A Two-Layer Network Model Reveals the Adhesion Scientist Career Stage and Research Topic in China

Abstract

Despite persistent efforts in untangling the mechanism of scientists switching between research topics, little is investigated the relationship of the career stage switch and the dynamic of research topics. Here, a two-layer network model, the coauthor collaboration network(α-layer) and the paper similarity network(β-layer) with the coupled information between them, is presented, and the relationship between the scientist career stage switch and research topics is analyzed. Utilizing the data of published papers and authors with at least one at Society of Management Science and Engineering of China, it is found that the career stages in α-layer display vivid stage switching behaviors respect to the publications, and β-layer has a clear community structure where each community represents a research topic. Computing the frequency of the state switch in α-layer, it is also found that scholars with a few topics in their early career have great probability to give up the research work, yet scholars with broad scopes would stay in research fields. The analysis result reveals that the coupling mechanism of the two layers is simulated by a linear correlation of the number of evolving topics with the changing career stages. The career stage switch is closely related with his/her collaborators in China, which the novice collaborating with the reputation scholar would help them find popular topics and projects, while the experienced scholars working with the novices would help to explore new knowledge frontiers. INDEX TERMS Two-layer network model, Markov chain, career stage, research topic, state transition.

TOSNet: A Topic-Based Optimal Subnetwork Identification in Academic Networks

“…The latter is "the difference between the number of nodes subsumed by a concept of a given community and neighboring communities." In another work, Ma et al [19] presented a network model that comprises two-layer such as collaboration network and paper similarity network layers. The latter is used to detect communities based on the similarity between each scholar's research topics in the collaboration network.…”

“…The latter is used to detect communities based on the similarity between each scholar's research topics in the collaboration network. According to the discussion in [19], the number of communities is equivalent to the number of research topics available in the network.…”