Selection in scientific networks

One of the most pressing and interesting actual scientific challenges deals with the analysis and the understanding of complex network dynamics. In particular, a major trend is the definition of new frameworks for the analysis, the exploration and the detection of the dynamics at play in real dynamic networks. In this paper, we focus in particular on scientific communities by targeting the social part of science through a descriptive approach that aims at identifying the social determinants behind the emergence and the resilience of scientific communities. We consider that scientific communities are at the same time through co-authorship communities of practice and that they exist also as representations in the scientists mind, since references to other scientists’ works are not merely an objective link to a relevant work, but they reveal also social objects that one manipulates and refers to. In fact, our analysis focuses on the coexistence of co-authorships and citation dynamics and how their interplay affects the shape, the strength and the stability of the scientific systems. Such an analysis—performed through the time-varying graphs (TVG) formalism and derived metrics—concerns the evolution of a scientific network extracted from a portion of the arXiv repository covering a period of 10 years of publications in physics. We detect an example of how the selection process of citations may affect the shape of the co-authorships network from a sparser and disconnected structure to a dense and homogeneous one.