduction	1	generative ai	5
15	interaction techniques	1	task oriented privacy	1	human-computer interaction	5
16	interactive design	1	user behavior	1	media	5
17	interactive storytelling	1	-	-	surveillance	5
18	interactivity	1	-	-	algorithm	4
19	knowledge-based systems	1	-	-	automation	4
20	motion capture	1	-	-	communication	4

F frequency.

Phase one (1995–2010): early foundations and conceptual development

The analysis of 63 publications in this phase yielded 150 author keywords, of which 144 appeared only once (see Fig. 3a). During the early 2002s, research in LIS-related AI applications focused on “interactive design”, addressing “virtual environments,” “immersive environments,” and “grammars” as key concepts. By 2004, the thematic scope expanded to include “interactive narratives”, emphasizing AI-driven content structuring in digital libraries and knowledge repositories. Around 2006, the focus shifted towards “virtual reality”, marking the beginning of AI-driven simulation and knowledge visualization tools. By 2008, research emphasis moved towards “planning,” particularly in the context of library automation and digital knowledge curation. By 2010, the primary focus was on “immersive entertainment”, incorporating “behavioral systems,” “synthetic characters,” and “motion capture”, demonstrating an early stage of AI applications in user-centered digital information systems.

[See PDF for image]

Fig. 3

The evolution of artificial intelligence and new media research topics across three key periods.

a Phase 1 (2002–2010): distribution and frequency of author keywords, highlighting the early thematic focus on interactive design, virtual reality, and planning. b Phase 2 (2011–2014): keyword analysis showing the expansion into data science, with emerging trends in location-based services, data mining, and classification. c Phase 3 (2015–2024): dominant research themes and high-frequency keywords reflecting the mature, AI-driven transformation in LIS, centered on machine learning, deep learning, and generative AI.

Phase two (2011–2014): expansion into data science and information retrieval

In the second phase, 57 articles contributed a total of 151 keywords (see Fig. 3b). Around 2012, research predominantly centered on “location-based services,” with a growing emphasis on user behavior analysis in digital knowledge platforms. This reflects the increasing integration of AI in digital libraries, research repositories, and scholarly indexing tools. By 2013, “data mining” emerged as the leading research trend, highlighting AI’s role in automated information retrieval, bibliometric analysis, and metadata management. By 2014, the research emphasis transitioned to “classification”, focusing on “feature selection” and “privacy”, which aligns with AI’s growing application in information security, knowledge categorization, and access control in digital repositories.

Phase three (2015–2024): maturity and AI-driven transformations in LIS

The third phase, spanning 2015–2024, included 340 publications, with 1317 keywords identified, of which 48 appeared more than three times (see Fig. 3c). Around 2021, research themes were strongly centered on “digital media” and “scholarly communication,” integrating AI-driven research analytics, knowledge dissemination platforms, and social network analysis in LIS. By mid-2021, a significant shift occurred towards “machine learning” and “human-computer interaction,” reflecting AI’s increasing role in automated reference management, AI-powered digital librarianship, and predictive analytics for academic publishing trends. From 2022 onwards, “artificial intelligence” became the most dominant research trend, closely related to “machine learning,” “digital knowledge repositories,” and “automated scholarly indexing.” By May 2022, the primary research themes involved “deep learning” and “big data,” both extensively applied to bibliometric analysis, AI-driven citation tracking, and research impact measurement systems. Finally, by 2023, LIS research on AI emphasized “human-machine communication,” “generative AI,” “algorithms,” and “chatbots”, illustrating the growing role of AI-driven virtual assistants, recommendation engines, and intelligent search functionalities in LIS research infrastructures.

Key insights from bibliometric trend analysis

This bibliometric analysis highlights the evolution of AI applications in LIS from conceptual foundations to large-scale automated knowledge systems. While early research (1995–2010) focused on AI-assisted classification and content curation, later stages (2011-2014) saw an expansion into data-driven information retrieval and privacy-preserving AI models. The most recent phase (2015–2024) reflects a rapid shift toward machine learning, deep learning, and AI-powered scholarly information systems, shaping the future of digital libraries, open-access knowledge repositories, and AI-driven academic publishing analytics.

Country publications based on bibliometric analysis

Analyzing the geographical distribution of AI-related research in LIS, digital media, and humanities provides crucial insights into global research trends, key contributing regions, and international collaboration networks. This bibliometric analysis helps comprehend how different nations participate in AI-powered developments in scholarly communication, digital repositories, and academic information retrieval. The integration of AI in LIS research varies across regions, and identifying research hotspots helps inform strategic academic partnerships and interdisciplinary collaboration in digital knowledge management.

The analysis of 460 AI-related LIS research papers, covering 59 countries (see Fig. 4), reveals that China leads with 143 publications (31.09%), reflecting its strong emphasis on AI applications in digital libraries, knowledge management, and academic indexing systems. The USA comes in next with 76 publications (16.52%), which underscores its achievements in the field of AI scholarly communication, automated metadata indexing, and citation analytics tools development. Canada (54, 11.74%) and Germany (40, 8.70%) also emerge as key contributors, with research focused on AI-enhanced bibliometric studies, digital repositories, and AI-based academic information retrieval systems.

[See PDF for image]

Fig. 4

Geographical distribution of AI-related publications in Library and Information Science (LIS) from 1995 to 2024.

International research collaborations play a crucial role in AI’s expansion in LIS, with China and the USA demonstrating the strongest co-authorship networks, followed by USA-UK and USA- Korea collaborations. These partnerships indicate a growing effort to integrate AI in LIS research, specifically in AI-driven digital knowledge ecosystems, automated research impact assessments, and intelligent library systems. Based on co-authorship characteristics, countries were grouped into six distinct research clusters. The first cluster, led by China, includes Canada, the United Kingdom, and Australia, reflecting strong collaboration in AI applications for academic libraries, knowledge organizations, and open-access repositories. The second cluster which is Germany’s sphere of interest with Austria, Japan, and Netherlands focuses on AI augmented library classification, bibliometric indicator analysis, and data mining through AI in LIS. The third cluster captures the interests of the USA along with India and Poland, focusing on knowledge dissemination, citation network intelligence, and research impact analytics through AI.

These findings depict the global trend for the use of AI technologies in LIS, media studies, and humanities. International partnerships will be crucial in the development of future AI-aided knowledge engineering, digital archiving, and scholarly communication systems. The wide regions where AI is used in LIS reflect the readiness of the global community for the effective application of AI in library automation, digital preservation, and academic information retrieval. The circumscribed research allocation of AI has to be dealt with carefully as it increases the responsibility of policymakers, academic institutions, and researchers to encourage more interdisciplinary collaboration and the application of AI-based technologies in the LIS fields of study.

