Probabilistic topic modeling of the russian text corpus on musicology

Abstract

The paper describes the results of experiments on the development of a statistical model of the Russian text corpus on musicology. We construct a topic model based on Latent Dirichlet Allocation and process corpus data with the help of the GenSim statistical toolkit. Results achieved in course of experiments allow us to distinguish general and special topics which describe conceptual structure of the corpus in question and to analyze paradigmatic and syntagmatic relations between lemmata within topics.