Scientific Mapping of the Field of Artificial Intelligence and Information Seeking Behavior: Analysis of Trends and Topic Clusters in the Web of Science
Pages 1-27
https://doi.org/10.22054/jks.2026.89953.1754
Yeganeh Dastar, mohammadreza khorramabadi arani, Ahmad Shabani, Fatemeh Darestani Farahani
Abstract Introduction
Information seeking behavior (ISB) is a complex, multidimensional process at the intersection of cognitive, affective, and technological dimensions. Traditional models describe ISB as a conscious, goal-oriented process. However, the advent of Artificial Intelligence (AI), including large language models, recommender systems, and conversational agents, has fundamentally transformed this paradigm. AI has shifted ISB from reactive, keyword-based searching to proactive, conversational, and algorithm-driven interactions. Despite these shifts, there is a lack of comprehensive, quantitative mapping of the global scientific structure at the intersection of AI and ISB. This study addresses this gap by employing scientometrics to map the intellectual structure, evolutionary trends, and emerging thematic clusters of this domain.
Research Questions
1. Which countries, institutions, authors, and journals have the highest scientific output in the domain of “AI in the transformation of ISB”?
2. What are the temporal trends, language distribution, and annual growth of publications from 1976 to 2025?
3. What is the conceptual structure, keyword co-occurrence network, and thematic clustering of this research field?
Literature Review
Foundational ISB models emphasize cognitive and affective stages, highlighting uncertainty and information anxiety. Later systemic models incorporated social and organizational barriers. In the digital era, scholars note that AI introduces new concepts such as “hidden information needs” and “conversational information interaction.” While previous studies have qualitatively explored specific AI systems, they lack a macro-level quantitative analysis. Scientometric approaches, utilizing databases like Web of Science (WoS), offer robust tools for mapping knowledge domains, identifying core contributors, and revealing hidden evolutionary patterns in scientific production, which this study leverages.
Methodology
This applied research utilized a scientometric approach. The statistical population comprised all documents indexed in the Web of Science (WoS) core collection, extracted on January 15, 2025, covering the period from 1976 to 2025. The search strategy combined AI-related terms with ISB terms. The initial search yielded 2, 147 records. After data cleaning (including removing duplicates, non-research documents, and standardizing author names) 1, 902 records were retained. Data analysis and visualization were conducted using Excel and R software with the bibliometrix package, employing co-word analysis, co-authorship networks, and citation analysis.
Results
The United States led scientific production with 392 records, followed by China (324) and the UK (162). Harvard University was the top institution (33 articles), followed by the University of Toronto (28). “IEEE” and “Lecture Notes in Computer Science” were the most prolific sources (56 articles each). Trend Analysis: English dominated the literature (98.26%). The publication trend exhibited two distinct phases: a slow growth phase (1976–2020, comprising 31% of outputs) and an exponential acceleration phase post-2020. The peak occurred in 2024 with 393 documents. Conceptual Structure: Computer Science was the dominant subject category (965 articles), while Information Science accounted for only 161. Co-word analysis revealed seven distinct thematic clusters. Cluster 1 focused on AI in digital health, trust, and bias. Cluster 2 covered deep learning. Cluster 3 centered on natural language processing. Cluster 4 (16 keywords) related to data mining applications, precision, feature extraction, machine learning, neural networks, information retrieval, and visualization. Cluster 5 highlighted generative AI and ethics. Cluster 6 (5 keywords) focused on recommender systems and personalization techniques such as collaborative filtering and case-based reasoning. Cluster 7 (3 keywords) related to algorithms and the music information retrieval process.
Discussion
The findings revealed that AI is no longer merely a retrieval tool but an active agent shaping user behavior, trust, and cognitive processes. The exponential growth post-2020 aligned with the rise of conversational AI. However, a significant disciplinary imbalance exists; the field is heavily dominated by computer science and engineering, while the human, cognitive, and ethical dimensions (traditionally the domain of information science and psychology) are marginalized. This technological determinism risks creating systems that are highly efficient but lack human-centric design. Furthermore, the geographical concentration of research in Western institutions highlighted a global scientific divide. The emergence of clusters focusing on “ethics” and “bias” indicated a growing awareness of socio-technical challenges.
Conclusion
The intersection of AI and ISB is a rapidly evolving, interdisciplinary field transitioning from traditional search to proactive, generative interactions. To ensure these technologies serve humanity effectively, future research must pivot towards three main axes: developing ethical frameworks for transparent AI systems, investigating the long-term cognitive impacts of AI on vulnerable groups like students, and fostering deep interdisciplinary collaborations. Only by integrating computer science with information science, psychology, and ethics can we design AI-driven information systems that are not only technologically robust but also socially equitable and human-centric.
Ethical Considerations
Author Contributions
All authors contributed equally to the conceptualization of the article and writing of the original and subsequent drafts.
Data Availability Statement
Data are available upon request from the authors.
Acknowledgements
Not applicable.
Ethical Considerations
The authors avoided data fabrication, falsification, plagiarism, and misconduct. As this study utilized publicly available, anonymized bibliometric data, no ethical approval was required.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Conflict of Interest
“The authors declare no conflict of interest.”