The map shows how many times a country has published around the world and visualizes the international collaboration networks. Dotted lines indicate the co-authorship and research partnerships between the major contributing countries with larger node sizes being indicative of higher research output. The findings of this analysis elaborate the worldwide effect of AI in digital libraries, scholarly publishing and knowledge management.

Sources, institution, and author analysis based on bibliometric analysis

To understand the scholarly landscape of AI applications in LIS, digital media, and humanities, we conducted a bibliometric analysis of 460 research publications, examining their article sources, institutional affiliations, and contributing authors. These publications were categorized into five thematic clusters, reflecting key research areas in AI-driven scholarly communication, digital library systems, and bibliometric analysis.

The bibliometric coupling analysis identified 262 publication sources, with 165 falling within five primary categories (see Fig. 5a). The top 10 sources collectively published 172 articles, accounting for 37.39% of all indexed publications (see Table 2). Among these, the Journal of LIS emerged as the most significant venue, followed by Scientometrics and Digital Library Perspectives, reflecting the growing academic interest in AI-enhanced scholarly communication, digital curation, and research evaluation. The presence of three major conference proceedings among the top ten sources indicates an increasing emphasis on AI’s role in LIS-focused knowledge dissemination, research impact assessment, and scholarly indexing systems.

[See PDF for image]

Fig. 5

Circular cluster distribution of publications in the fields of artificial intelligence and new media.

a Sources: Bibliographic coupling network of the top publication venues (e.g., journals, conferences). b Organizations: Collaborative network and research output of leading contributing institutions. c Authors: Co-authorship network and productivity clusters of the most active researchers.

Table 2. The top 10 publication sources, organizations and authors for artificial intelligence and new media.

Categories	NO.	Name	NP	NTC	NAC
Sources (262)	1	New Media & Society	57 (33.14%)	1102	19.33
	2	Advances In Artificial Intelligence, Canadian AI 2014	43 (25%)	393	9.14
	3	Smart Graphics, Proceedings	32 (18.60%)	205	6.41
	4	Convergence-The International Journal Of Research Into New Media Technologies	15 (8.72%)	183	12.20
	5	Wireless Communications & Mobile Computing	7 (4.07%)	13	1.86
	6	Hci International 2022—Late Breaking Papers: Interaction In New Media, Learning And Games	5 (2.91%)	1	0.20
	7	Ieee Access	4 (2.33%)	12	3.00
	8	Artnodes	3 (1.74%)	8	2.67
	9	Public Relations Review	3 (1.74%)	54	18.00
	10	Scientific Reports	3 (1.74%)	3	1.00
Organizations (574)	1	Seoul Natl Univ	15 (27.78%)	104	6.93
	2	Natl Univ Singapore	10 (18.52%)	135	13.50
	3	Sungkyunkwan Univ	6 (11.11%)	63	10.50
	4	Seoul Media Inst Technol	4 (7.41%)	24	6.00
	5	Texas Tech Univ	4 (7.41%)	13	3.25
	6	Chinese Univ Hong Kong	3 (5.56%)	6	2.00
	7	MIT	3 (5.56%)	2706	902.00
	8	Northwestern Univ	3 (5.56%)	10	3.33
	9	Univ Oxford	3 (5.56%)	36	12.00
	10	Univ So Calif	3 (5.56%)	10	3.33
Authors (1300)	1	Shukla, Jainendra (IIIT-Delhi)	6	38	6.33
	2	Yoon, Sungroh (Seoul Natl Univ)	6	17	2.83
	3	Fu, Guohong (Soochow Univ)	5	64	12.80
	4	Shah, Rajiv Ratn (IIIT-Delhi)	5	12	2.40
	5	Zhang, Meishan (Tianjin Univ)	5	71	14.20
	6	Appel, Markus (Univ Wurzburg)	4	32	8.00
	7	Jo, Jeonghee (Seoul Natl Univ)	4	4	1.00
	8	Li, Xuanya (Baidu Inc)	4	12	3.00
	9	Liu, An-An (Tianjin Univ)	4	12	3.00
	10	Natale, Simone (Loughborough Univ)	4	111	27.75

NP represents the number of publications; NTC is the total number of citations; NAC refers to the average number of citations per publication. Percentages in parentheses indicate the share within the top 10. The author is followed by their institution in parentheses.

The 460 indexed publications were associated with 80 institutions, highlighting a global and interdisciplinary research network (see Fig. 5b). Seoul National University led with 15 publications (27.78%), followed by MIT with 3 publications and Hanyang University with 2. Seoul National University and Hanyang University demonstrated strong research collaborations in AI-driven digital libraries, semantic search technologies, and intelligent knowledge organization systems. The second leading category includes the National University of Singapore (10 publications), the Chinese University of Hong Kong (3), and the University of Oxford (3), contributing significantly to AI-driven citation analysis, automated metadata extraction, and predictive analytics for LIS research. The fourth category, led by Sungkyunkwan University (6 publications) and Northwestern University (3 publications), reflects interdisciplinary approaches to AI-powered digital repositories, knowledge retrieval, and academic profiling systems.

Regarding authorship patterns, 1300 authors contributed to the 460 indexed publications, with 815 classified into five key research clusters (see Fig. 5c). Among the top 10 most active authors, 40% were affiliated with Chinese institutions, while India and Korea each accounted for 20% of LIS-related AI research output. The leading contributors include Shukla Jainendra and Yoon Sungroh, with Natale Simone’s publications recording the highest citation impact (see Table 2). Additionally, corporate researchers such as Li Xuanya, representing AI-driven information services, contributed significantly to AI’s integration into LIS workflows. The strongest co-authorship collaborations were observed between Yoon Sungroh and Jo Jeonghee (first category—blue), Li Xuanya and Liu An-An (first category—blue), and Markus Appel (third category—yellow), emphasizing AI’s role in institutional knowledge systems. The second category (purple) is dominated by Shukla Jainendra, while Natale Simone (fifth category—green) contributed significantly to AI-enhanced scholarly impact assessments.

