This paper proposes a new method of detection of research directions, as well as a way of automatic tuning of its parameters. The method defines a function of assessment of the similarity of scientific publications that takes into account thematic similarity of texts, cited references and co-authorship. In contrast to other methods for calculation of thematic similarity of texts thesauruses and ontologies are used. That allows taking into account the specifics of the subject areas and identifying multiword-combinations with syntactic and semantic dependencies that characterizes research directions in a best way.
PDF at the Institute for Systems Analysis of Russian Academy of Sciences website (in Russian): http://www.isa.ru/aidt/images/documents/2015-02/53_59.pdf
Artificial Intelligence and Decision Making magazine: http://www.aidt.ru/index.php?option=com_content&view=article&id=686&lang=en
D. A. Devyatkin, Y. E. Danik, I. A. Tikhomirov, A. V. Shvets. Method of selecting directions of scientific research on the basis of analysis of full texts of publications // Artificial intelligence and decision-making. 2015. p. 53-59