Next Article in Journal
Decision-Making and Contract Coordination of Closed-Loop Supply Chain with Dual-Competitive Retail and Recycling Markets
Previous Article in Journal
TimeWeaver: Time-Aware Sequential Recommender System via Dual-Stream Temporal Network
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Systems
Volume 13
Issue 10
10.3390/systems13100859
systems-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Mapping Impact and Influence in AI-Driven Advertising: A Scientometric and Network Analysis of Knowledge Ecosystem

by
Camille Velasco Lim
1 and
Han-Woo Park
2,*
1
Department of Media and Communications, YeungNam University, Gyeongsan-si 38541, Republic of Korea
2
Interdisciplinary Graduate Programs of Digital Convergence Business, East Asian Cultural Studies, Cyber Emotions Research Center, Big Local Big Pulse Lab, Department of Media and Communication, YeungNam University, Gyeongsan-si 38541, Republic of Korea
*
Author to whom correspondence should be addressed.
Systems 2025, 13(10), 859; https://doi.org/10.3390/systems13100859
Submission received: 1 July 2025 / Revised: 7 September 2025 / Accepted: 15 September 2025 / Published: 29 September 2025
Browse Figures
Versions Notes

Abstract

Artificial Intelligence (AI) has become deeply embedded in the advertising industry, presenting both opportunities and challenges. Understanding this transformation requires mapping the dyad’s structure and identifying the conditions that shape its evolution. This study applies a scientometric framework, integrating bibliometric network analysis with statistical modeling, to examine AI advertising as a knowledge ecosystem. By analyzing patterns of collaboration, thematic convergence, and structural centrality, we interpret how scholarly networks generate, connect, and diffuse ideas in ways that influence both academic and industry practices. The findings reveal that the field’s growth is underpinned by interconnected clusters of expertise with strategic opportunities emerging from interdisciplinary integration and global collaboration. Simultaneously, consolidating influence among a few dominant actors raises questions about diversity, access, and the balance between innovation and ethical responsibility. Statistical analyses conducted in SPSS Statistics version 29.0.2.0 further identify the bibliometric and structural factors that most predict citation impact, strengthening the study’s contribution to understanding how influence is built and sustained in AI-driven advertising research.

1. Introduction

The advertising industry is undergoing constant evolution, moving from traditional formats to digital marketing and, more recently, to AI-driven advertising. Artificial Intelligence (AI), a crucial element in the digital landscape, has demonstrated its capacity to make advertising more efficient, effective, and personalized through market prediction and targeting, highly personalized advertising experiences, recommendation systems, retargeting, and rapid data compression for optimized strategies [1]. Increasingly complex advertising strategies also characterize this era, and businesses may face difficulties in creating and implementing optimized strategies [2]. AI systems can further enhance efficiency by integrating machine learning with adaptive algorithms to modify advertising content across platforms [3]. Specifically, AI assists advertisers through ad optimization, automated ad generation, and personalization using natural language processing, image recognition, speech recognition, machine learning (ML), natural language generation, and image and speech generation [4]. For example, Salesforce (U.S.) utilizes AI-driven sights that improve their lead conversion and revenue growth [5,6] while Amazon uses recommendation system for an enhanced customer experience [7]. As of 2023, the value of AI advertising stands at USD 43,557.36 million, with a Compound Annual Growth Rate (CAGR) of 18.94% [8], and global online advertising investments is also anticipated to extend to USD 517.51 billion; 80% of these investments accounts for AI [9].
While AI provides numerous strategic advantages, including computational advertising that enables greater matching accuracy, efficient customization, and contextual interaction between users and ads [10,11], the literature remains fragmented. Current studies tend to emphasize (1) technical applications such as Click Through Rate (CTR) predication, recommendation systems and automated bidding; (2) consumer-facing outcomes, such as engagement, personalization, and brand loyalty; or (3) risk consideration, such as data privacy, algorithm bias, and transparency [12,13,14]. Nevertheless, integrative analyses that map how these themes connect within the broader knowledge ecosystem of AI advertising are lacking. A knowledge ecosystem is being designed as the dynamic network of actors, institutions, and technologies that collectively generate, share, and apply knowledge [15]. Moreover, while scientometric approaches have been used in related field [16], predictors of high-impact research in AI advertising, such as collaboration patterns, institutional influence, and thematic positioning, remain underexplored.
This study addressed three gaps in the literature by mapping the conceptual, collaborative, and citation structure of AI advertising research, identifying strategic prospects and potential risks embedded in the knowledge ecosystem, and determining bibliometric predictors of influential research contributions. Accordingly, it investigates the following research questions:
  • RQ1: What knowledge sources, such as authors, institutions, countries, and journals, are shaping the development of AI advertising research?
  • RQ2: How has research on generative AI in advertising evolved, and what strategic prospects and potential risks can be identified through scientometric analysis?
  • RQ3: Which factors predict high-impact or essential research contributions in AI-driven advertising, and how can these insights inform to future research agendas?
The study answers these questions by applying a mixed method scientometric approach integrating semantic network mapping, co-authorship and citation analysis, bibliographic coupling, and statistical modeling of bibliometric indicators. Data were sourced from Web of Science Core Collection and analyzed using VOSviewer 1.6.20, NodeXL Pro, and IBM SPSS Statistics version 29.0.2.0. By linking structural mapping with citation impact analysis, this research advances theoretical understanding of the AI advertising knowledge and provides actionable insights. The findings can inform strategic collaborations, highlight emerging high-impact themes, and support responsibly integrating AI in advertising, ensuring innovation is balanced with ethical considerations.

1.1. Theoretical Background

1.1.1. Knowledge Ecosystem in AI Advertising Studies

Building on the identified gaps, it is essential to situate AI advertising research within a broader framework that explains how knowledge is generated, exchanged, and applied across diverse stakeholders. In innovation and technology research, the knowledge system comprises an evolving network of interconnected actors, resources, and activities that collectively produce and transform knowledge to address shared challenges [17]. Rather than functioning as static repositories, these ecosystems are adaptive systems in which participants, such as firms, universities, policymakers, and intermediaries, continuously codevelop knowledge in response to shifting technological and market conditions [18].
A knowledge ecosystem is differentiated from other collaborative structures by its emphasis on diversity of expertise, reciprocal relationships, and the coexistence of both competition and cooperation [19]. Actors operate through formal mechanisms, such as contractual partnerships, as well as informal exchanges, such as communities of practice, enabling the circulation of codified data and tacit know-how across organizational and disciplinary boundaries [20]. This interplay ensures that knowledge flows are not merely transferred but are adapted, recombined, and applied in context-specific ways, making the system more resilient and innovative.
In this study, the knowledge ecosystem is understood as an adaptive, loosely coupled collective of autonomous yet interdependent actors that share the aim of creating, exchanging, and applying knowledge [21,22]. Unlike traditional organizations bound by rigid hierarchies, knowledge ecosystems operate as meta-organizational structures in which participants such as advertisers, AI developers, and platform providers retain independence while operating in coordination toward a common objective [23].
Flexibility is essential in AI advertising, where collaborations often involve stakeholders with different incentives but complementary resources. Drawing on the partial organization perspective, such ecosystems can be coordinated through selected elements, such as shared standards or rules, without the need for full formal authority. This facilitates the fragile integration of emerging AI technologies, rapid experimentation, and context-specific adaptation, all while preserving the autonomy needed for innovation.
Applied to advertising, the knowledge ecosystem extends beyond traditional academic–industry collaborations. This expanded collaborative logic aligns with quadruple helix perspective, which recognizes that innovation systems increasingly require the integration of diverse actor types, including civic society, media, and cultural organizations alongside academia, industry, and government, to address complex societal and technical changes [24].
It incorporates platform providers, AI developers, creative agencies, advertisers, regulatory modes, and user communities, each influencing and being influenced by the flow of algorithmic innovations, market intelligence, and innovative strategies. Social network analysis has revealed this type of conceptual architecture in the marketing fields, uncovering key clusters and subdomains [25]. Understanding their interplay is critical for evaluating not only the pace and direction of AI adoption in advertising but also its ethical, societal, and economic implications.
Several studies highlight the value of mapping knowledge flows in marketing and advertising. Social network analysis has been used to reveal the conceptual architecture, uncovering key clusters and subdomains of inquiry. Researchers have also proposed frameworks to trace how emerging technologies integrate into advertising ecosystems, arguing that conceptual clarity and methodological innovation are essential for leveraging the potential for these tools.
Recently, algorithmic systems themselves have become both outputs and enablers of the knowledge ecosystem. A notable example is the algorithm for marketing strategy decision-making (AMSDM), developed by [26], illustrating how AI technologies—specifically Azure Text Analysis—can transform unstructured customer feedback into actionable marketing intelligence by identifying sentiment, subdomains, keywords, etc. Such systems function as embedded knowledge infrastructures, filtering, interpreting, and operationalizing consumer sentiment within strategic decision-making processes. Integrating AMSDM into advertising practice shows how AI enhances the adaptive capacity of the knowledge ecosystem, linking technological innovation with communicative precision, and enabling the ongoing recalibration of knowledge and strategies through sustained interaction with intelligent systems.

