• Title of article

    USING THE LINK GRAMMAR PARSER IN THE STUDY OF TURKIC LANGUAGES

  • Author/Authors

    Batura, T.V. Institute of Informatics Systems - Russian Academy of Sciences Siberian Branch - Novosibirsk State University, Russia , Murzin, F.A. Institute of Informatics Systems - Russian Academy of Sciences Siberian Branch - Novosibirsk State University, Russia , Semich, D.F. Institute of Informatics Systems - Russian Academy of Sciences Siberian Branch - Novosibirsk State University, Russia , Sagnayeva, S.K. Gumilyov Eurasian National University, Astana, Kazakhstan , Tazhibayeva, S.Zh. Gumilyov Eurasian National University, Astana, Kazakhstan , Bakiyev, M.N. Gumilyov Eurasian National University, Astana, Kazakhstan , Yerimbetova, A.S. Gumilyov Eurasian National University, Astana, Kazakhstan , Bakiyeva, A.M. Novosibirsk State University, Novosibirsk, Russia

  • Pages
    9
  • From page
    14
  • To page
    22
  • Abstract
    Growing amount of information on the Internet and rapid development of social networks make the task of text processing increasingly actual. In this paper we propose an algorithm for the comparison of sentences and introduce certain measures of the closeness (similarity) between the sentences. The estimation of the relevance of documents should be based on the context of a search query and should not be limited only by keywords, their similarity or frequency. So proposed measures take into account lexical, syntactic and semantic relations between words. One of the problems we solve in the current time is the development of a parser like Link Grammar Parser for Turkic languages most frequent in the Internet, such as Kazakh, Uzbek (Cyrillic and Roman alphabets), and Turkish. The results of our research are planned to be used in different information retrieval systems.
  • Keywords
    natural language processing , syntactic analysis , Link Grammar Parser , rele- vance , Turkic languages
  • Journal title
    Eurasian Journal of Mathematical and Computer Applications
  • Serial Year
    2016
  • Record number

    2601162