Text segmentation method and apparatus, text segmentation program, and storage medium storing text segmentation program

摘要

PROBLEM TO BE SOLVED: To make only the boundary of semantic paragraphs settable as a right answer from a text neither too much nor too little. SOLUTION: The text is divided into words by morpheme analysis, a vector corresponding to each of words provided by morpheme analyzing processing is acquired by retrieving a concept base storing vectors expressing the meanings of words, and word strings as set of words of a certain number are taken before and after the boundary of words. Then a word string coupling degree is calculated from information on the vectors of words comprising each of word strings as similarity scale or distance scale of preceding and following word strings and a minimum word boundary when the word strings coupling degree is similarity scale or maximum word boundary when is distance scale, is recognized as a boundary of semantic paragraphs of the text.