This bibliometric analysis emphasizes AI’s transformative role in scholarly publishing, digital curation, and knowledge management systems, emphasizing its global impact on AI-driven research retrieval, automated indexing, and intelligent recommendation systems within LIS, digital humanities, and academic communication. The findings highlight the growing significance of interdisciplinary collaborations in AI-driven LIS advancements, with institutions and researchers contributing to machine learning applications in citation analysis, AI-powered library automation, and predictive analytics for academic impact assessment.

A thematic comparison of bibliometric and web-based text analysis

The bibliometric and web-based text analysis conducted from 1995 to August 2024 identified 1586 keywords, among which 61 appeared more than three times (see Fig. 6a). The analysis categorized these keywords into five distinct thematic clusters, reflecting the evolution of AI applications in LIS, digital media, and scholarly communication.

[See PDF for image]

Fig. 6

Thematic evolution based on comparative analysis.

a Bibliometric analysis (1995–August 2024): co-occurrence network of keywords derived from the literature, clustered into five primary research themes (e.g., artificial intelligence, digital repositories, deep learning). b Web-based text analysis (August 2024): high-frequency terms extracted from contemporary online discourse, highlighting current public and professional focus areas related to AI and new media.

The first cluster (red) is centered around “artificial intelligence,” which exhibits a strong co-occurrence relationship with “machine learning,” “digital knowledge management,” “automated indexing,” and “predictive analytics”. This highlights the integration of AI in scholarly communication, digital curation, and research evaluation. The second cluster (purple) focuses on “digital repositories” and “AI-powered knowledge organization”, with close ties to “human-computer interaction,” “automated information retrieval,” and “knowledge dissemination systems”. The third cluster (blue) is primarily centered on “AI in digital libraries”, showing strong co-occurrence with “machine learning,” “text analysis,” and “automated recommendation systems”. The fourth cluster (yellow) revolves around “deep learning”, with discussions on “training,” “neural networks,” and “bibliometric AI applications”, demonstrating AI’s growing role in citation analysis, scholarly profiling, and impact assessment in LIS research. The fifth cluster (green) is linked to “AI-driven scholarly publishing,” encompassing themes such as “digital preservation,” “automated academic indexing,” “privacy in digital knowledge ecosystems,” and “AI-assisted metadata tagging.”

A comparative analysis between bibliometric data and web-based text analysis (see Table 3; Fig. 6b) highlights key high-frequency terms, providing further insights into AI’s research trajectory in LIS. According to frequency statistics of the top 60 keywords from web-based analysis, the combination of “AI” (166) and “digital knowledge” (134) dominates scholarly discussions, aligning with bibliometric trends capturing similar high-frequency terms such as “content” (75), “data” (37), “internet” (30), and “information retrieval” (42). These findings reinforce the interdisciplinary nature of AI research in LIS, emphasizing its applications in digital repositories, knowledge classification, and academic publishing.

Table 3. Top 60 high frequency words statistics for AI and new media in online texts.

NO.	Words	F	NO.	Words	F	NO.	Words	F
1	AI	166	21	internet	19	41	graduates	11
2	digital knowledge	134	22	application	19	42	hand	11
3	content	75	23	time	17	43	platforms	11
4	technology	56	24	continuous	16	44	read	11
5	development	51	25	impact	16	45	learning	11
6	news	49	26	traditional	16	46	majors	11
7	writing	44	27	future	15	47	brand	11
8	information retrieval	42	28	improve	14	48	experience	10
9	media	37	29	people	14	49	model	10
10	data	37	30	journalism	14	50	generate	10
11	network	36	31	program	13	51	trend	10
12	intelligent	34	32	challenges	13	52	provide	10
13	internet	30	33	field	13	53	analysis	10
14	user	27	34	direction	12	54	employment	10
15	models	27	35	video	12	55	training	10
16	university	26	36	technologies	12	56	social	10
17	production	23	37	efficiency	12	57	human	10
18	professional	21	38	major	12	58	enterprises	10
19	users	21	39	opportunities	12	59	realize	10
20	new	21	40	difficult	11	60	addition	10

F frequency.

Additionally, as detailed in Table 3, new emerging keywords from web-based discussions reflect evolving concerns and expectations regarding AI’s role in LIS research and digital transformation. Key terms such as “technology” (56), “development” (51), “future” (15), “challenges” (13), “graduates” (11), “majors” (12), and “employment” (10) indicate ongoing discussions about the future of AI-driven digital information systems, skill development for LIS professionals, and the ethical challenges of AI integration in scholarly publishing.

This combined bibliometric and text analysis approach highlights AI’s progressive role in LIS research and digital knowledge management. Early bibliometric trends primarily emphasized AI’s applications in information retrieval and classification, whereas recent themes focus on AI-enhanced scholarly communication, automated indexing, and predictive citation analytics. The convergence of machine learning, digital libraries, and AI-assisted metadata organization illustrates the growing demand for intelligent knowledge systems in LIS research. As AI technologies continue to evolve, their integration into bibliometric analysis, academic publishing, and research impact assessment will shape the future of digital knowledge ecosystems.

Sentiment analysis based on network data

Sentiment analysis provides valuable insights into academic and professional discourse regarding AI’s impact on LIS, scholarly communication, and digital knowledge management. By analyzing network data from scholarly discussions and professional forums, we can assess how LIS researchers, information professionals, and academic institutions perceive AI’s role in digital transformation (Song et al., 2021).

The sentiment analysis results indicate that neutral sentiment (score = 0) appeared most frequently, totaling 55 instances (see Fig. 7a), suggesting that a significant portion of LIS discussions remain objective regarding AI’s implications. Sentiment scores of 0.4767 and 0.4588 appeared 13 and 12 times, respectively, within the mildly positive range, reflecting moderate support for AI’s integration into LIS research workflows, digital repositories, and citation analysis tools.

[See PDF for image]

Fig. 7

Sentiment analysis of discourse on AI and new media.

a Top ten frequency statistics of sentence scores: distribution of the most frequently occurring sentiment scores within the analyzed text data. b Share ratio of five sentiment categories: pie chart showing the proportion of text segments categorized as Positive Extreme (PosExt), Positive Moderate (PosMod), Neutral (Neu), Negative Moderate (NegMod), and Negative Extreme (NegExt).

The overall sentiment distribution indicates a predominant preference for positive sentiment, with 43.55% categorized as extremely positive and 23.55% as moderately positive (see Fig. 7b). These findings suggest that most LIS researchers and professionals acknowledge AI’s transformative potential in digital libraries, automated research indexing, and bibliometric evaluations. Meanwhile, 21.94% of the discussions remained neutral, reflecting a balanced perspective on both AI’s advantages and challenges in LIS applications.