1.1.2. Scientometric Analysis of AI in Advertising

Within the knowledge ecosystem described in Section Data Collection, scientometric analysis is a critical method for systematically mapping how research themes, actors, and intellectual linkages evolve in AI advertising. By quantifying relationships among publications, authors, institutions, and keywords, scientometrics operationalizes the flow of knowledge across the ecosystem and identifies the conceptual clusters that underpin its development.
Several scientometric studies have evaluated the evolving role of AI in advertising, identifying core research themes, such as consumer sentiment analysis, brand management, trust building, and innovation in service delivery [27]. These studies also provide extensive literature assessments, highlight key contributors, and identify future research avenues, including leveraging semantic networks and ML, to deepen insights into consumer behavior [28,29]. While much of the literature focuses on AI’s potential, recent work has also focused on critical ethical and regulatory risks associated with AI. Concerns include algorithmic bias, data privacy breaches, copyright issues, and the increasing spread of misinformation [30,31]. Wach et al. [32] add that generative AI may also contribute to technostress, poor quality outputs, and regulatory gaps. In turn, scholars have emphasized the need for clearer legislation, stronger ethical guidelines, and coordinated efforts among researchers, policymakers, and industry leaders to ensure responsible use [33]. Positioning these themes within the knowledge ecosystem framework highlights how scientometric mapping is not only a descriptive but also a strategic tool that helps stakeholders anticipate emerging opportunities and risks that permeate the system. These risks underscore the importance of mapping both the strategic opportunities in AI advertising and the emerging challenges to be addressed.

1.1.3. Predictors of Research Impact in AI Advertising

Researchers have called for clearer definitions of what constitutes essential or high-impact research in the evolving field of AI advertising.
Research success is determined by citation frequency and visibility in influential journals. However, AI-generated advertising remains methodologically fragmented, with most studies still concentrated in areas such as e-commerce, social media, and targeted advertising [34,35]. Despite this growth, there is a noticeable gap in empirical research that identifies the predictors of impact outcomes in this domain. Drawing from the broader scientometric literature, factors such as international co-authorship, interdisciplinarity, institutional prestige, and research funding have been shown to influence scholarly impact across disciplines [36,37]. Nevertheless, these variables have not been systematically applied to the field. Better metrics and frameworks for evaluating research impact in this area would help clarify the field’s intellectual structure, identify influential research trajectories, and inform future funding and publication strategies. This gap requires what is being studied, how it is studied, and why specific research studies achieve greater academic and societal traction. These are the questions this study addresses in its investigation of high-impact trends in AI-generated advertising.
Research has explored consumer behavior, branding, and AI-driven innovation, yet the literature remains fragmented. Links between ethical risks, and long-term strategy are rarely examined, and factors driving influential research are not well understood. Using knowledge ecosystem perspective, this study investigates how scholarly outputs, collaboration patterns, and thematic trends interact to generate, circulate, and validate knowledge in AI advertising.

1.1.4. Exploring Trends in Advertising Research Through a Framework

Figure 1 depicts the limitation of previous research by providing a comprehensive illustration of AI advertising publications sourced from the Web of Science (WoS). This enables in-depth analysis of the knowledge ecosystem and the factors influencing citation networks within the field. The proposed framework applies scientometric and bibliometric approaches to examine both conceptual and structural dimensions of AI advertising research.
The left side of the framework (as shown in Figure 1) focuses on mapping the advertising knowledge ecosystem by identifying key trends, potential risks, and dominant knowledge sources. This is achieved through semantic network analysis, co-authorship mapping (by country and affiliation), citation analysis, and bibliometric techniques. These methods provide a structured view of the cognitive foundations of the field, identify how topics evolved, and determine the influential publications and research gaps. They also aid in constructing taxonomies of concepts, determining authors’ areas of expertise, and enabling exploration browsing through visualizations of research interconnections [38]. Color-coded themes—green for strategic prospects and red for potential risks —further clarify the assessment of opportunities and threats. The green signifies positive potential and red serving as a cautionary market of fluctuating outcomes that may be neutral, negative or manageable depending on the context. In this study, strategic prospects refer to research themes, technologies, or methodological direction in AI advertising that demonstrate growth potential, cross disciplinary relevance, or capacity to shape future practice and scholarship.
As AI continues to transform advertising strategies and knowledge production, understanding the dynamics of the research ecosystem becomes essential. Traditional literature reviews may be inadequate to capture the scale and complexity of this rapidly evolving domain. In contrast, bibliometric and scientometric approaches enable systematic, data-driven insights into how knowledge is produced, disseminated, and applied—making these methods particularly suited for identifying intellectual, collaborative patterns and epistemic blind spots. This framework adopts these approaches to map the development of AI advertising research, both in terms of opportunity and risk. In doing so, it is essential to recognize that overreliance on automated analyses, whether statistical or AI-driven, can diminish the exercise of critical, context-sensitive judgment, a factor integral to producing robust and meaningful empirical finding [39].
On another note, the right side (as seen in Figure 1) examines factors that influence research impact, using bibliometric indicators and a statistical method to identify citation-related predictors. Citations are often used as proxies for research quality and influence, and analyzing patterns of citation can help clarify which research trajectories have gained academic traction. This layer of analysis provides insight into how high-impact scholarship has evolved and aids in assessing how research in AI advertising is evolving. Previous studies note that the use of metaphors in knowledge-based systems reflects underlying theoretical attempts to navigate complex interdisciplinary dynamics [40], further justifying the layered analytical approach used in this study.

2. Materials and Methods

Data Collection

This study employs scientometric analysis, comprising network and bibliometric analysis, to examine the intellectual structure, collaboration patterns, and thematic evolution of AI-generated advertising. Network analysis explores citation patterns to reveal the structure and influence within and upon the communication field [41]. In contrast, bibliometric analysis analyzes large volumes of scientific data to uncover emerging trends in article and journal performance, research constituents, and patterns of collaborations [42]. These methods are well-suited to the study’s objectives because they enable a comprehensive, data-driven assessment of both the strategic opportunities and emerging challenges in the field, which are capabilities that have been demonstrated in previous research on AI advertising. For example, He (2022) [43] identified “big data,” “precision marketing,” and “digital marketing” as key areas of research, and [44] employed bibliometric and social network analysis to identify influential authors and significant publications.
A dataset of 4880 English-language papers was retrieved from the Web of Science (WoS) database, a prominent platform known for its extensive coverage of high-impact research [45], covering the period 1990–2023. Only scholarly document types such as proceeding papers, articles, editorial content, revie articles, and early access were included to ensure the relevance and quality of the samples. Searches were limited to titles and abstracts, using an iteratively refined set of keywords, “artificial intelligence,” “AI advertising,” and “Machine Learning Advertising”, to maximize precision and topical relevance. The restriction to English-language publications is acknowledged as a limitation and addressed in the Conclusion Section.
VOSViewer and NodeXL Pro tools were used to construct and analyze the network [46] to identify growth and trends, identify topics and contributors [47], and map and visualize scientific domains. VOSviewer, in particular, provides advanced bibliometric mapping capabilities that convert large and complex datasets into clear visual representations, enabling the detection of thematic clusters, collaboration patterns, etc. These capabilities are enhanced when integrated with AI-assisted interpretation to improve research outcomes [48]. It focuses on graphics for large maps, uses visualization of similarities mapping, and creates label, density, cluster density, and scatter views for extensive map analysis [49].
NodeXL Pro “https://www.smrfoundation.org/ (accessed on 5 June 2025)”, on the other hand, visualizes and analyzes networks, providing tools for data import, social media integration, various metrics calculations, and automated reporting [50]. Both tools have the capacity to process large datasets, detect thematic clusters, and visualize relationships among authors, institutions, and countries [51,52,53,54].
Throughout the analysis process, centrality metrics, clustering techniques, and visualization methods facilitate a comprehensive understanding of the intricate connections and structures within complex networks. Recognizing crucial metrics such as degree, betweenness, and closeness centrality helps identify influential nodes and power dynamics within the network [55,56,57,58], which are of significant interest. The analysis applied parameters such as citation analysis, including all 4880 documents (minimum citations = 0), selecting the top 1000 by total link strength. Co-authorship analysis used complete counting for both countries (maximum 25 per document; all 16 included) and organizations (maximum 25 per document; minimum five documents per sutor, 456 of 5159 authors included). Bibliographic coupling included all documents, with the top 1000 selected by link strength. Term co-occurrence analysis (titles and abstracts) applied a minimum occurrence of 10, yielding 1944 qualifying terms, and the 1166 most relevant (60%) were selected. Several studies have investigated term clustering for identifying topics and trends [59,60]. Clustering enables coherent grouping of communities of nodes that share common themes or interconnected interests, allowing for the detection of cohesive groups of nodes [61,62]. For reference, Table 1, Table 2, Table 3 and Table 4, as well as Figure 2, Figure 4, Figure 5, Figure 6 and Figure 8, were generated using VosViewer. All bibliometric mapping and network visualizations were conducted using VOSViewer with complete counting, applying the parameters detailed above.
IBM SPSS Statistics was used to ascertain the characteristics that influence citations. The researchers employed correlation analysis to evaluate the associations between bibliometric variables, ANOVA to investigate variations in research relevance across categories, and regression analysis to construct a predictive model for research importance [63]. Table 7 and everything in the Appendix A were generated using IBM SPSS.

