To generate a corresponding typical natural language representation from a structure of a frequent appearance pattern having similar text meaning in document processing.
A language analyzing means 21 reads a set of sentences from a text DB 11 of a storage device 100 and performs a morphological analysis and a modification analysis of each sentence, a frequent appearance pattern extracting means 22 extracts a partial structure that frequently appears from a tree structure, and a typical representation generating means 23 scores a morpheme string corresponding to the partial structure prepared by the frequent appearance pattern extracting means 22, generates natural language with the highest score or a plurality of natural language representations ranked in a score order and outputs the natural language representations to an output device 300.
COPYRIGHT: (C)2008,JPO&INPIT
JP2004102397A | ||||
JP2001084250A |
Akitaka Kimura
Toshio Asai