The Impact of International Collaborations with Highly Publishing Countries in Computer Science

Alberto Gómez-Espés, Michael Färber, Adam JatowtarXiv 2025 · DOI · arXiv

Introduction

In the rapidly evolving landscape of Computer Science (CS), international collaboration has long been recognized as a catalyst for innovation. However, the specific dynamics of who collaborates with whom—and the resulting impact on research quality—have remained largely anecdotal. A recent large-scale bibliometric study reveals a global research ecosystem dominated by three "Publishing Superpowers": China, the European Union (EU), and the United States. This research demonstrates that these regions act as gatekeepers of scientific integrity and impact, with collaborations involving them yielding significantly higher citations and lower retraction rates.

The Hegemony of Publishing Superpowers

Analyzing over 3.1 million research papers from the OpenAlex database published between 1990 and 2021, the study identifies a stark concentration of research activity. Either China, the EU, or the US participated in a staggering 90.15% of all international CS collaborations.

To better understand these dynamics, the researchers categorized papers into three distinct relation groups:

  1. US-EU-CN: Collaborations exclusively involving these three regions.
  2. Mixed: Collaborations between at least one superpower and other countries.
  3. Other Countries: Collaborations that excluded the three superpowers entirely.

While the "Mixed" group accounted for nearly half of all publications (49.18%), suggesting an active outreach from superpowers to developing regions, the highest citation impact remains concentrated within the US-EU-CN bubble.

Scientific Integrity and Citation Impact

The study highlights a clear correlation between collaboration patterns and research quality. Papers produced within the US-EU-CN cluster averaged 27.02 citations, nearly double the 15.26 citations seen in papers involving only "Other" countries.

Perhaps more critically, international collaboration with superpowers appears to serve as a safeguard for scientific integrity. The retraction rate for the "Other Countries" group was 0.12%—four times higher than the 0.03% rate observed in the US-EU-CN cluster. This suggests that the established research infrastructures, rigorous peer-review traditions, and institutional scrutiny prevalent in the superpowers help mitigate errors and misconduct before or after publication.

Socio-Economic Factors: The HDI Proxy

Beyond geographic regions, the study utilized the United Nations' Human Development Index (HDI) to assess how a country’s socio-economic status influences its research output. The findings indicate that HDI is a potent predictor of research quality. Papers involving at least one "Very High Developed" country received 46% more citations than those without such a partner. This suggests that the resources, education levels, and institutional stability reflected in high HDI scores translate directly into more impactful and reliable scientific contributions.

The Shifting Landscape of CS Topics

The research also tracked the evolution of CS topics over three decades. A pivotal shift occurred in 2017, when Artificial Intelligence (AI) surpassed Mathematics and Physics to become the primary focus of global computer science research. This "AI Pivot" was observed across all regions, signaling a universal shift toward machine learning and automated systems.

Furthermore, the data captured the immediate impact of global events on research priorities. Since 2019, there has been a notable spike in CS research related to Medicine. This trend was likely driven by the global response to the COVID-19 pandemic, which accelerated the intersection of bioinformatics, data science, and healthcare.

Conclusion and Future Directions

The dominance of the US, EU, and China in computer science research creates a high-impact "inner circle" that benefits from superior citation rates and higher integrity standards. While the "Mixed" collaboration group shows that knowledge is being shared globally, the disparity in retraction rates and citations suggests that developing regions still face significant hurdles in achieving the same level of perceived quality.

Future research is needed to determine whether the lower retraction rates in superpower-led collaborations are due to better error detection or a lower incidence of deliberate fraud. For now, the data is clear: in the world of computer science, who you collaborate with is just as important as what you research.

Summary updated

BibTeX
@misc{gomezespes2025impact,
  author = {Alberto Gómez-Espés and Michael Färber and Adam Jatowt},
  title = {{The Impact of International Collaborations with Highly Publishing Countries in Computer Science}},
  year = {2025},
  eprint = {2505.09776},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2505.09776},
}
Figure