Although negative sentiment was minimal, moderate negative views accounted for 9.03%, and extreme negative perceptions represented 1.94%, indicating some concerns related to AI-driven scholarly communication, ethical considerations, and automation challenges in LIS research. These concerns likely stem from issues such as AI bias in metadata tagging, transparency in citation metrics, and potential employment disruptions in digital libraries due to AI automation.

The predominantly positive sentiment toward AI’s role in LIS, academic publishing, and digital knowledge management highlights the widespread recognition of AI’s benefits in scholarly indexing, automated information retrieval, and impact assessment. The small proportion of negative sentiment reflects concerns about AI’s ethical implications, transparency in scholarly evaluation, and data privacy in AI-driven research workflows. As AI technologies continue to evolve, ongoing discussions in LIS research communities will likely address these ethical considerations while refining AI’s role in knowledge organization and scholarly publishing.

Discussion

The impact of artificial intelligence on the volume of publications in library and information science (LIS)

LIS, digital knowledge management, and (scholarly) communication are research areas in which the research in AI has expanded in recent decades. Similar to Kankanhalli MS (2003) research on multimedia synthesis technology, this study contributed to theoretical development through empirical sampling and analogical reasoning. The study’s findings show that the application of AI in LIS during the first phase (1995–2010) was in its early stages, characterized by slow growth and a focus on “interactive narrative” and “planning”. This slow progress may have been due to the technological limitations of the era. As academic knowledge speaks through numbers and digital storage increases, researchers started investigating the world of AI in digital information systems, which is similar to the research conducted by other scholars (Sala, 2002; Kassel et al., 2005). This may be due to the fact that early research mainly concentrated on AI experiments related to knowledge classification, semantic search, and metadata extraction within the field of LIS.

The second phase, 2011 and 2014 ushered in a growing interest in AI as the practical applications materialized, especially through machine learning and NLP techniques, to use them in LIS. During this period, research publications progressed steadily, demonstrating the part that AI plays in automated library cataloging, digital knowledge repositories, and AI-assisted citation analysis. This is akin to how machine learning, particularly convolutional neural networks, is applied in medicine, driving research advancements and resulting in consistent improvements. (Batch et al., 2013; Chu and Krzyzak, 2014). In addition, researchers examined the use of AI to develop user interaction models for digital libraries (developing recommendation and search functionalities). International co-work also develops, speeding up technological advancement and spreading knowledge to the larger population, Molnar’s (2014) study also makes reference to this. Therefore, by 2014, the highest number of AI-related LIS publications was reflected, indicating the fast advancement of AI applications in digital knowledge systems and academic impact measurement tools.

In the third phase (2015–2024) of AI research related to LIS, the number of publications increased rapidly, echoing the trend in previous research. Lundberg et al. (2024) examined how machine learning can be applied in bibliometrics, highlighting its potential to enhance the efficiency of literature analysis. The findings from this phase not only advanced this direction but also encouraged the broader use of AI in areas like academic indexing, publishing, and digital preservation. Adigun and Igboechesi (2024) highlighted the importance of AI in intelligent knowledge retrieval and discovery, showing that AI technology can greatly enhance the accuracy and efficiency of information retrieval while also benefiting the assessment of academic influence. In contrast to other studies, this research mainly showcases the swift development and application of AI in LIS, illustrating a gradual transition from theory to practice. This study also found that additional validation was achieved by examining AI in LIS journals and conferences. The community’s focus on AI’s role in scholarly communication and digital curation is evident in the prominence of AI-LIS research conferences and proceedings.

In this paper, bibliometric analysis is increasingly shown to emphasize the importance of AI in LIS, digital libraries, and scholarly communication concerning how AI technologies are changing processes of research workflow, metadata management, and digitized knowledge systems. However, if only English-language literature is used for the analysis when screening documents, important research findings in other languages may be overlooked, which would limit the comprehensiveness of the research (Politzer-Ahles et al., 2020). In the next phase of LIS research, AI is likely to continue evolving to facilitate the automation of scholarly indexing, enhance digital search functionalities, and improve citation metrics.

The impact of artificial intelligence on scholarly communication in library and information science (LIS)

AI applications are used in developing LIS, digital repositories, and academic publishing which has consequently shaped interdisciplinary research. Maltby et al. (2024) highlighted that integrating AI-driven knowledge management, automated information retrieval, and predictive analytics has opened up significant opportunities in the LIS field for discovery, academic exchange, and data-driven decision-making. Our research aligns with Maltby et al.’s findings, demonstrating how AI technology enhances information processing efficiency and academic exchange, likely due to the ongoing evolution and widespread adoption of AI. Furthermore, Rogers (1995) and Morie (2002) examined the early development of digital classification and virtual knowledge systems, while Ivanova et al. (2014), Estupinan et al. (2018), and Wang et al. (2024) provided further analysis on AI-driven information recommendation, interactive narrative analysis, and AI-processed metadata. These studies showcase the shift from traditional information management methods to the adoption of modern AI applications. While early research primarily focused on foundational information organization methods, this study highlights AI’s crucial role in boosting academic influence and optimizing information retrieval processes, likely driven by advancements in AI technology and improved data analysis capabilities.

Morie (2002) specializes in interactive design and virtual knowledge environments, creating worlds that leverage expectations, interests, and interactions with the natural world to construct “narratives”. This study, similar to Morie’s findings, notes that around 2002, early research on the application of AI in LIS also focused on “virtual environments”, The development of computer-assisted knowledge systems and human-computer interaction technologies led researchers to create AI-based metadata organization platforms, digital repositories, and knowledge dissemination systems. By 2006, the research focus shifted to “virtual reality”, which was connected to Schroeder’s (2006) research on shared virtual environments. Research on shared virtual environments laid the groundwork for integrating other data media research. By 2010, research shifted toward AI-driven academic indexing, with studies on “behavioral systems” and “motion capture” aligning with the increasing demand for automated citation analysis, behavioral analysis, and academic exchange platforms .

From 2011 to 2014, LIS research emphasized “location-based service” and “behavior analysis”, which relates to the increasing popularity of AI-driven personalization services for platforms and automated content recommendation systems, as studied by Kuehn (2013). By 2013, “data mining” emerged as a dominant research theme, due to a growing emphasis on big data analytics for scholarly impact assessments, research evaluation, and academic profiling systems, which is consistent with Lee’s (2022) research. The emergence of “classification” in 2014 raised concerns about AI’s role in digital information processing and research privacy protection, coinciding with the increasing importance of data security in AI-driven LIS platforms (Guo and Vargo, 2015; Kysela, 2015).

