DEVICE AND METHOD FOR FULL-TEXT REGISTERED WORD RETRIEVAL

摘要

PURPOSE: To reduce the file capacity of a main text to be retrieved and retrieve full-text registered words by connecting retrieval indexes and the compressed main text file of a compressed block. CONSTITUTION: A reference unit division part 11 divides the main text into reference units sectioned with tags. Then a word position extraction part 12 extracts a pair of a word and the position of the reference unit in the main text where the word appears. A main text compressing process part 15, on the other hand, compresses the main text in the respective divided reference units into blocks. Then a classifying process mechanism 13 generates a word position set of the positions of all compressed blocks where the word appears for the word. A retrieval index generation part 14 generates retrieval indexes 4 for obtaining the position set of the compressed blocks from the word for the word position set. Further, a retrieval index compressed main text connection part 16 connects the corresponding retrieval indexes 4 to the compressed main text file 5 of the compressed blocks to generate a compressed main text file 5 to which the retrieval indexes 4 are added.