| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To acquire a character string based on a context of an input sentence regarding a symbol in the input sentence, and to reduce the cost of creating a dictionary for converting the symbol to a character string. <P>SOLUTION: In a symbol converter 1, a role specification means 13 specifies a role of a symbol on the basis of the sentence feature quantity of the symbol, which is a feature vector composed of a part of speech of a word in an input sentence 19 and an appearing position of the symbol in the input sentence as components, and a classification model 18, which is a classifier for specifying the role of the symbol from an arbitrary feature quantity. A conversion character string selection means 14 derives a word meaning set corresponding to the role of the symbol specified by the role specification means 13 from a word meaning dictionary storing the word meaning sets corresponding to the roles of the symbols, and determines the word meaning having a largest similarity between the sentence feature of the input sentence in which the symbol is substituted with a word meaning and the word meaning in the derived word meaning set to be a conversion character string 20 of the symbol. <P>COPYRIGHT: (C)2011,JPO&INPIT</p> |
