| 摘要 |
PROBLEM TO BE SOLVED: To provide a retrieval device for preventing the retrieval leak or erroneous retrieval even when a block where the reliability of voice recognition is low is mixed with content data constituted of video data and voice data. SOLUTION: As for each block of content data, meta information including element data describing the identifier of the block, voice text being the voice recognition result of the voice of the block, either the high reliability of the voice text or the low reliability of the voice text and keywords extracted from the voice text and a category to which the keywords belong is stored, and when the first character string and the second character string are retrieval conditions, the meta information including the element data describing the category having the same sounds as those of the first character string and keywords having the same sounds as those of the second character string is retrieved from the highly reliable meta information, and when only the second character string is the retrieval conditions, the meta information including one of a plurality of third character strings having sounds which are the same as or similar to those of the second character string in the voice text is retrieved. COPYRIGHT: (C)2005,JPO&NCIPI
|
