ИСТИНА |
Войти в систему Регистрация |
|
ИПМех РАН |
||
Techniques for determining the time-point of an event in the fiction text is cruial in the field of advertizing as well in the second language acquisition.. Formal detection of events in the narrative based on text itself, without considering its perception by a reader is proposed. Our hypothesis is as follows: the description of the state is defined by the number of pairs of antonyms one of which is a word of the sentence under consideration and the other one occurs in the rest of the story. This article involves examination of this hypothesis and the results obtained. The essence of the method of detection of events is as follows: For each adjective or adverb in a single sentence all its antonyms within the rest of the text of the story are found and counted. Then we divide the number of the found antonyms by the number of words in the sentence to avoid the dependence of the results from the length of the sentence. The results are presented in graphical form on the X-Y plane. Each point of the graph has coordinates (x, y) where x = Ns is the serial number of the sentence starting from the beginning of the story; y is the count of the antonyms found in the whole story for all the words of the sentence Ns. The resulting graph of antonyms undergoes smoothing and interpretation.