3. Results

3.1. Investigating Co-Occurrences of Terms

We performed keyword co-occurrence analysis to understand how research on generative AI in advertising has evolved. Keywords, such as machine learning, customer engagement, automation, and brand personalization, emerged as central nodes across multiple clusters. This trend highlights an increased scholarly focus on performance-driven application in the field.
Figure 2 visualizes 871 nodes representing co-occurring terms clustered into five groups. These terms were extracted from publications and clustered into specific themes, as shown in Table 1. These were formed by analyzing the interconnected terms. A detailed breakdown of subtopics is shown in Table 2. The cluster with the highest degree of centrality belongs to Clusters 4 and 5, where terms such as model (858), algorithm (805), performance (798), user (769) and prediction (734) are found. These terms underscore fundamental importance in advancing the understanding, application, and development of AI in advertising research, with models and algorithms at the forefront. Researchers use AI models and algorithms to analyze data, predict user behavior, and optimize advertising performance. This methodological significance may interest researchers in publishing papers evaluating AI-driven campaigns’ effectiveness, which makes these terms more frequent. Users reflect the user-centric approach of AI Advertising research as AI enables advertisers to better understand user behaviors and preferences better, leading to more personalized and targeted advertising experiences. “Prediction”, which refers to predictive capabilities, may indicate deployment of AI models and algorithms that can anticipate user actions and preferences, enabling advertisers to deliver timely, relevant advertisements, making it a hotspot for many researchers. Lastly, as part of the top five terms with the highest centrality, “core concepts” form the foundation for theoretical frameworks, empirical studies, and practical applications, which are likely to be the central discourse of this field.
Table 1 involved a two-stage process. First, co-occurring terms from the publications were identified and clustered using VOSviewer. VOSviewer’s analysis generated five distinct clusters of interconnected terms based on their co-occurrence frequency within the dataset. These clusters were represented by color which are automatically assigned by VOSViewer per group. Second, after the VOSviewer clustering, we performed a manual thematic analysis on the resulting clusters to determine overarching themes and concepts and to assign meaningful topic labels. These themes were then summarized and organized into the five main topics presented in Table 1. This manual assignment ensured that the resulting issues in Table 1 were coherent, interpretable, and accurately reflected the content of the co-occurring terms identified in the initial VOSviewer analysis. The detailed breakdown of subtopics within each main topic is presented in Table 2.
Following topic detection from Table 1, concepts and themes were manually analyzed and grouped into two overarching categories: strategic prospects and potential risks. This categorization adhered to the framework established by Fulton et al. (2024) [64], which analyzes AI’s multifaceted impacts across eight categories: societal, economic, ethical, political, environmental, data, technological, and organizational. Each concept or theme from Table 1 was classified as either a benefit (strategic prospect) or a risk, depending on its context. The resulting data were then organized into a smart chart (Figure 3), revealing that contributors to both strategic prospects and potential risks mainly came from Clusters 2, 3, and 5, as these clusters emphasize application-focused dimensions of AI advertising, making them key drivers of both challenges and opportunities in the field. The smart chart further highlights AI’s central role as a technology driving advancements in AI advertising.
AI enables strategic targeting and personalization, real-time bid placement and targeting optimization, creative automation, and content development in advertising. AI-driven advertising boosts customer engagement and conversion rates, while real-time optimization of budget allocation and ROI. AI improves marketing success, user behavior, and decision-making, while chatbots improve customer satisfaction and assistance. This quickens the process and reduces costs. In the aspect of AI in advertising, designing compelling campaigns is possible with collected data.
This shift indicates that the advertising knowledge ecosystem is rapidly adapting to the operational logic of AI technologies, emphasizing efficiency, segmentation, and personalization. However, as demonstrated in broader studies of knowledge–society interfaces, effective technological adoption also depends on fostering transparent, inclusive engagement between expert communities and the wider public to ensure societal trust and accountability [65]. The relative absence of terms related to governance, regulation, or long-term societal impacts signals a potential blind spot. While researchers are eager to examine the benefits of AI, the strategic risks, such as algorithmic bias or ethical constraints, remain underdeveloped as thematic clusters.
These opportunities also entail risks, as vast volumes of user data will create privacy concerns, requiring strong data protection mechanisms. Furthermore, AI may reinforce biases, leading to unfair targeting and discrimination. Some AI models lack transparency, reducing online advertising transparency and accountability, compromising accuracy. AI systems are vulnerable to fraud and manipulation, making them exploitable. Finally, AI advertising automation raises concerns about job loss and extinction.

3.2. Investigating Knowledge Sources Through Co-Authorship Analysis (Countries and Institutions)

Worldwide scientific collaboration involves the USA, China, Germany, England, and France in co-authorship networks of scientific papers in all subjects [66]. In advertising, this includes examining this network’s centrality measurements, showing the U.S.’s dominance in citations and centrality categories, indicating it is a leading publication and information exchange hub. The U.S. mediates between study domains in the network, illustrating that its work is widely acknowledged and influences how other regions approach AI in advertising.
As seen in Figure 4, China regions’ high citation count suggests it is becoming a major participant in the sector; however, England, Germany, and other countries are also strong. Intending to become a global AI leader by 2030, the Chinese government invested USD 52 billion in science and research in 2024, up 10% from 2023 [67]. Its huge market size gives big data an economic advantage. They have led technological progress and market applications, but they are still developing university–industry relations, hence the latest investments [68]. Other countries may be seen in Table 2.
Figure 5 and Table 3 show the dynamic collaboration between industry and academia in AI and advertising research. For example, Tsinghua University in China and Harvard University in the USA are renowned for their high citation rates and research output. Tsinghua University has the highest competitive score for betweenness centrality and closeness centrality, and it is supported by the Ministry of Education of China. The Chinese Academy of Science has published extensively on CTR prediction and deep learning, both of which are crucial for online advertising. They have collaborated with CTR prediction institutions and internet advertising giants Tencent and Alibaba. In contrast, Harvard University focuses on understanding user behavior for targeted advertising.

3.3. Understanding Publication Influences and Shared Intellectual Landscape Through Citation Analysis and Bibliographic Coupling Analysis

Advertising is increasingly interested in AI applications, with AI optimizing advertising through CTR prediction. Jordan and Mitchell’s “Machine Learning: Trends, Perspectives, and Prospects” provides insights into supervised, unsupervised, and reinforcement learning. AI algorithms are employed in predicting advertising campaign, bidding tactics, and analyzing client behavior. Davenport et al.’s article explores AI’s impact on marketing, including campaign automation and AI-powered effectiveness measurement. Finally, Zhou et al. (2018) [69] explain how they train deep learning models using user data and click history to predict ad clicks, potentially improving overall ROI.
Figure 6 and Figure 7 illustrate how citation weight shapes scholarly influence in the field, addressing RQ1. The most highly cited works are by Jordan and Mitchell (2015), Autor (2015), and Zhou (2018), which reflect different but complementary types of impact. Jordan and Mitchell establish the foundational vision of ML that underpins most AI systems in use today. Autor incorporates a labor and policy perspective, connecting AI development to workforce transformations, which is a growing concern in advertising automation. Zhou’s work on deep interest modeling, already deployed in Alibaba’s ad systems, represents high-impact innovation with precise industrial application. These citation patterns suggest that influence in this space comes from more than disciplinary relevance alone. What distinguishes these works is their ability to define, reframe, or operationalize how AI is applied across systems and sectors, including advertising. This shows how this network of authors shapes how the field understands and applies AI itself.
In addressing RQ1, Figure 8 provides a structural perspective by examining bibliographic coupling, complementing citation-based analyses through identifying authors whose works form the intellectual backbone of the field. Quantitative reference of Figure 8 can be seen in Table 4. The dominance of Cluster 1, including top authors such as Peltier (2024), Hermann (2022), and Vlačić et al. (2021) reveals a core group whose work is united both by their high total link strength and their shared emphasis on defining AI marketing, exploring ethical considerations, and future-proofing the field.
Specifically, Peltier et al. (2024) provides a conceptual framework integrating AI into interactive marketing, providing a roadmap for other scholars to follow. Herman (2022) introduces an ethical lens, focusing on how AI can be leveraged for social good, resonating with ongoing discussions on responsible AI. Meanwhile, Vlačić et al. (2021) synthesize the evolving role of AI through a systematic review, influencing how subsequent researchers categorize and study capabilities. The dense interlinkages among these publications are demonstrated through their shared reference structures, indicating how they collectively scaffold the field’s dominant discourse. Thus, their influence is not just based on their high citation count, but, more importantly, on their conceptual authority, as evidenced by their embeddedness in the network. This shows that influence is determined not only by which researchers are cited, but by who shapes the research logic of AI in advertising, offering a deeper understanding of scholarly influence grounded in intellectual coalignment.

