А. А. Печников The author proposes an approach to study special interest communities based on the graph of researchers carrying up studies within the same discipline. The co-authorship graphs are the constructs of scientific collaborations popular owing the initial apparent “acquaintance” of the researchers documented in their articles. Many real networks are characterized by the clusterization, which means that the graph topology, as a real network model, is organized in communities, i. e. subgraphs with more internal connections than external ones. The analysis of communities resulted from co-authorship graph breakdown enables to identify the basic characteristics of the communities, e. g. their type (research thematic lines), number of researchers in the community and their interconnections. In the case of several communities of the same special interest, their consolidation determines the main lines of studies within the scientific discipline and generalized data like the total number of communities and researchers in each consolidated community. In their turn, these data can be used for administrative decisions on stimulating relevant and actionable studies. The author discusses the results of testing of the proposed approach on the basis of Math-Net.Ru portal data. Practically, the testing results prove the need to stimulate the studies in robotics and robotic systems, combustion and explosion, information protection methods and systems. The testing results evidence on the adequacy of the used mathematic models and potentiality of the approach direct transfer to other disciplines. The key is to have the complete and reliable basic bibliographic information on co-authorship within the scientific discipline under the study for the large enough time period.
Научная статья 
сообщество