Abstract
Purpose: This study investigates the structural patterns of scientific influence among highly cited publications by Iranian researchers in Computer Science using Ego-Centered Citation Networks (ECCNs). Moving beyond conventional citation-based indicators, it focuses on dependency and independence metrics to capture qualitative dimensions of scientific impact, including intellectual cohesion, convergence, dispersion, and the relative autonomy of citation flows.
Methodology: This applied study used network-based scientometric indicators within a quantitative–analytical design. The study population consisted of highly cited papers authored by Iranian computer science researchers who, according to the Essential Science Indicators (ESI), were ranked among the top 1% worldwide. In total, 45 researchers and 219 highly cited papers were identified. The papers were published between 2013 and 2023, and their citation data were retrieved from the Web of Science Core Collection (WOSCC). Data required to construct the ECCNs were collected and stored through web scraping using Mimfa Scraper and JavaScript. The extracted data covered bibliographic information for focal papers (egos), references cited by each ego, citations received by the ego (alters), and references cited by the alters. Altogether, 1,886,000 records were extracted. Network processing and indicator computation were performed using Mimfa Ravar Datalab and C#. Absolute dependency was defined as the number of alters that cited the ego’s reference sources; relative dependency as the ratio of this number to all alters; absolute independence as the number of alters that did not cite the ego’s sources; and relative independence as the ratio of this number to all alters. Descriptive statistical analyses and scatterplot-based examinations of indicator distributions were conducted in EXCEL, while network and chart visualizations were produced using Gephi and Flourish.
Findings: The Ego-Centered Citation Networks of highly cited Iranian Computer Science papers exhibit diverse patterns of scientific influence in terms of dependency and independence. Papers with higher dependency values, particularly those in domains such as metaheuristic optimization algorithms and medical imaging, exhibited more cohesive citation structures and more concentrated intellectual trajectories. In these cases, citing papers not only referred to the focal paper but also substantially overlapped with its reference base, indicating shared theoretical and methodological foundations around the ego paper. Such configurations suggest a concentrated form of scientific influence in which a focal paper contributes to the consolidation of a research line, the stabilization of a methodological tradition, or the formation of a coherent citation cluster. By contrast, interdisciplinary and emerging papers, especially those situated at the intersection of artificial intelligence, health, energy systems, and the Internet of Things, displayed higher levels of independence. This pattern suggests that ideas introduced or synthesized by these focal papers may have diffused through more diverse and relatively independent citation pathways beyond the original reference base of the ego paper. The descriptive examination of scatterplots further indicated that papers receiving larger numbers of citations often showed higher values of either absolute dependency or absolute independence. This suggests that citation volume may be associated with two different forms of influence: in some cases, a higher number of citations is accompanied by stronger overlap with the ego’s foundational references; in other cases, it is accompanied by a broader set of independent citing paths. Accordingly, highly cited papers do not necessarily exert influence through a single structural mechanism. Some function as intellectual anchors within dense and cohesive clusters, whereas others operate as bridges across multiple and less tightly connected research areas.
Conclusion: This study examined 219 highly cited papers authored by 45 Iranian computer science researchers between 2013 and 2023 to identify structural patterns of dependency and independence within ECCNs. The results indicate that the scientific influence of highly cited Iranian Computer Science papers is multidimensional and cannot be adequately understood through citation counts alone. Dependency and independence indicators provide a more nuanced view of how a paper influences subsequent research by distinguishing between cohesive, reference-sharing citation structures and more dispersed, autonomous citation pathways. Papers with higher dependency values appear to contribute to concentrated and cumulative forms of influence, particularly in mature and specialized domains such as optimization and medical imaging. In these contexts, focal papers and their citing papers tend to draw on a shared intellectual base, suggesting continuity, convergence, and the possible emergence of paradigm-building research traditions or schools of thought. Conversely, papers with higher independence values, especially in interdisciplinary and emerging topics, appear to generate wider and more heterogeneous patterns of uptake. These papers may stimulate research beyond their original citation environment and support the horizontal diffusion of ideas across domains. Overall, dependency and independence metrics can complement traditional citation indicators by offering a more discriminating basis for identifying foundational works, intellectual leaders, interdisciplinary bridges, and early signals of innovation in Computer Science.