Constructing Scientific Publication Profiles Based on Texts and Coauthorship Connections (in the Field of Control Theory and Its Applications)

Abstract. The calculation of scientific publication profiles is crucial in the systematization of scientific knowledge and support for scientific decision-making. This paper proposes a method for forming publication profiles in the field of control theory, based on the integration of text analysis and coauthorship network analysis. We describe a basic algorithm that analyzes publication texts by a thematic classifier as well as its enhanced version that considers network connections within a heuristic approach. The methods are examined using expert assessments and quantitative metrics; according to the examination results, combining textual and network data significantly improves the accuracy of publication profiles. Hypotheses about a relationship between the thematic similarity and network proximity of publications are tested, and the approach proposed is validated accordingly. In addition, directions for further research are identified.

Keywords: publication network, publication profile, control theory, graph neural networks, text analysis.

PDF (English)

Cite this paper

Gubanov, D.A. and Melnichuk, V.S., Constructing Scientific Publication Profiles Based on Texts and Coauthorship Connections (in the Field of Control Theory and Its Applications). Control Sciences 1, 39–44 (2025).