DATA ANALYSIS DEVICE, ITS METHOD AND DATA ANALYSIS PROGRAM

摘要

PROBLEM TO BE SOLVED: To automatically extract the structure of a story from a section in which a voice recognition result can not be obtained from contents.

SOLUTION: A character string specified by contents data 10 is extracted by pattern recognition processing, the extracted character string is divided in each unit integrated as meaning and a topic boundary point data 11 specifying division points (topic boundary points 11a-11e) are extracted. Sound signal change point data 12 and video signal change point data 13 for specifying physical change points (sound signal change points 12a-12i and video signal change points 13a-13i) of the contents are extracted. Then story candidate sections 14a-14d are specified by using these extracted data and the topic boundary points 11a-11e, the sound signal change points 12a-12i or the video signal change points 13a-13i satisfying a prescribed condition are specified as story boundary points 15a-15d in each of the specified story candidate sections 14a-14d.

COPYRIGHT: (C)2004,JPO&NCIPI