3.4. Factors Affecting Citation Networks in AI in Advertising Research: Focused on Non-Quality Factors—Variables

The study examines 4886 WoS publications’ bibliographic data through the variables found in Table 5. As shown in Table 6, only top five from the complete lists were shown. The missing percentages of these classifications contribute to other research areas, countries and years that gave lower publication contributions, as this focused on top five variables only. Notably, a total of 127 research areas, 117 countries, and 35 years (1990–2022) comprise the totality of each classification.
Results demonstrate a dominance in computer science and engineering publications, suggesting a strong emphasis on technology alongside its implementation. The presence of business economics indicates that there is an interest in commercial viability and concern for the economic impact of AI technologies, involving research on economic effects. Topics include AI algorithms, ML techniques, and data infrastructure.
It is evident even from the co-authorship analysis that global collaboration has been strong in countries shown in Table 6. However, as shown in Figure 9, the publications, particularly in 2022 and 2023, indicate growing interest in investment of resources to AI advertising research, as these are years where AI is most prominent.
Figure 9 illustrates the increase in publications in the AI advertising domain in the period studied. This chart indicates a significant uptick in publication volume starting in 2018, signaling an inflection point when scholarly interest in AI marketing rapidly intensified. This demonstrates a transition from early-stage conceptual exploration to mature, multidimensional inquiry. For instance, Zhou et al. (2018) introduced deep interest network, significantly advancing AI-enabled personalization in advertising. In that same period, Wirtz et al. (2018) explored the implications of service robots, broadening AI’s role from backend automation to consumer-facing interactions.
Growth accelerates during 2021–2024, marked by high centrality contributions such as Vlačić et al. (2021), who developed a comprehensive research agenda, and Manis and Madhavaram (2023), who conceptualized the hierarchy of AI marketing capabilities, linking technological development with strategic capability building. During these years, several authors also researched ethical and consumer protection concerns, AI inclusivity, and vulnerability, further diversifying the thematic spectrum. These works are recent and structurally significant in the network, demonstrating both topical relevance and citation centrality. This figure reflects not just a quantitative increase in scholarly output but a qualitative diversification of scholarship on AI advertising, as research themes expanded from performance metrics to ethics, inclusion, and strategic implementation. The prominence of post-2018 works in both citation metrics and network centrality reinforces the field’s transition into a mature and multidimensional domain.
Table 7 indicates that AI in advertising is a fast-growing research area, with most publications appearing after 2019. The average author (3.57) and affiliation counts (4.85) highlight a strong culture of cross-institutional collaboration. The number of highly cited reference (51) and moderate paper lengths (14 pages) suggest research builds on previous work while maintaining concise delivery. Citation impact is highly uneven; few landmark studies dominate the field, reflecting that the domain is in the early stage and has yet to be consolidated. Keywords and research area counts indicate a balance between specialized studies and broader, multitopic exploration, underscoring both depth and diversity in the field’s knowledge base.
The correlation analysis (as shown in Appendix A, Table A1) reveals several statistically significant relationships among bibliometric factors. Notably, the publication year shows a positive correlation with the cited reference count (r = 0.255), indicating that more recent publications tend to include a greater number of references, likely reflecting the cumulative knowledge building in the field. Similarly, publication year correlates with both author and affiliation count (r = 0.200), suggesting that newer studies are increasingly collaborative, spanning multiple institutions.
A strong relationship is observed between the number of pages and the number of cited references (r = 0601), implying that longer studies tend to draw from a broader base of the literature. Moderate correlations between the number of affiliations, authors, keywords, and cited references (r = 0.139 and r = 0.173, respectively) further suggest that broader institutional and thematic coverage enhance scholarly visibility. Conversely, the weak negative correlations involving research area count indicate that diversification across research areas does not necessarily translate into higher citation performance.

3.5. Regression Analysis

The regression results (as shown in Appendix A, Table A2 and Table A3) indicate that recent publications and more cited references are significant predictors of citation impact. Each additional year since a publication results in an estimated 2.3 fewer citations, underscoring the time-sensitive nature of academic visibility. Conversely, every additional reference cited in a paper is associated with a modest but significant increase in citations received, suggesting that integrating relevant previous work can improve a study’s uptake in subsequent research.
Other variables, such as the number of authors, research areas, and pages, have no statistically significant effect in this model, indicating that collaboration scale and article length alone are insufficient to predict impact once recency and reference depth have been accounted for. These findings indicate that strategic citation practices and timely publication in emerging topics may be more influential drivers of research visibility than the structural attributes of the paper.

3.6. Multicollinearity and Variable Relationships

The correlation and covariance diagnostics (see Appendix A, Table A4) indicate generally low-to-moderate intervariable correlations, suggesting minimal multicollinearity concerns for the regression models. The number of pages is strongly positively correlated with the number of cited references (r = 0.574), which reflects an expected relationship where longer studies typically integrate more references. Similarly, the significant negative correlation between the number of research areas and WoS categories (r = −0.758) suggests that articles on multiple research areas tend to have lower word counts, potentially due to the concise reporting required in writing across interdisciplinary topics.
The publication year has a weak-to-moderate negative correlation with both the number of cited references (r= 0.160) and author keywords (r = −0.166), potentially indicating that newer studies are more focused and keyword-efficient, possibly due to refined thematic targeting in AI advertising research. Other correlations are also weak, implying that most predictors contribute uniquely to model variance without substantial overlap. Covariance values are negligible across variables, reinforcing the absence of problematic redundancy.

4. Discussions

Integrating AI into advertising is not only altering how campaigns are implemented but also reshaping the knowledge ecosystem that underpins the field. At the center of this transformation lies AI’s capacity for enhanced targeting and personalization, achieved by aggregating and interpreting vast datasets [70,71,72]. This enables advertising to become more adaptive, context-aware, and emotionally resonant, driving measurable gains in brand perception and conversion. However, these advancements do not exist in isolation; instead, they are embedded in a broader sociotechnical system comprising regulatory frameworks, cultural contexts, and evolving industry norms.
Addressing the first research question on the strategic opportunities and risks of AI in advertising, the study’s thematic analysis reveals that opportunities stem from data-enabled precision and personalization, whereas risk clusters around ethical, privacy, and interpretability concerns [73]. These risks are not peripheral challenges, as they shape the trustworthiness and long-term viability of AI-driven advertising for policymakers, prompting proactive safeguards that ensure AI deployments are secure, fair, and transparent. For advertising practitioners, they highlight the strategic necessity of embedding ethical governance into innovation pipelines to sustain brand equity. Notably, while this study maps structural and thematic patterns in the AI advertising literature, it does not examine the ethical and societal ramifications of deploying AI in the field. However, concerns about algorithmic discrimination, consumer anatomy, and transparency remain vital areas of debate.
International collaboration has significantly increased over the past 30 years, shifting the scientific research landscape toward international networks, with the percentage of internationally co-authored publications doubling over 20 years from 10% in 1990 [74]. This network structure illustrates that the knowledge ecosystem is anchored by a few high-capacity hubs, particularly in the U.S. and China, that set much of the research and application agenda. Their influence is amplified by strong university–industry linkages and significant R&D investments, enabling rapid transformation from theoretical models to market-ready tools. Leading universities such as Harvard and MIT have been helping develop this knowledge, as they have been making significant contributions to integrating AI advertising, merging theoretical discoveries in universities with practical applications in technology corporations, such as IBM, Adobe, and more. China’s regions, such as the Chinese Mainland, Hong Kong, and Macao, play a significant role in this network because they have invested heavily in research and development, with twice as many as those in the U.S. They have steadily increased their investment in this field as they aim to be global AI leaders by 2023. These two countries have been competing with China significantly, aiming to overtake the U.S. through strategic programs to foster AI development, with most of their research publications having scientific relevance and having applied value in industries that would contribute to economic growth. Tsinghua University and the Chinese Academy of Sciences prove this theory, as their outputs provide evidence of their collaboration with industry giants like Alibaba Group, Tencent, and Baidu. They also collaborate with international universities with a focus on CTR predictions, with practical implications in the field of online advertising.
India, the UK, Germany, and France are among the top collaborators in advertising innovation. India is known as the world’s tech backbone, with its naturally growing consumer base and tech talent pool making it an attractive hub for AI advertising. Like China, India’s goal is to increase its research investments to USD 17 billion by 2027 [75]. Results show that India has been gearing up for improvements in various fields. Their advertising goes beyond the limitations of marketing, linking efforts with other fields. In Germany, public awareness of the use of AI technologies is high [76]. Their publications mainly focus on engineering. However, given their literacy, many are concerned about privacy and ethics since Germany’s GDPR policy is foundational to the country. The UK’s financial and marketing hub, particularly London, offers opportunities for AI research. Universities with strong computer science and marketing programs contribute to AI in advertising. The UK’s focus on ML and consumer behavior-centric research is crucial.
How knowledge production and dissemination are shaping the AI advertising domain is illuminated by the interplay between high-impact academic publications and industry uptake. Influential works bridging AI’s technical foundations with practical marketing applications [77,78] demonstrate the cyclical nature of the knowledge ecosystem. This shows that theoretical insight informs tools and strategies, whose market adoption generates new data and questions for research. In this cycle, central actors contribute knowledge as well as curate the agenda for future inquiry, reinforcing their position within the global network.
Taken together, these findings extend the conceptual framework of the knowledge ecosystem by illustrating how AI in advertising functions as a dynamic, multilayered system. It is driven by feedback loops between research practice and policy, shaped by the distribution of power in global collaborations, and moderated by ethical and cultural constraints [79,80,81]. The health of this ecosystem depends on maintaining balance, particularly fostering innovation without public trust, diversifying participation without fragmenting standards, and ensuring that the benefits of AI’s capabilities are equitably distributed across markets and stakeholders.