Between 2015 and 2024, research increasingly focused on AI in scholarly publishing, digital curation, and automated research impact measurement. Terms such as “machine learning,” “AI-powered digital repositories,” and “automated citation tracking” illustrated the deep integration of AI technologies in research assessment and digital knowledge organization. By 2021, AI had gained considerable influence in digital LIS platforms, enabling automated literature reviews, AI-enhanced academic content indexing, and predictive analytics for research trends, which is consistent with the changes in research topics analyzed by Feio and Oliveira (2024). In 2023, research emphasized “artificial intelligence” and “generative AI,” supported by deep learning and big data analytics, further advancing AI applications in metadata tagging, AI-powered knowledge classification, and intelligent research retrieval systems. Similar findings can also be found in Lee (2022) and Zhao (2024).

This bibliometric analysis underscores AI’s transformative role in scholarly publishing, digital knowledge curation, and LIS-driven academic impact assessment. The transition from early AI-assisted content classification (pre-2010) to automated citation indexing (2011–2014) and finally to AI-enhanced digital research evaluation (2015–2024) illustrates the increasing reliance on AI technologies for scholarly communication, academic profiling, and digital repository automation. As AI research in LIS progresses, future advancements will likely focus on AI ethics in scholarly publishing, transparency in citation tracking, and predictive modeling for research impact evaluation.

The impact of artificial intelligence on research collaboration in digital media information dissemination

Trends in the geographical distribution and international collaboration networks can be seen in research in digital media information, digital knowledge management, and scholarly communication. The increasing integration of AI technologies in academic research has led to the establishment of cross-institutional partnerships in many diverse fields including new media, social media, and interdisciplinary collaborations as well as global academic engagement.

Studies by Seol et al. (2012) and Woo et al. (2014) reveal that Seoul National University is a leader in applying AI to digital media information. Their collaboration with Hanyang University exemplifies how the Korean academic community is pioneering AI-driven knowledge systems. This aligns with our research, emphasizing the significance of inter-university collaboration in advancing AI applications, which can be attributed to the complementary resources and technologies among institutions. Meanwhile, Abadi et al. (2020) highlighted the financial investments in AI research in both China and the United States, illustrating the interconnectedness of institutions like MIT, the Chinese University of Hong Kong, and Tianjin University on an international scale. Although MIT publishes fewer papers on media applications compared to other top 15 institutions, its influence arises from many highly cited projects, which aligns with our research findings. This indicates that high-quality research impacts the academic community more significantly than sheer quantity, potentially improving contributions to academic exchanges and the efficiency of information management. It can be attributed to the institution’s continuous exploration and innovation in cutting-edge AI technology.

Molnar (2014) observed that the adoption of novel educational informatics methodologies, such as Web 2.0 and e-learning 2.0, within Hungarian and international collaborative practices, has facilitated intelligent positioning, big data analytics, and the implementation of emerging technologies. This trend parallels the collaborative research exemplified by Sungkyunkwan University and Northwestern University in this study. Both cases underscore the global integration of AI and digital technologies within education and academia. This convergence has driven innovation in learning environments and knowledge dissemination models, particularly in response to the evolving learning styles and needs of digital-native learners. Consequently, it provides new technological support and methodological frameworks for the efficient advancement of higher education. In applying AI to scholarly publishing and knowledge dissemination systems, Shukla Jainendra has worked on the individual level of AI-driven user interaction model for the human-computer interaction, while Yoon, Sungroh is engaged with AI application towards big data analytics, as mentioned in Newman (2002). The combined expertise of the two facilitates the technological support of integrating AI in digital repositories, automated tracking of citations, and digital media information dissemination.

Multiple studies highlight Natale Simone’s substantial contributions to AI theory and media convergence research (Lesage and Natale, 2019; Natale and Ballatore, 2020). This established body of work likely accounts for the high citation frequency of his publications observed in this study’s results. Furthermore, academic institutions and industry leader collaboration for advancing AI applications in information dissemination is becoming increasingly vital (Nie et al., 2016). Therefore, Li Xuna, a Baidu figurehead in AI-driven knowledge services, collaborated with Professor Liu An’an, an expert in multimedia research. This synergistic partnership effectively bridged theoretical knowledge and practical applications, catalyzing technological innovation and advancing media information dissemination. This finding aligns with the research of Shang et al. (2023).

The impact of integrating bibliometrics, web text and sentiment analysis in dynamic environments

This study combined bibliometric clustering analysis and network text analysis to discover the relationships and research trends on the topic of AI applications in LIS, digital knowledge management, and scholarly communication in a dynamic environment. Both methodologies yield results of a strong co-occurrence relationship between AI and other themes like scholarly indexing, machine learning, and predictive analytics for research evaluation. This finding signals the dramatic change in digital repositories, citation analysis, and academic knowledge organization systems caused by AI, which is similar to the research trends of Ma (2021) on human-computer interaction system data information under deep learning mechanisms. We hypothesize that this is due to AI altering the organization of knowledge and the processing of information, thereby driving changes in the academic field. Like many studies analyzing keyword frequency in cited literature to identify AI research trends, this study finds that ethical concerns—such as data privacy and algorithm transparency—are common in AI applications across various fields (Kysela, 2015; Feio and Oliveira, 2024). However, unlike earlier research, this study introduces more dynamic and diverse perspectives by analyzing texts from emerging social media platforms focused on online customers.

Among the high-frequency terms in text analysis is “content”, “data”, and “digital knowledge” which point out the problem in AI-driven content classification, automated research indexing, and credibility of AI-generated metadata in the digital library, which echoes the findings of Shahzad et al. (2022). In line with this study’s findings on network text analysis, Wu (2022) explored how AI and virtual reality can deliver realistic image experiences to users through interactive methods. Frequent terms like “technology” and “development” highlight AI’s role in improving academic search functions, boosting automated recommendation systems, and promoting digital knowledge dissemination. Furthermore, it identified terms like “future”, “challenges”, “university”, and “employment”, suggesting that traditional roles are being supplanted, resulting in shifts in the job market. This is consistent with Ma’s (2023) findings. Therefore, in today’s data-rich environment, LIS professionals should not confine themselves to traditional library storage systems when formulating policies for AI-driven digital knowledge systems, research funding, and career development pathways. Instead, they should incorporate digital media and digital humanities to unify diverse information and fulfill the objective of knowledge dissemination.

