| 摘要 |
PURPOSE: To realize an equivalently quick full text search by realizing a connected character component table free from noise due to hashing by a practical capacity and narrowing down the range of retrieval object documents with a high precision even at the time of designating a word consisting of the combination of alphabets and words as a retrieval term. CONSTITUTION: Text data 103 is divided into words, and n-character strings are extracted from an added word at every m characters, and a connected character component table 105 is generated where information indicating the existence of each character string is recorded in the entry of a character component table corresponding to the character string, and n-character strings are extracted from the retrieval term at every m characters, and the connected character component table 105 is searched with a retrieval control program 209, and thereby, documents which are not related to the retrieval term are excluded by connected character component table search before retrieval of a condensed text 104, thus realizing quick full text search. |
