Science of science (Chair: MATTEO CINELLI)

Citation analysis is a standard bibliometric tool for assessing the impact of different papers, researchers, institutions, and countries [1]. Achieving fair citation analysis remains challenging due to the diverse citation practices across individual researchers and research disciplines. Indeed, the number of citations received by papers often exhibits imbalances in terms of the author's country of affiliation and gender. Gender imbalance persists in computer science across various aspects, including educational attainment, faculty representation, and career progression. While recent studies have highlighted citation imbalances related to authors’ gender in neuroscience and physics [2, 3], the extent of gendered citation imbalance in computer science remains largely unknown. Here, by deploying a family of reference models for citation networks, we quantify gender imbalance in citations between papers published in computer science conferences [4].

Scientific collaborations are shaped by many factors: common scientific interests, background, social interactions, availability, and so on. But what are the core mechanisms that drive collaborations? In our research, we focused on how researchers collaborate based on the similarity or complementarity of their skills. What we found is that these two mechanisms (similarity and complementarity) have distinct topological properties: similar collaborations have a higher tendency to form triangles since similarity is transitive and may lead to triangle closure, while complementary collaborations are characterized by the high density of quadrangles because complementarity leads to the formation of even-length motifs if the number of complementary domains is small. Our findings highlight the organization of collaborations in different fields of science. These results may help domain experts identify plausible new collaborations and also better understand the mid-scale structure of scientific fields, including communities ofcollaborators and relationships between scientific problems.

Scientific citations track the circulation of ideas and reflect the collective attention of the academic community. However, our understanding of how scientific horizons evolve amidst a surge in publications remains contradictory. To address this gap, we investigate the spatiotemporal scale of citation flows within the knowledge space, uncovering evidence of narrowing citation mobility and growing constraints of knowledge distance. These findings offer new insights into the diversity and efficiency of knowledge diffusion in science.Drawing on millions of papers in the field of physics, we apply the Doc2Vec algorithm leveraging the semantics of papers, to map the citation network into the embedded knowledge landscape of physics, quantifying the spatiotemporal dynamics of citation flows. We first observe an exponential distribution of citing distances, characterized by numerous short-distance flows and a few long-distance ones. Citation mobility is proximity-based, with diffusion likelihood decreasing as epistemic distance increases. Further simulations using the Gravity model confirm that epistemic distance and research popularity are key push-and-pull factors in citation mobility. To explore how citation mobility evolved, we group citation flows into different decades based on the citing date. Despite an increase in the volume of citation flows, we observe a decline in the proportion of cross-grid citation flows. Fitting citation flows by decade with Gravity models reveals a steady increase in the distance-damping exponent, indicating a growing barrier effect of distance on citing behavior over time. Finally, examining individual papers, we find that even as their citation counts increase, the breadth of their impact decreases over decades. Our results reflect increasingly localized and myopic information foraging in modern scientific practices. This trend underscores the urgent need to promote diverse knowledge exchange, ensuring long-term scientific innovation.

Prizes confer credibility to individuals, ideas, and disciplines; provide financial incentives; and foster community-building celebrations. Despite considerable efforts to study the influence of prizes on scientists and scientific growth, most research focuses on individual prizes. Little is known about the broader recognition ecosystem and how prizes interact with one another. Gaining insight into these interactions can significantly enhance our understanding of how collective recognition, fame, and credit are allocated, amplified, and disseminated within the scientific community. It can also help us predict future winners and emerging hotspots in science. In this study, we curated three large-scale datasets on prizes, documenting the prize-winning records of more than 13,000 recipients across over 6,000 prizes. This allowed us to construct a dynamical scientific prize network based on co-prizewinners and develop a minimal selection model that accurately captures both individual prize-winning trajectories and the collective behavior of the entire prize system. Our analysis further reveals a concerning trend: many prizes founded after 2000 disproportionately emphasize winners of already established prizes and ignoring new emerging talents. This overemphasis jeopardizes diversity and inclusivity in science, potentially stifling groundbreaking discoveries. By understanding and addressing these patterns, we can ensure that the scientific recognition system evolves to better support innovation and equity in the years to come.

There exists a growing discourse around the domination of Big Tech on the landscape of artificial intelligence (AI) research, yet our comprehension of this phenomenon remains cursory. This paper aims to broaden and deepen our understanding of Big Tech's reach and power within AI research. It highlights the dominance not merely in terms of sheer publication volume but rather in the propagation of new ideas or memes. Current studies often oversimplify the concept of influence to the share of affiliations in academic papers, typically sourced from limited databases such as arXiv or specific academic conferences.The main goal of this paper is to unravel the specific nuances of such influence, determining which AI ideas are predominantly driven by Big Tech entities. By employing network and memetic analysis on AI-oriented paper abstracts and their citation network, we are able to grasp a deeper insight into this phenomenon. By utilizing two databases: OpenAlex and S2ORC, we are able to perform such analysis on a much bigger scale than previous attempts.Our findings suggest that while Big Tech-affiliated papers are disproportionately more cited in some areas, the most cited papers are those affiliated with both Big Tech and Academia. Focusing on the most contagious memes, their attribution to specific affiliation groups (Big Tech, Academia, mixed affiliation) seems equally distributed between those three groups. This suggests that the notion of Big Tech domination over AI research is oversimplified in the discourse.