摘要 |
PURPOSE: To obtain sufficient nonlanguage knowledge by extracting important words, which are not limited to technical terms, from an original text automatically and widely. CONSTITUTION: When the input document 1 to be translated is inputted through an input means 3, a machine translating means 4 analyzes the syntax an segments it into words. A term extraction rule is applied so as to extract a word string formed by connecting, for example, several nouns. In this case, a term extracting means 7 retrieves all word strings of successive nouns from the input document. An important word string which is repeatedly used can be selected by collecting the statistics of how many times a work string like them is used in the input document 1. |