HIERARCHICAL PRESEARCH-TYPE DOCUMENT RETRIEVAL METHOD, APPARATUS THEREFOR, AND MAGNETIC DISC DEVICE FOR THIS APPARATUS.

摘要

A document information retrieval method of effecting full text search, an apparatus therefor, and a magnetic disc device used therefor, wherein two-step presearch of documents is effected with respect to a keyword for the retrieval. In the first step (step 402) of the presearch, a character table (500) describing, by documents, the presence or absence of all the character codes included in a group of text data of the documents stored is generated in advance, the character table is searched using all character codes that constitute the keyword, and only the documents including the character codes are picked up. In the second step (step 403), compressed text data excluding annexed words contained in the text data and repetitively appearing words are generated, and documents containing the keyword as a word are picked up out of the documents picked up in the first step. After the second step (step 403), a text search (step 404) is effected according to proximity condition, context condition, etc. A dedicated hardware (1106) for character string collation based on the finite automation system is employed as character string collation means. As for different expressions and synonyms, an inputted character string is developed for a different expression through a different expression developing unit (2601), and reference is made to a synonym dictionary (2612) for each of the character strings developed for different expression in order to develop the synonyms through a synonym developing unit (2602). Then, the result of synonym development is developed through the different expression developing unit (2603) according to a conversion rule table (2603). The text data for document retrieval are stored by a plurality of magnetic disc devices (1) operable in parallel. These devices are simultaneously driven, and the output data thereof are systematically processed.