Search

KEYWORD DISTINGUISHING METHODS IN THE PROCESS OF SCIENTIFIC TEXTS RENDERING // Tomsk State Pedagogical University Bulletin. 2018. Issue 8 (197). P. 45-50

The article deals with the problem of adequate keyword distinguishing from scientific texts in order to retain the meaning of the text and to lessen the risk of misinterpretation of information rendered by it. There is a great need for the effective methods of automatic keyword extraction due to the excessiveness of the data. The article presents the review of up-to-date approaches to text rendering and keyword distinguishing in foreign and Russian research. This research is essential for further investigation of the problem as it shows the advantages and disadvantages of the approaches found. The existing methods are subdivided into four groups: statistical approaches, natural language processing, machine training methods, hybrid methods combining the above mentioned three groups. The second part of the article presents a hybrid method for extracting keywords from scientific texts in the process of rendering law texts. It employs statistical and empiric approaches at its beginning taking into consideration the formal features of keywords (frequency, information weight, position). The next stages include linguistic methods of lexico-semantic analysis, building of word chains, the part-of-speech analysis. The final stage of extracting the categorical component of the meaning of the keywords is essential for describing the concept of the text and allows to restore the meaning of the whole text out of the few keywords. The research presents a list of categorical components of keywords extracted from abstracts of scientific texts of law discourse, which is the result of the application of the suggested method.

Keywords: rendering, scientific discourse, categorical component of the word meaning, text comprehension