| 摘要 |
PURPOSE:To reduce the number of times of access to a dictionary and to accelerate the conversion speed of KANA (Japanese syllabary)/KANJI (Chinese character) conversion by building up instance information in a word dictionary. CONSTITUTION:In a step s4, the word dictionary is referred to and an input read string is divided into clauses from a sentence head side by using the logic of a two-clause maximum likelihood evaluation method for instance. Whether or not an instance is valid between the two adjacent clauses is obtained from the instance information in the step s5, normal KANA/KANJI conversion is performed in the step s6 when the instance is not valid and the KANA/KANJI conversion based on the instance information is performed in the step s7 when the instance is valid. When the word dictionary is referred to in the step s4, the instance information is simultaneously obtained. Thus, the need of accessing the dictionary again at the time of referring to the instance information in the step s5 is eliminated and the conversion speed can be substantially accelerated. |
