DEVICE AND METHOD FOR DOCUMENT ANALYSIS

摘要

PROBLEM TO BE SOLVED: To securely analyze a document including a religion-relative terms. SOLUTION: A KANJI (Chinese character) and KANA (Japanese syllabary) mixed document (input document) which is inputted is analyzed by a normal term analysis part 131 by using a word dictionary 16 and divided into words. A word or phrase which is not found in the word dictionary 16 is regarded as an unkown word and a religion mode determination part 132 compares and matches the word in the input document right after the unknown word against a trigger dictionary 19 to determine whether or not the word is coupled with a religion-relative term behind it and, therefore, whether or not a religion mode wherein the unknown word is analyzed as a religion-relative term is entered. When the religion mode is determined, the said unknown word is analyzed by a religion-relative term analysis part 133 by being compared and matched against a religion term dictionary 18; when it is decided that the religion mode is not entered, the said unknown word is analyzed by a normal unknown word analysis part 134 by using a single-KANJI dictionary 17.