Topic Analysis in Collection of Keyword Tuples

S. A. Afonin, Associate Professor, Researcher, e-mail: serg@msu.ru, Moscow State University, Institute of Mechanics, K. V. Lunev, Postgraduate Student, e-mail: kirilllunev@gmail.com, Institute of Mechanics Lomonosov Moscow State University

The paper presents algorithms for topic identification in collection of keywords tuples. The analysis of the keywords, namely the degree of their abstractness, is also carried out. The methods used in the paper are based on graph algorithms. The document is presented by the tuple of its keywords. The graph of keywords is built according to the given set of documents. The nodes of the graph fit the keywords, and the edges show the fact of occurrence of words in one tuple. With the help of the graph thematic relationship between documents and thematic characteristics of a document are identified.

Keywords: topic modelling, natural language processing, graph algorithms

pp. 29–39