5. Conclusions

The mapping of AI-driven advertising research reveals a field that is not merely expanding but structurally reorganizing. The rise in interdisciplinary clusters, cross-national collaborations, and high-centrality authors indicates that the knowledge ecosystem is moving toward greater integration, though unevenly. This integration remains concentrated within a limited set of countries, institutions, and thematic domains, shaping which problems are prioritized, how solutions are framed, and whose perspectives are embedded in the operational logic of AI advertising. For scholars, the network patterns and bibliometric predictors identified here suggest aligning research with emerging high-impact themes and embedding work in collaborative, internationally connected clusters to enhance both visibility and influence. For policymakers, the blind spots which are particularly in governance, regulation, and long-term societal impact, signal the urgency of policy frameworks that can keep pace with technical adoption. For industry practitioners, the trajectory of AI advertising research underscores that competitive advantage will increasingly hinge on integrating ethical safeguards, transparency measures, and inclusive design into AI-enabled strategies, not solely on optimizing performance metrics.
Ultimately, the health of this knowledge ecosystem depends on maintaining balance. This entails fostering innovation while building public trust, diversifying participation without fragmenting standards, and ensuring AI’s benefits are equitably distributed across markets and stakeholders. Achieving this balance requires embedding principles such as transparency, ethical safeguards, and sustained human oversight into the design and deployment of generative AI, as underscored by recent scholarship on responsible advertising practices [82].

Limitations

This study has several limitations. First, the dataset was extracted exclusively from the Web of Science (WoS) database, which, while reputable, may not encompass all relevant publications indexed in other sources such as Scopus or Google Scholar. Second, the analysis focuses solely on English-language publications due to the authors’ language proficiency, potentially underrepresenting contributions from other linguistic regions. Third, although automated topic modeling techniques such as Latent Dirichlet Allocation (LDA) were not employed, the study relied on manual topic detection, guided by expert interpretation patterns, keyword occurrence, and centrality data, while this provided meaningful thematic insights, the inclusion of Delphi-based expert validation or algorithmic modeling in future research could further strengthen the conceptual robustness. Finally, the study does not directly address ethical and societal dimensions of AI in advertising, such as algorithmic discrimination or erosion of consumer trust. Future studies should adopt a multi-database, multilingual, and mixed-method approaches while integrating critical perspective on ethics, autonomy, and transparency as AI becomes increasingly embedded in persuasive communication.

Author Contributions

Conceptualization, H.-W.P. and C.V.L.; methodology, H.-W.P. and C.V.L.; software, C.V.L.; validation, H.-W.P. and C.V.L.; formal analysis, C.V.L.; investigation, C.V.L. and H.-W.P.; resources, C.V.L.; data curation, C.V.L.; writing—original draft preparation, C.V.L.; writing—review and editing, C.V.L. and H.-W.P.; visualization, C.V.L.; supervision, H.-W.P.; project administration, H.-W.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

The data that support the findings of this study are available on request from the first author.

Acknowledgments

The authors thank MDPI editors and reviewers for thoroughly reviewing the paper and providing insightful comments.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

The tables below consist of variable definitions and statistical analysis results to support findings in Section 4.
Table A1. Definition of Terms.
Table A1. Definition of Terms.
VariableDefinition
A_CountTotal number of authors per document
G_CountNumber of countries affiliated with document authors
RA_CountNumber of research areas assigned to a document
AK_CountNumber of author keywords assigned to a document
WC_CountNumber of Web of Science categories assigned
Number of PagesTotal page length of a document
Publication YearYear in which the document was published
Cited Reference CountNumber of references cited within the document
Table A2. Correlation analysis.
Table A2. Correlation analysis.
Times Cited, All DatabasesPublication YearA_CountG_CountAK_CountRA_CountNumber of PagesCited Reference Count
Times Cited, All DatabasesPearson Correlation1−0.113 * −00.019−00.17−00.17−00.0190.046 **0.084 **
Sig. (2-tailed) <0.00100.1800.24300.24300.18800.001<0.001
N48804880488048804880488048804880
PublicationPearson Correlation−0.113 * 1−00.0180.200 **0.200 **0.057 **0.188 **0.255 **
Sig. (2-tailed)<0.001 00.211<0.001<0.001<0.001<0.001<0.001
N48804880488048804880488048804880
A_CountPearson Correlation−00.019−1810000.005−00.009 −0.029 *
Sig. (2-tailed)00.1800.211 00.99900.99900.72200.5080.044
N48804880488048804880488048804880
G_CountPearson Correlation−00.0170.200 **0110.000 **0.063 **0.139 **173 **
Sig. (2-tailed)00.243<0.00100.999 <0.001<0.001<0.001<0.001
N48804880488048804880488048804880
AK_CountPearson Correlation−00.0170.200 **010.000 **10.063 **0.139 **0.173 **
Sig. (2-tailed)00.243<0.00100.999<0.001 <0.001<0.001<0.001
N48804880488048804880488048804880
RA_CountPearson Correlation−00.0190.057 **00.0050.063 **0.063 **1 −00.032 *0.003
Sig. (2-tailed)00.001<0.00100.722<0.001<0.001 00.0240.85
N48804880488048804880488048804880
Number of PagesPearson Correlation0.086 **0.188 **−00.0090.139 **0.139 ** −0.032 *10.601 **
Sig. (2-tailed)00.001<0.00100.508<0.001<0.00100.024 <0.001
N48804880488048804880488048804880
Cited Reference CountPearson Correlation0.084 **0.255 **0.029 *0.173 **0.173 **00.0030.601 **1
Sig. (2-tailed)<0.001<0.00100.044<0.001<0.00100.85<0.001
N48804880488048804880488048804880
* Correlation is significant at the 0.05 level (2 tailed). ** Correlation is significant at the 0.01 level (2 tailed).
Table A3. Coefficients.
Table A3. Coefficients.
Unstandardized CoefficientsStandardized Coefficients
Beta		tSig.95.0% Confidence Interval for B
ModelBStd. ErrorLower BoundUpper Bound
(Constant)4696.708480.892 9.767<0.0013753.9425639.474
A_Count−0.5680.435−0.018−1.3040.192−1.4210.286
RA_Count−1.2262.176−0.012−0.5640.573−5.4923.039
AK_Count−0.2180.435−0.007−0.5010.616−1.0700.635
WC_Count−1.5491.756−0.019−0.8820.378−4.9921.893
Number of Pages−0.0030.1280.000−0.0200.984−0.2530.247
Publication Year−2.3190.238−0.145−9.730<0.001−2.786−1.851
Cited Reference Count0.1850.0280.1216.667<0.0010.1310.239
(Constant)4696.708480.892 9.767<0.0013753.9425639.474
A_Count−0.5680.435−0.018−1.3040.192−1.4210.286
RA_Count−1.2262.176−0.012−0.5640.573−5.4923.039
AK_Count−0.2180.435−0.007−0.5010.616−1.0700.635
WC_Count−1.5491.756−0.019−0.8820.378−4.9921.893
Number of Pages−0.0030.1280.000−0.0200.984−0.2530.247
Publication Year−2.3190.238−0.145−9.730<0.001−2.786−1.851
Cited Reference Count0.1850.0280.1216.667<0.0010.1310.239
Table A4. Coefficient correlations.
Table A4. Coefficient correlations.
Model Publication YearNumber of PagesWC_CountCited ReferenceRA
Count		A_
Count		AK_Count
CorrelationsCited Reference Count−0.160−0.5740.0521.000−0.6000.028−0.080
RA_Count−0.016−0.003−0.758−0.6001.000−0.220−0.055
A_Count0.014−0.0090.0260.028−0.0221.000−0.006
AK_Count−0.166−0.0400.008−0.800−0.055−0.0061.000
Publication Year1.000−0.0320.081−0.160−0.1600.014−0.166
Number of Pages−0.0321.0000.0410.574−0.003−0.009−0.400
WC_Count0.0810.0411.0000.052−0.7580.0260.008
CovariancesCited Reference Count−0.001−0.0020.0030.001−0.0040.000−0.001
RA_Count−0.009−0.001−2.895−0.0044.735−0.021−0.052
A_Count0.001−0.0010.0200.000−0.0210.190−0.001
AK_Count−0.017−0.0020.006−0.001−0.052−0.0010.189
Publication Year0.057−0.0010.034−0.001−0.0090.001−0.170
Number of Pages−0.0010.0160.009−0.002−0.001−0.001−0.002
WC_Count0.0340.0093.0840.003−2.8950.0200.006

