This work considers the networks of scientists, which take into account not only the relationship of co-authorship, but also the thematic proximity of their scientific interests. The unique feature of the presented approach is its use of a typical scientometric service and consideration of tags or descriptors of topics attributed by scientists to themselves and other authors of articles indexed by this scientometric service. During the implementation of this approach, a special algorithm is used to scan the resources of the scientometric service and obtain a representative set of authors or co- authors as network nodes. The weight of the connection between scientists in the considered network is determined by the meaningful correlation of their scientific fields, which is measured by the number of matching descriptors. Clustering algorithms enable the identification of groups of highly connected nodes that correspond to scientific schools and teams of scientists capable of collaborating on joint projects. The software implementation of the proposed approaches and methods uses the Perl and Python programming languages, publicly available information scanning utilities, and Gephi graph analysis and visualization software.

Keywords

Network of scientists, co-authorship network, scientometric service, information network scanning, topic descriptors, cluster analysis, cybersecurity.