APPLIED ASPECTS OF COMPUTER SCIENCE

Short text messages classification method

E. Mbaykodzhi^a, A. A. Dral^b, I. V. Sochenkov<sup>cd

a</sup> Peoples' Friendship University of Russia, Moscow
^b Lomonosov Moscow State University
^c Institute for Systems Analysis of Russian Academy of Sciences
^d LLC "TSA"

Abstract: The article covers short text classification method based on different measures of similarity word distribution. The work presents the text classification method based on a new lexical descriptor information gain concept and a topic importance characteristic. The results of experiments are presented.

Keywords: short text classification, metadata based classification, SVM, TF, IDF, topic importance characteristic.