References

  1. Habil, S.; El-Deeb, S.; El-Bassiouny, N. AI-Based Recommendation Systems: The ultimate solution for market prediction and targeting. In Springer eBooks; Springer: Cham, Switzerland, 2023; pp. 683–704. [Google Scholar]
  2. Jeong, J.; Hong, D.; Youm, S. Optimization of the Decision-Making System for advertising strategies of small Enterprises—Focusing on Company A. Systems 2022, 10, 116. [Google Scholar] [CrossRef]
  3. Yu, Y. The Role and Influence of Artificial Intelligence on Advertising Industry. In Advances in Social Science, Education and Humanities Research; Atlantis Press: Dordrecht, The Netherlands, 2022. [Google Scholar]
  4. Ford, J.; Jain, V.; Wadhwani, K.; Gupta, D.G. AI advertising: An overview and guidelines. J. Bus. Res. 2023, 166, 114124. [Google Scholar] [CrossRef]
  5. Salesforce. State of the Connected Customer Report. 2022. Available online: https://www.salesforce.com/form/conf/socc2022-report/ (accessed on 5 June 2025).
  6. SPTechUSA. AI and Salesforce: How Artificial Intelligence Is Shaping CRM. 2025. Available online: https://sptechusa.com/blog/ai-and-salesforce-how-artificial-intelligence-is-shaping-crm (accessed on 22 September 2025).
  7. Linden, G.; Smith, B.; York, J. Amazon.com recommendations: Item-to-item collaborative filtering. IEEE Internet Comput. 2003, 7, 76–80. [Google Scholar] [CrossRef]
  8. Technology; Irb Trends And Reports. 2024 Artificial Intelligence in Marketing Market: Anticipated to Reach ~USD 123320.38 Million by 2032, with Rising CAGR of 18.94%|104 Pages Report. Available online: https://www.linkedin.com/pulse/2024-artificial-intelligence-marketingmarket-nyace (accessed on 5 June 2025).
  9. Argan, M.; Dinç, H.; Kaya, S.; Argan, M.T. Artificial Intelligence (AI) in Advertising. Adv. Distrib. Comput. Artif. Intell. J. 2022. [Google Scholar] [CrossRef]
  10. Huh, J.; Nelson, M.R.; Russell, C.A. ChatGPT, AI Advertising, and Advertising Research and Education. J. Advert. 2023, 52, 477–482. [Google Scholar] [CrossRef]
  11. Enache, M.C. AI for advertising. Ann. Dunarea Jos Univ. Galati Fasc. I Econ. Appl. Inform. 2020, 26, 28–32. [Google Scholar] [CrossRef]
  12. Haleem, A.; Javaid, M.; Qadri, M.A.; Singh, R.P.; Suman, R. Artificial intelligence (AI) applications for marketing: A literature-based study. Int. J. Intell. Netw. 2022, 3, 119–132. [Google Scholar] [CrossRef]
  13. De Bruyn, A.; Viswanathan, V.; Beh, Y.S.; Brock, J.K.-U.; von Wangenheim, F. Artificial intelligence and marketing: Pitfalls and opportunities. J. Interact. Mark. 2020, 51, 91–105. [Google Scholar] [CrossRef]
  14. Rahman, W.F.W.A.; Che Fauzi, A.A.; Wan Husain, W.S.; Che Hassan, S.H.; Nik Kamaruzaman, N.U.S.; Wan Abdul Aziz, W.N.H. The usage of artificial intelligence in marketing automation: Potentials and pitfalls. J. Math. Comput. Sci. 2020, 6, 1–8. [Google Scholar]
  15. Järvi, K.; Almpanopoulou, A.; Ritala, P. Organization of knowledge ecosystems: Prefigurative and partial forms. Res. Policy 2018, 47, 1523–1537. [Google Scholar] [CrossRef]
  16. Kandoth, S.; Shekhar, S.K. Scientometric visualization of data on artificial intelligence and marketing: Analysis of trends and themes. Sci. Talks 2024, 9, 100309. [Google Scholar] [CrossRef]
  17. Sorenson, O.; Fleming, L. Science as a vocation: Knowledge ecosystems and the organization of science. Res. Policy 2020, 49, 103994. [Google Scholar]
  18. Vodă, A.I.; Bortoș, S.; Şoitu, D.T. Knowledge ecosystem: A sustainable theoretical approach. Eur. J. Sustain. Dev. 2023, 12, 47. [Google Scholar] [CrossRef]
  19. Fragidis, G.; Mavridis, A.; Vontas, A.; Koumpis, A.; Tarabanis, K. A proposed conceptual framework for the study of research ecosystems. In Proceedings of the 2008 2nd IEEE International Conference on Digital Ecosystems and Technologies, Phitsanuloke, Thailand, 26–29 February 2008; IEEE: New York, NY, USA, 2008. [Google Scholar]
  20. Altman, M.; Cohen, P.N. The scholarly knowledge ecosystem: Challenges and opportunities for the field of information. Front. Res. Metr. Anal. 2022, 6, 751553. [Google Scholar] [CrossRef]
  21. Scott, W.R.; Davis, G.F. Organizations and Organizing: Rational, Natural and Open Systems Perspectives; Routledge: New York, NY, USA, 2007. [Google Scholar]
  22. Gulati, R.; Puranam, P.; Tushman, M. Meta-organization design: Rethinking design in interorganizational and community contexts. Strateg. Manag. J. 2012, 33, 571–586. [Google Scholar] [CrossRef]
  23. Ahrne, G.; Brunsson, N. Organizations and meta-organizations. Scand. J. Manag. 2005, 21, 429–449. [Google Scholar] [CrossRef]
  24. Park, H.W.; Stek, P. Measuring helix interactions in the context of economic development and public policies: From triple to quadruple and N-tuple helix vs. N-tuple and quadruple helix to triads. Triple Helix 2022, 9, 43–53. [Google Scholar] [CrossRef]
  25. Lee, H. A study on knowledge structure and cognitive mapping of marketing using social network analysis. J. Glob. Sch. Mark. Sci. 2013, 24, 39–64. [Google Scholar] [CrossRef]
  26. Stancu, A.; Panait, M. Marketing strategy metamorphosis under the impact of artificial intelligence services. Systems 2025, 13, 227. [Google Scholar] [CrossRef]
  27. Mustak, M.; Salminen, J.; Plé, L.; Wirtz, J. Artificial intelligence in marketing: Topic modeling, scientometric analysis, and research agenda. J. Bus. Res. 2021, 124, 389–404. [Google Scholar] [CrossRef]
  28. Wisetsri, W. Systematic analysis and future research directions in artificial intelligence for marketing. Turk. Comput. Math. Educ. J. 2021, 12, 43–55. [Google Scholar]
  29. Verma, S.; Sharma, R.; Deb, S.; Maitra, D. Artificial intelligence in marketing: Systematic review and future research direction. Int. J. Inf. Manag. Data Insights 2021, 1, 100002. [Google Scholar] [CrossRef]
  30. Hagendorff, T. Mapping the ethics of generative AI: A comprehensive scoping review. Minds Mach. 2024, 34, 39. [Google Scholar] [CrossRef]
  31. Al-Kfairy, M.; Mustafa, D.; Kshetri, N.; Insiew, M.; Alfandi, O. Ethical challenges and solutions of generative AI: An interdisciplinary perspective. Informatics 2024, 11, 58. [Google Scholar] [CrossRef]
  32. Wach, K.; Duong, C.D.; Ejdys, J.; Kazlauskaitė, R.; Korzynski, P.; Mazurek, G.; Paliszkiewicz, J.; Ziemba, E. The dark side of generative artificial intelligence: A critical analysis of controversies and risks of ChatGPT. Entrep. Bus. Econ. Rev. 2023, 11, 7–30. [Google Scholar] [CrossRef]
  33. Wang, X.; Wu, Y.C. Balancing innovation and regulation in the age of generative artificial intelligence. J. Inf. Policy 2024, 14, 385–416. [Google Scholar] [CrossRef]
  34. Gkikas, D.C.; Theodoridis, P.K. Artificial intelligence (AI) impact on digital marketing research. In Strategic Innovative Marketing and Tourism; Springer: Cham, Switzerland, 2019; pp. 1251–1259. [Google Scholar]
  35. Shaik, M. Impact of artificial intelligence on marketing. East Asian J. Multidiscip. Res. 2023, 2, 993–1004. [Google Scholar] [CrossRef]
  36. Rodgers, S. Why we need better measures of research impact in advertising. J. Advert. Res. 2018, 58, 385–389. [Google Scholar] [CrossRef]
  37. So, M.; Kim, J.; Choi, S.; Park, H.W. Factors affecting citation networks in science and technology: Focused on non-quality factors. Qual. Quant. 2014, 49, 1513–1530. [Google Scholar] [CrossRef]
  38. Zhao, D.; Strotmann, A. Bibliometric factor maps for knowledge discovery in digital libraries. In Proceedings of the International Conference on Electronic Publishing (ELPUB 2009), Milan, Italy, 10–12 June 2009; pp. 501–512. [Google Scholar]
  39. Phillips, F. Judgment, AI, and statistics. ROSA J. 2025; online first. [Google Scholar]
  40. Amaral, M.; Cai, Y.; Perazzo, A.R.; Rapetti, C.; Piqué, J.M. The Legacy of Loet Leydesdorff to the Triple Helix as a Theory of Innovation. Triple Helix 2024, 10, 161–187. [Google Scholar] [CrossRef]
  41. Barnett, G.A.; Huh, C.; Kim, Y.; Park, H.W. Citations among Communication Journals and Other Disciplines: A Network Analysis. Scientometrics 2011, 88, 449–469. [Google Scholar] [CrossRef]
