摘要 |
PROBLEM TO BE SOLVED: To execute illustration and retrieval at high speed by retrieving a discriptor index in a database by means of the discriptor of a target document which is prepared based on the pattern of an in/inter-word label and adopting it as the coincident document when it is the specified one. SOLUTION: A fetching processor 12 receives plural input documents 14, preserves them in a document database 16, generates respective input document discriptors in the period and preserves them in a discriptor database 18. A collating processor 20 receives a sample page 22, uses an attached discriptor generator 40 and generates a sample page discriptor. Then, the discriptor is collated with the discriptor database 18, a document identifier for identifying the document with the discriptor being common to the sample page 22 is received, the document database 16 is retrieved based on the document identifier and the document having more discriptors common to the target document 24 than the other documents is presented as the coincident document.
|