摘要 |
PROBLEM TO BE SOLVED: To provide a text classification program, method and apparatus that can automatically classify text information with high accuracy. SOLUTION: Input text data are morphologically analyzed into morphemes. Modification relations between the morphemes in the divided text data are analyzed, and a set of morphemes in a given modification relation is extracted. A distance is calculated between the morphemes forming the extracted morpheme set in the text data. According to a standard value calculated in advance and the distance calculated in the text data, similarity is calculated between reference text data and the text data. According to the calculated similarity between the reference text data and the text data, the input text data are classified. COPYRIGHT: (C)2005,JPO&NCIPI
|