Sentiment analysis gives us a glimpse into what scholars, researchers, and LIS professionals believe is AI’s function concerning academic research and preservation in the digital world. The results show that the most frequent sentiment was neutral (score = 0) indicating that a considerable amount of scholarly research about AI in LIS research is objective, which is consistent with the findings of Machová et al. (2023). Most researchers admit AI can be used to automate dozens and more tasks in LIS, automated indexing, and research impact analysis, as evidenced by the majority of opinions being mildly positive. Nevertheless, the majority of studies also indicate that negative emotions exist (Ma, 2023; Ring, 2024; Cheng, 2024). These studies reveal that people are worried about the ethical implications of AI, see the need for greater scholarly transparency for AI impact assessments, and that AI may have a detrimental effect in the field of LIS job displacement.

Thematic analysis helps us comprehend the topics that followers discussed about the Dubai World Cup, and sentiment analysis enables us to analyze the tone of public sentiments related to the World Cup and the city of Dubai. Bibliometric and sentiment analysis together show that AI has two sides both as technical advancement in LIS research and ethical issues. Sentiment analysis of online texts fills a gap in bibliometrics regarding sentiment and attitude, which is an important area of research. Though AI-enabled automation of research retrieval, digitization of data contribution, and citation trace is improved, there are still concerns about the AI bias, transparency of academic profiling, and the requirement to achieve robust protection of user privacy in various domains of LIS application of AI. Additionally, we hypothesize that the differences between the results of bibliometric analysis and web text analysis stem from the fact that web text analysis captures dynamic, cutting-edge discussions reflecting practical issues and emerging concerns, whereas bibliometrics emphasizes stable academic research outcomes. Regarding ethical AI governance in LIS, future research should focus on ethical AI governance in LIS, how to further improve AI transparency in SC, and AI’s role in the career development of information professionals.

Conclusion

This research states that AI integration in LIS, scholarly communication, and digital knowledge management has transformed from initial theoretical conceptualization to extensive use. It has changed academic publishing, digital library automation, and research impact on assessment, facilitating the role of AI in knowledge organization optimization, bibliometric analysis, and automatic research retrieval systems.

Scholarly workflows are being redefined by AI technologies, facilitating intelligent metadata indexing, prediction of citation and review analytics, as well as AI-based recommendation systems in digital repositories. Although the academic community has generally recognized AI’s positive role in LIS, the problems of transparency, ethical governance of AI, and the transformation of the workforce have not been solved. This research should be followed up by addressing these challenges in areas like AI bias in scholarly indexing, privacy in digital archiving, and ethical issues of AI-driven automation when evaluating research.

Combining bibliometric and web-based text analysis, this study investigates the co-developed pathways of AI-related research with LIS. This approach uses structured bibliometric data in conjunction with real-time network text analysis to overcome the limitations of direct use of the bibliometric database and provides a dynamic point of view on how AI is shaping scholarly communication and LIS research trends. The research also brings new methodological support to AI knowledge management studies and opens the path to further research on digital library automation, research evaluation supported by AI, and ethical AI adoption in LIS workflows. However, this study has certain limitations. The restriction to English-language publications may exclude relevant research disseminated in other languages, potentially introducing a cultural or regional bias. Additionally, despite following a systematic protocol, the manual screening process remains susceptible to subjective judgment, which might have influenced the final selection of literature. Future research could address these constraints by incorporating multi-language literature sources and implementing more structured screening frameworks, such as multi-reviewer validation processes.

Author contributions

BP wrote the main manuscript text. BP prepared Figs. 1–7. DL conducted the data analysis and provided feedback on the manuscript. BP assisted in literature review and drafting sections of the introduction. All authors reviewed and approved the final manuscript.

Data availability

The data used to support the findings of this study are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Ethical statement

Ethical approval was not required for this study as it does not involve human participants, personal data, or experimental procedures.

Informed consent

No informed consent was needed for this paper. This article does not contain any studies with human participants performed by any of the authors.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1057/s41599-025-06372-9.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

Abadi, HHN; He, Z; Pecht, M. Artificial intelligence-related research funding by the US national science foundation and the national natural science foundation of China. IEEE Access; 2020; 8, pp. 183448-183459. [DOI: https://dx.doi.org/10.1109/ACCESS.2020.3029231]

Ade, F. Computer-vision and artificial-intelligence. Helv Phys Acta; 1988; 61, pp. 36-41.

Adigun, TA; Igboechesi, GP. Exploring the role of generative artificial intelligence in enhancing information retrieval and knowledge discovery in academic libraries. IJLISS; 2024; 10, pp. 1-14. [DOI: https://dx.doi.org/10.37745/ijliss.15/vol10n2114]

Batch, Y; Yusof, MM; Noah, SA. ICDTag: a prototype for a web-based system for organizing Physician-Written blog posts using a hybrid Taxonomy-Folksonomy approach. J Med Internet Res; 2013; 15, e41. [DOI: https://dx.doi.org/10.2196/jmir.2353] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23470419][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3636292]

Bircan, T; Salah, AAA. A bibliometric analysis of the use of artificial intelligence technologies for social sciences. Math-basel; 2022; 10, 4398. [DOI: https://dx.doi.org/10.3390/math10234398]

Cheng, I-H; Lee, ST. The impact of ethics instruction and internship on students’ ethical perceptions about social media, artificial intelligence, and ChatGPT. J Media Ethics; 2024; 39, pp. 114-129. [DOI: https://dx.doi.org/10.1080/23736992.2024.2333486]

Chu JL, Krzyzak A (2014) Analysis of feature maps selection in supervised learning using convolutional neural networks. In: Sokolova M, VanBeek P (eds) Advances in artificial intelligence, Canadian AI 2014. Springer-Verlag Berlin, Berlin, pp 59–70

Deo, RC. Artificial intelligence and machine learning in cardiology. Circulation; 2024; 149, pp. 1235-1237. [DOI: https://dx.doi.org/10.1161/CIRCULATIONAHA.123.065469] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38620085]

Di Vaio, A; Palladino, R; Hassan, R; Escobar, O. Artificial intelligence and business models in the sustainable development goals perspective: a systematic literature review. J Bus Res; 2020; 121, pp. 283-314. [DOI: https://dx.doi.org/10.1016/j.jbusres.2020.08.019]

Donthu, N; Kumar, S; Mukherjee, D et al. How to conduct a bibliometric analysis: an overview and guidelines. J Bus Res; 2021; 133, pp. 285-296. [DOI: https://dx.doi.org/10.1016/j.jbusres.2021.04.070]