  42. Donthu, N.; Kumar, S.; Mukherjee, D.; Pandey, N.; Lim, W.M. How to Conduct a Bibliometric Analysis: An Overview and Guidelines. J. Bus. Res. 2021, 133, 285–296. [Google Scholar] [CrossRef]
  43. He, W. Research hotspot and prospect of artificial intelligence in marketing in China—Visual analysis based on CiteSpace. BCP Bus. Manag. 2022, 20, 472–478. [Google Scholar] [CrossRef]
  44. Chen, J.; Ablanedo-Rosas, J.H.; Frankwick, G.L.; Jiménez Arévalo, F.R. The state of artificial intelligence in marketing with directions for future research. Int. J. Bus. Intell. Res. 2022, 12, 1–26. [Google Scholar] [CrossRef]
  45. Kejžar, N.; Korenjak Černe, S.; Batagelj, V. Network analysis of works on clustering and classification from Web of Science. In Studies in Classification, Data Analysis, and Knowledge Organization; Springer: Berlin/Heidelberg, Germany, 2010; pp. 525–536. [Google Scholar]
  46. Khan, G.F.; Lee, S.; Park, J.Y.; Park, H.W. Theories in Communication Science: A Structural Analysis Using Webometrics and Social Network Approach. Scientometrics 2016, 108, 531–557. [Google Scholar] [CrossRef]
  47. Gamit, R.M.; Chaudhari, S.P.; Trivedi, D.; Mandalia, S.; Vanik, H.; Chak, A.D. Sixteen Years of Research Journey of Webology: Bibliometric Visualization Approach. Towards Excell. 2022, 14, 1306–1324. [Google Scholar] [CrossRef]
  48. Lim, C.V.; Park, H.W. Evaluating the Functionality and Integration of AI Tools in VOSviewer for Enhancing Research Outcomes. ROSA J. 2025, 2, 13–20. [Google Scholar] [CrossRef]
  49. Van Eck, N.J.; Waltman, L. Software Survey: VOSviewer, a Computer Program for Bibliometric Mapping. Scientometrics 2009, 84, 523–538. [Google Scholar] [CrossRef]
  50. Social Media Research Foundation. The Home of NodeXL: Your Social Network Analysis Tool for Social Media. Available online: https://www.smrfoundation.org/ (accessed on 5 June 2025).
  51. Nazar, R.; Lindemann, D. Terminology extraction using co-occurrence patterns as predictors of semantic relevance. In Proceedings of the Workshop on Terminology in the 21st Century: Many Faces, Many Places, Marseille, France, 20–25 June 2022. [Google Scholar]
  52. Mbaye, R.; Gebeyehu, R.; Hossmann, S.; Mbarga, N.; Bih-Neh, E.; Eteki, L.; Ohene-Agyei, T.; Oyerinde, A.; Kiti, G.; Mburu, Y.; et al. Who Is Telling the Story? A Systematic Review of Authorship for Infectious Disease Research Conducted in Africa, 1980–2016. BMJ Glob. Health 2019, 4, e001855. [Google Scholar] [CrossRef]
  53. Dimitris, M.C.; Gittings, M.; King, N.B. How Global Is Global Health Research? A Large-Scale Analysis of Trends in Authorship. BMJ Glob. Health 2021, 6, e003758. [Google Scholar] [CrossRef]
  54. Alzboon, M.S.; Aljarrah, E.; Alqaraleh, M.; Alomari, S.A. NodeXL tool for social network analysis. Turk. J. Comput. Math. Educ. 2021, 12, 202–216. [Google Scholar]
  55. Brandes, U. A Faster Algorithm for Betweenness Centrality. J. Math. Sociol. 2001, 25, 163–177. [Google Scholar] [CrossRef]
  56. Li, C.; Li, Q.; Van Mieghem, P.; Stanley, H.E.; Wang, H. Correlation Between Centrality Metrics and Their Application to the Opinion Model. Eur. Phys. J. B 2015, 88, 65. [Google Scholar] [CrossRef]
  57. Kourtellis, N.; Alahakoon, T.; Simha, R.; Iamnitchi, A.; Tripathi, R. Identifying High Betweenness Centrality Nodes in Large Social Networks. Soc. Netw. Anal. Min. 2012, 3, 899–914. [Google Scholar] [CrossRef]
  58. Wandelt, S.; Shi, X.; Sun, X. Complex Network Metrics: Can Deep Learning Keep up with Tailor-Made Reference Algorithms? IEEE Access 2020, 8, 68114–68123. [Google Scholar] [CrossRef]
  59. Wartena, C.; Brussee, R. Topic detection by clustering keywords. In Proceedings of the 2008 19th International Workshop on Database and Expert Systems Applications, Turin, Italy, 1–5 September 2008; pp. 54–58. [Google Scholar]
  60. Oh, K.; Lim, C.-G.; Kim, S.S.; Choi, H.-J. Research trend analysis using word similarities and clusters. Int. J. Multimed. Ubiquitous Eng. 2013, 8, 185–196. [Google Scholar]
  61. dos Santos, C.K.; Evsukoff, A.G.; de Lima, B.S.L.P. Cluster analysis in document networks. In Data Mining IX; WIT Press: Southampton, UK, 2008. [Google Scholar]
  62. Toivonen, R.; Kivelä, M.; Saramäki, J.; Viinikainen, M.; Vanhatalo, M.; Sams, M. Networks of Emotion Concepts. PLoS ONE 2012, 7, e28883. [Google Scholar] [CrossRef]
  63. Qin, X.; Jiang, Z. The Impact of AI on the Advertising Process: The Chinese Experience. J. Advert. 2019, 48, 338–346. [Google Scholar] [CrossRef]
  64. Fulton, R.; Fulton, D.; Hayes, N.; Kaplan, S. The Transformation Risk-Benefit Model of Artificial Intelligence: Balancing Risks and Benefits Through Practical Solutions and Use Cases. arXiv 2024. [Google Scholar] [CrossRef]
  65. Phan, Q.A.; Ho, M.-T.; Vuong, Q.-H.; Pham, H.-H.; Vu, M.H.; Nguyen, T.T.H.; Phan, T.T.T. Science Communication Matters: An Exploratory Study of Academic Public Engagement in Vietnam Using Bayesian Statistics. J. Contemp. East. Asia 2024, 23, 35–57. [Google Scholar]
  66. Isfandyari-Moghaddam, A.; Saberi, M.K.; Tahmasebi-Limoni, S.; Mohammadian, S.; Naderbeigi, F. Global Scientific Collaboration: A Social Network Analysis and Data Mining of the Co-Authorship Networks. J. Inf. Sci. 2021, 49, 1126–1141. [Google Scholar] [CrossRef]
  67. Neumann, C. China’s New Push toward Research, Innovation, and AI: From the World’s Factory to the World’s Research Lab? Available online: https://www.linkedin.com/pulse/chinas-new-push-toward-research-innovation-ai-from-worlds-neumann-arsse (accessed on 5 June 2025).
  68. Li, D. Is China Emerging as the Global Leader in AI? Harvard Business Review. 18 February 2021. Available online: https://hbr.org/2021/02/is-china-emerging-as-the-global-leader-in-ai (accessed on 5 June 2025).
  69. Zhou, J.; Theesfeld, C.L.; Yao, K.; Chen, K.M.; Wong, A.K.; Troyanskaya, O.G. Deep Learning Sequence-Based Ab Initio Prediction of Variant Effects on Expression and Disease Risk. Nat. Genet. 2018, 50, 1171–1179. [Google Scholar] [CrossRef] [PubMed]
  70. Strycharz, J.; Segijn, C.M. The Future of Dataveillance in Advertising Theory and Practice. J. Advert. 2022, 51, 574–591. [Google Scholar] [CrossRef]
  71. Kamila, M.K.; Jasrotia, S.S. Ethical Issues in the Development of Artificial Intelligence: Recognizing the Risks. Int. J. Ethics Syst. 2023, 41, 45–63. [Google Scholar] [CrossRef]
  72. Oakes, M. AI: The End of Corporate Communications as We Know It? Available online: https://www.linkedin.com/pulse/ai-end-corporate-communications-we-know-michael-oakes/ (accessed on 5 June 2025).
  73. Davenport, T.; Guha, A.; Grewal, D.; Bressgott, T. How Artificial Intelligence Will Change the Future of Marketing. J. Acad. Mark. Sci. 2020, 48, 24–42. [Google Scholar] [CrossRef]
  74. Wagner, C.S.; Park, H.W.; Leydesdorff, L. The Continuing Growth of Global Cooperation Networks in Research: A Conundrum for National Governments. PLoS ONE 2015, 10, e0131816. [Google Scholar] [CrossRef]
  75. Das, M.R. India’s AI Market to Reach $17 Billion by 2027 as Investments, Talent Surge, Reveals Report. Firstpost. 21 February 2024. Available online: https://www.firstpost.com/tech/indias-ai-market-to-reach-17-billion-by-2027-as-investment-talent-surge-reveals-report-13740117.html (accessed on 5 June 2025).
  76. Kozyreva, A.; Herzog, S.M.; Lorenz-Spreen, P.; Hertwig, R.; Lewandowsky, S. Artificial Intelligence in Online Environments: Representative Survey of Public Attitudes in Germany. Mpg Open Sci. 1 January 2020. Available online: https://pure.mpg.de/rest/items/item_3188061_10/component/file_3195148/content (accessed on 5 June 2025).
  77. Dwivedi, Y.K.; Hughes, L.; Ismagilova, E.; Aarts, G.; Coombs, C.; Crick, T.; Duan, Y.; Dwivedi, R.; Edwards, J.; Eirug, A.; et al. Artificial Intelligence (AI): Multidisciplinary Perspectives on Emerging Challenges, Opportunities, and Agenda for Research, Practice and Policy. Int. J. Inf. Manag. 2021, 57, 101994. [Google Scholar] [CrossRef]
  78. van Reijmersdal, E.A.; Rozendaal, E.; Smink, N.; van Noort, G.; Buijzen, M. Processes and Effects of Targeted Online Advertising among Children. Int. J. Advert. 2017, 36, 396–414. [Google Scholar] [CrossRef]
