DEVICE AND METHOD FOR PREPARING INDEX, DEVICE, METHOD AND SYSTEM FOR RETRIEVING DOCUMENT, DEVICE AND METHOD FOR PREPARING DATABASE, AND STORAGE MEDIUM

摘要

PROBLEM TO BE SOLVED: To output a document part fitting a retrieval condition as a retrieval result by dividing a structured document into segments from a structure and a content and presenting a segment including a given retrieval key as a retrieval result. SOLUTION: A document is divided into segments by a specific tag (S301), as for a segment including an image, the degree of association with an adjacent segment is calculated and the segment including an image is united with an adjacent segment having a prescribed degree of association (S302). A header is detected by a header tag and header information is added to a segment included in the range of the header (S303). After the segment is divided into documents, the documents are combined in accordance with the degree of association between the documents, and indexes are prepared for the header of the segment and respective parts other than it (S305). Retrieval is performed to two indexes, the goodness of it is calculated by weighting a retrieved result to the indexes, and a retrieved result decided by the goodness of fit is outputted in a segment unit.