Duan, R; Wang, Y; Qin, H. Artificial intelligence speech recognition model for correcting spoken English teaching. J Intell Fuzzy Syst; 2021; 40, pp. 3513-3524. [DOI: https://dx.doi.org/10.3233/JIFS-189388]

Estupinan S, Andkjaer KI, Szilas N (2018) Engagement in interactive digital storytelling: sampling without spoiling. In: Clua E, Roque L, Lugmayr A, Tuomi P (eds) Entertainment computing—ICEC 2018. Springer International Publishing AG, Cham, pp 248–253

Feio C, Oliveira L (2024) News consumption and political participation of young people in new media—a narrative literature review. Commun Rev. https://doi.org/10.1080/10714421.2024.2419693

Goodell, JW; Kumar, S; Lim, WM; Pattnaik, D. Artificial intelligence and machine learning in finance: identifying foundations, themes, and research clusters from bibliometric analysis. J Behav Exp Financ; 2021; 32, [DOI: https://dx.doi.org/10.1016/j.jbef.2021.100577] 100577.

Guo, L; Vargo, C. The power of message networks: a big-data analysis of the network agenda setting model and issue ownership. Mass Commun Soc; 2015; 18, pp. 557-576. [DOI: https://dx.doi.org/10.1080/15205436.2015.1045300]

Hassan-Montero Y, De-Moya-Anegón F, Guerrero-Bote VP (2022) SCImago Graphica: a new tool for exploring and visually communicating data. Prof Inf e310502. https://doi.org/10.3145/epi.2022.sep.02

Hassan T, Ahmad A, Anwar M, et al. (2021) Basic structure of natural language processing applied on multiple domains: review. In: 2021 International conference on innovative computing (ICIC). pp 1–7

Ivanova, MO; Couceiro, MS; Martins, FML. Merging the fields of swarm robotics and new media: perceiving swarm robotics as new media. Maejo Int J Sci Technol; 2014; 8, pp. 143-160.

Kankanhalli MS (2003) Multimedia analysis and synthesis. In: Gedeon TD, Fung LCC (eds) AI 2003: Advances in artificial intelligence. Springer-Verlag Berlin, Berlin, pp 41–52

Kassel S, Grebenstein K, Tittmann C (2005) A knowledge-based decision-support-system for e-commerce. In: AlAkaidi M, Rothkrantz L (eds) Euromedia ’2005: 11th annual Euromedia conference. Eurosis, Ghent, pp 146–150

Kuehn KM (2011) Prosumer-citizenship and the local: a critical case study of consumer reviewing on Yelp.com. Ph.D., The Pennsylvania State University

Kutlugun E, Eyupoglu C (2020) Artificial intelligence methods used in computer vision. In: 2020 5th international conference on computer science and engineering (UBMK). IEEE, New York, pp 214–218

Kysela J (2015) Analysis of privacy erosion of geosocial networks. In: 2015 16th IEEE international symposium on computational intelligence and informatics (CINTI). IEEE, New York, pp 199–202

Lee, D. Symbolic model of new media art expression based on artificial intelligence big data. Wirel Commun Mob Comput; 2022; 2022, pp. 1-10.2022CMaPh.390..1L [DOI: https://dx.doi.org/10.1155/2022/9630255]

Lesage, F; Natale, S. Rethinking the distinctions between old and new media: Introduction. Convergence; 2019; 25, pp. 575-589. [DOI: https://dx.doi.org/10.1177/1354856519863364]

Lin, P; Chen, L; Luo, Z. Analysis of tourism experience in Haizhu National Wetland Park based on web text. Sustainability; 2022; 14, 3011.2022Sust..14.3011L [DOI: https://dx.doi.org/10.3390/su14053011]

Liu, C; Li, W; Xu, J et al. Global trends and characteristics of ecological security research in the early 21st century: a literature review and bibliometric analysis. Ecol Indic; 2022; 137, [DOI: https://dx.doi.org/10.1016/j.ecolind.2022.108734] 108734.

Lundberg, L; Boldt, M; Borg, A; Grahn, H. Bibliometric mining of research trends in machine learning. AI; 2024; 5, pp. 208-236. [DOI: https://dx.doi.org/10.3390/ai5010012]

Lu, Z; Nam, I. Research on the influence of new media technology on internet short video content production under artificial intelligence background. Complexity; 2021; 2021, [DOI: https://dx.doi.org/10.1155/2021/8875700] 8875700.

Ma L (2021) Realization of artificial intelligence interactive system for advertising education in the era of 5G integrated media. Wireless Netw. https://doi.org/10.1007/s11276-021-02700-4

Ma, Y. AI-Based employment situation analysis of innovative talents in the new media era. J Multimed Inf Syst; 2023; 10, pp. 169-178. [DOI: https://dx.doi.org/10.33851/JMIS.2023.10.2.169]

Machová, K; Szabóova, M; Paralič, J; Mičko, J. Detection of emotion by text analysis using machine learning. Front Psychol; 2023; 14, 1190326. [DOI: https://dx.doi.org/10.3389/fpsyg.2023.1190326] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37799520][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10548207]

Maltby, J; Rayes, T; Nage, A et al. Synthesizing perspectives: crafting an Interdisciplinary view of social media’s impact on young people’s mental health. PLoS ONE; 2024; 19, e0307164.1:CAS:528:DC%2BB2cXitFOitbnK [DOI: https://dx.doi.org/10.1371/journal.pone.0307164] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39008509][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11249244]

Mehta, N; Devarakonda, MV. Machine learning, natural language programming, and electronic health records: The next step in the artificial intelligence journey?. J Allergy Clin Immun; 2018; 141, pp. 2019-2021.e1. [DOI: https://dx.doi.org/10.1016/j.jaci.2018.02.025] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29518424]

Molnar G (2014) Modern ICT Based teaching and learning support systems and solutions in higher education practice. In: Turcani M, Drlik M, Kapusta J, Svec P (eds) Divai 2014: 10th international scientific conference on distance learning in applied informatics. Wolters Kluwer Cr a S, Praha 3, pp 421–430

Morie JF (2002) Coercive narratives, motivation and role playing in virtual worlds. In: Callaos N, Bica M, Sanchez M (eds) 6th world multiconference on systemics, cybernetics and informatics, vol XII, Proceedings: industrial systems and engineering II. International Institute Informatics & Systemics, Orlando, pp 473–479