  79. Mogaji, E.; Olaleye, S.; Ukpabi, D. Using AI to personalise emotionally appealing advertisement. In Digital and Social Media Marketing; Springer: Cham, Switzerland, 2019; pp. 137–150. [Google Scholar]
  80. Deng, S.; Tan, C.-W.; Wang, W.; Pan, Y. Smart Generation System of Personalized Advertising Copy and Its Application to Advertising Practice and Research. J. Advert. 2019, 48, 356–365. [Google Scholar] [CrossRef]
  81. Voorveld, H.A.M.; Meppelink, C.S.; Boerman, S.C. Consumers’ Persuasion Knowledge of Algorithms in Social Media Advertising: Identifying Consumer Groups Based on Awareness, Appropriateness, and Coping Ability. Int. J. Advert. 2023, 43, 960–986. [Google Scholar] [CrossRef]
  82. Sands, S.; Campbell, C.; Ferraro, C.; Demsar, V.; Rosengren, S.; Farrell, J. Principles for advertising responsibly using generative AI. Organ. Dyn. 2024, 53, 101042. [Google Scholar] [CrossRef]
Systems 13 00859 g001
Figure 1. Multi-dimensional analytical framework—AI in advertising research. Source: Author’s work.
Figure 1. Multi-dimensional analytical framework—AI in advertising research. Source: Author’s work.
Systems 13 00859 g001
Systems 13 00859 g002
Figure 2. Co-occurring terms network visualization.
Figure 2. Co-occurring terms network visualization.
Systems 13 00859 g002
Systems 13 00859 g003
Figure 3. Identifying strategic prospects and potential risks.
Figure 3. Identifying strategic prospects and potential risks.
Systems 13 00859 g003
Systems 13 00859 g004
Figure 4. Network visualization of co-authoring countries.
Figure 4. Network visualization of co-authoring countries.
Systems 13 00859 g004
Systems 13 00859 g005
Figure 5. Network visualization of co-authoring institutions.
Figure 5. Network visualization of co-authoring institutions.
Systems 13 00859 g005
Systems 13 00859 g006
Figure 6. Network visualization of citations.
Figure 6. Network visualization of citations.
Systems 13 00859 g006
Systems 13 00859 g007
Figure 7. Citation weight count.
Figure 7. Citation weight count.
Systems 13 00859 g007
Systems 13 00859 g008
Figure 8. Network visualization of bibliographic coupling.
Figure 8. Network visualization of bibliographic coupling.
Systems 13 00859 g008
Systems 13 00859 g009
Figure 9. Visual trend of published paper in publication years.
Figure 9. Visual trend of published paper in publication years.
Systems 13 00859 g009
Table 1. Terms clustering.
Table 1. Terms clustering.
ClusterCluster Name Concept and Themes
1Socioeconomic impact of technological advancementImpact on Work and Environment, Societal and Economic Transformation, Government and Policy Changes, International Collaboration, Human Elements
2Marketing and Advertising in the Age of AIConsumer Behavior and AI, AI Tools and Techniques, Effectiveness and Performance, Ethical Considerations and Societal Impact
3AI and Machine Learning in HealthcareMedical Imaging and Data Analysis, AI-powered Healthcare Tools and Systems, Ethical Considerations and Challenges
4AI and Machine Learning in Energy EnvironmentPrediction and Forecasting, Optimization and Efficiency, Tools and Systems Model Development and Evaluation
5Algorithmic Optimization and Challenges in Online AdvertisingAlgorithmic Techniques and Models, Performance Optimization and Metrics, Real-world Applications and Datasets, Challenges and Concerns, Evaluation and Comparisons
Table 2. Co-authorship of organization network analysis data.
Table 2. Co-authorship of organization network analysis data.
Unit of AnalysisRegions with Corresponding Data Points
No. of PublicationsChinese Mainland, Hong Kong, and Macao (26,090), USA (13, 307), India (8614), England (4924), Germany (4924)
No. of CitationsUSA (26, 090), Chinese Mainland and Hong Kong and Macao (13,307), England (8614), Germany (4924), Australia (4924)
Degree CentralityUSA (71), England (64), India (62), France (57), Chinese Mainland, Hong Kong, and Macao (56)
Betweenness CentralityUSA (331.78), England (213.104), India, (189.41), France (137.707), Chinese Mainland, Hong Kong, and Macao (134.968)
Closeness CentralityUSA (331.78), India (213.104), England (189.412), Canada (137.707), Chinese Mainland, Hong Kong, and Macao (134.968)
Table 3. Co-authorship of organizations network analysis data.
Table 3. Co-authorship of organizations network analysis data.
Unit of AnalysisOrganization with Corresponding Data Points
No. of PublicationsChinse Academy of Sciences (35), Tsinghua University (35), Shanghai Jiao Tong University (33), Alibaba Group (30), UCL (28)
No. of CitationsUniversity of California Berkeley (17), Carnegie Mellon University (17), MIT (21) Alibaba Group (30), The University of Queensland (12)
Degree CentralityHarvard University (40), Tsinghua University (38), MIT (32), TBS Business School (30), Newcastle University (29)
Betweenness CentralityTsinghua (5105.352), UCL (4447.872), Harvard University (4039.349), University of Zagreb (3598.482), Zhejiang University (3338.334)
Closeness CentralityTsinghua University (5105.352), UCL (4447.872), Harvard University (4039.349), University of Zagreb (3598.482), Zhejiang University (3338.334)
Table 4. Bibliographic coupling network analysis data.
Table 4. Bibliographic coupling network analysis data.
Unit of AnalysisPublicationData
Total Link Strength/CentralityPeltier (2024)1667
Hermann (2022)1561
Vlacic (2021)1434
Manis (2023)1422
Hermann (2023)1276
Table 5. Bibliometric analysis.
Table 5. Bibliometric analysis.
FactorsWeb of Science Available FactorsQuantitative MeasurementsFactor Selection
Author Attribute
Countryox
Authorox
Number of Co-authorsoo•
Number of Co-affiliationsoo•
Genderxx
Agexo
Careerxo
Number of Previous Studiesxo
Affiliation Attribute
Collaborationox
International Cooperationox
Funding▲x
Research Attribute
Languageox
Document Typeox•
Novelty of Titleox
Number of Cited Referenceoo•
Number of Figures and Tablesxo
Number of Pagesoo•
Number of Author Keywords oo•
Number of Citations (Times Cited, in all databases)oo⦾
Journal Characteristics
Reputation▲o
Research Areasox⋆
Year Publishedox⋆
Web of Science Categoriesox
Methods
Keyword Analysisxo
Co-Citationxo
Application Levelxx
Legend: (o) Possible. (x) Impossible. ▲ Possible coupling with another database. ⦾ Dependent variable. • Independent variable; ⋆ Test the moderating effect.
Table 6. Frequency analysis.
Table 6. Frequency analysis.
N = 4880
ClassificationFrequency%
Research Areas
Computer Science195340.02
Engineering117224.016
Business Economics105221.557
Telecommunications3356.865
Science Technology Other Topics2445
Countries
China Mainland, Hong Kong, and Macau98120.102
USA91818.811
India4459.119
England3727.623
Germany2625.369
Years
202295719.611
202395119.488
202174415.246
202053410.943
20194839.898
Table 7. Descriptive analysis of variables.
Table 7. Descriptive analysis of variables.
N = 4880MinimumMaximumMeanStd. Deviation
Times Cited, All Databases 0 4108 15.33 77.457
A_Count 1 35 3.57 2.516
G_Count 1 31 4.85 2.599
Number of Pages 1 239 14.19 10.780
Publication Year 1990 2024 2019.61 4.845
AK_Count 1 31 4.85 2.599
RA_Count 1 6 1.58 0.775
Cited Reference Count 0 658 51.09 50.453
Times Cited, WoS Core 0 3703 14.23 70.042
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Lim, C.V.; Park, H.-W. Mapping Impact and Influence in AI-Driven Advertising: A Scientometric and Network Analysis of Knowledge Ecosystem. Systems 2025, 13, 859. https://doi.org/10.3390/systems13100859

AMA Style

Lim CV, Park H-W. Mapping Impact and Influence in AI-Driven Advertising: A Scientometric and Network Analysis of Knowledge Ecosystem. Systems. 2025; 13(10):859. https://doi.org/10.3390/systems13100859

Chicago/Turabian Style

Lim, Camille Velasco, and Han-Woo Park. 2025. "Mapping Impact and Influence in AI-Driven Advertising: A Scientometric and Network Analysis of Knowledge Ecosystem" Systems 13, no. 10: 859. https://doi.org/10.3390/systems13100859

APA Style

Lim, C. V., & Park, H.-W. (2025). Mapping Impact and Influence in AI-Driven Advertising: A Scientometric and Network Analysis of Knowledge Ecosystem. Systems, 13(10), 859. https://doi.org/10.3390/systems13100859

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop