Abstract
The article presents a method for forming project groups for joint scientific activities based on the analysis of elements of scientific publications using machine learning methods, scientometric analysis and semantic technologies. The proposed information system integrates data from external sources, such as Scopus and Web of Science, and uses clustering and ranking algorithms for scientific entities. One of the key modules of the system is the creation and use of a thesaurus based on the data of publications, which makes it possible to more objectively assess scientific activities, identify areas of contribution of scientists and effectively form interuniversity scientific communities. Testing of the method led to a significant reduction in the time of searching for partners and an increase in the productivity of scientific projects. The study is of interest to scientific communities focused on cooperation and partnership in scientific research.