Mostafa, MM; Feizollah, A; Anuar, NB. Fifteen years of YouTube scholarly research: knowledge structure, collaborative networks, and trending topics. Multimed Tools Appl; 2023; 82, pp. 12423-12443. [DOI: https://dx.doi.org/10.1007/s11042-022-13908-7] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36157354]

Natale, S; Ballatore, A. Imagining the thinking machine: technological myths and the rise of artificial intelligence. Convergence; 2020; 26, pp. 3-18. [DOI: https://dx.doi.org/10.1177/1354856517715164]

Newman, J. In search of the videogame player—the lives of Mario. N Media Soc; 2002; 4, pp. 405-422. [DOI: https://dx.doi.org/10.1177/146144402320564419]

Nie, W; Liu, A; Zhu, X; Su, Y. Quality models for venue recommendation in location-based social network. Multimed Tools Appl; 2016; 75, pp. 12521-12534. [DOI: https://dx.doi.org/10.1007/s11042-014-2339-x]

Ning, H; Yin, R; Ullah, A; Shi, F. A survey on Hybrid Human-Artificial intelligence for autonomous driving. IEEE Trans Intell Transp Syst; 2022; 23, pp. 6011-6026. [DOI: https://dx.doi.org/10.1109/TITS.2021.3074695]

Panigrahi N, Asha T (2019) Aspect-Level sentiment analysis on hotel reviews. In: Behera HS, Nayak J, Naik B, Abraham A (eds) Computational Intelligence in data mining. Springer, Singapore, pp 379–389

Patil S, Kulkarni S (2018) Mining social media data for understanding students’ learning experiences using memetic algorithm. In: Materials today-proceedings. Elsevier Science Bv, Amsterdam, pp 693–699

Politzer-Ahles, S; Girolamo, T; Ghali, S. Preliminary evidence of linguistic bias in academic reviewing. J Engl Acad Purp; 2020; 47, [DOI: https://dx.doi.org/10.1016/j.jeap.2020.100895] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33088213][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7575202]100895.

Ravi, K; Ravi, V. A survey on opinion mining and sentiment analysis: tasks, approaches and applications. Knowl-Based Syst; 2015; 89, pp. 14-46. [DOI: https://dx.doi.org/10.1016/j.knosys.2015.06.015]

Ring, A. The internet is relational: networked connectivity and the “sovereign individual” in videos by brenda lien. Oxf Ger Stud; 2024; 53, pp. 58-77. [DOI: https://dx.doi.org/10.1080/00787191.2024.2315882]

Rogers, A. Virtual-reality—the new media. BT Technol J; 1995; 13, pp. 132-137.

Sala NMR (2002) New media, Internet, learning, and communication. In: Callaos N, Loutfi M, Justan M (eds) 6th world multiconference on systemics, cybernetics and informatics, vol viii, proceedings: concepts and applications of systemics, cybernetics and informatics II. International Institute Informatics & Systemics, Orlando, pp 529–534

Schroeder, R. Being there together and the future of connected presence. Presence; 2006; 15, pp. 438-454. [DOI: https://dx.doi.org/10.1162/pres.15.4.438]

Seol, H; Park, G; Lee, H; Yoon, B. Demand forecasting for new media services with consideration of competitive relationships using the competitive Bass model and the theory of the niche. Technol Forecast Soc Chang; 2012; 79, pp. 1217-1228. [DOI: https://dx.doi.org/10.1016/j.techfore.2012.03.002]

Shahzad, K; Khan, SA; Ahmad, S; Iqbal, A. A scoping review of the relationship of big data analytics with context-based fake news detection on digital media in data age. Sustainability; 2022; 14, 14365.2022Sust..1414365S [DOI: https://dx.doi.org/10.3390/su142114365]

Shang, J; Zhang, K; Liu, S. The relationship between knowledge potential matching and innovation performance of university-enterprise cooperation: the moderating effect of geographic proximity. Technol Anal Strateg Manag; 2023; 35, pp. 1278-1295. [DOI: https://dx.doi.org/10.1080/09537325.2021.2001453]

Sivapalaratnam, S. Artificial intelligence and machine learning in haematology. Brit J Haematol; 2019; 185, pp. 207-208. [DOI: https://dx.doi.org/10.1111/bjh.15774]

Song Z, Song K, Cheng N, et al. (2021) Research on internet public opinion recognition method based on high frequency co-occurrence. In: 2021 IEEE/ACIS 20th international fall conference on computer and information science (ICIS Fall). pp 237–242

Tao, J; Qiu, D; Yang, F; Duan, Z. A bibliometric analysis of human reliability research. J Clean Prod; 2020; 260, [DOI: https://dx.doi.org/10.1016/j.jclepro.2020.121041] 121041.

Wang, X; Cai, L; Xu, Y. Creation mechanism of new media art combining artificial intelligence and internet of things technology in a metaverse environment. J Supercomput; 2024; 80, pp. 9277-9297. [DOI: https://dx.doi.org/10.1007/s11227-023-05819-7]

Woo, J; Choi, JY; Shin, J; Lee, J. The effect of new media on consumer media usage: an empirical study in South Korea. Technol Forecast Soc Chang; 2014; 89, pp. 3-11. [DOI: https://dx.doi.org/10.1016/j.techfore.2014.09.001]

Wu, Y. Application of artificial intelligence within virtual reality for production of digital media art. Comput Intell Neurosci; 2022; 2022, pp. 1-10. [DOI: https://dx.doi.org/10.1155/2022/3781750]

Xu D, Feng S, Huang Z (2024) Positive and negative factors of the influence of new media and the digital environment on ideal and belief education among university students in China. J Gen Psychol. https://doi.org/10.1080/00221309.2024.2385105

Yang, J; Liu, X; Li, Z; Zhou, H. Group counterattack to destination infringement events: media empowerment and emotional conduction regarding interactive ritual theory. Curr Issues Tour; 2024; 27, pp. 2230-2249. [DOI: https://dx.doi.org/10.1080/13683500.2023.2224551]

Zhang, L; Wang, S; Liu, B. Deep learning for sentiment analysis: a survey. WIREs Data Min Knowl Discov; 2018; 8, [DOI: https://dx.doi.org/10.1002/widm.1253] e1253.

Zhao, Y. The synergistic effect of artificial intelligence technology in the evolution of visual communication of new media art. Heliyon; 2024; 10, [DOI: https://dx.doi.org/10.1016/j.heliyon.2024.e38008] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39328541][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11425173]e38008.