| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a KANA(Japanese syllabary)/KANJI(Chinese character) converting device for applying a highly likelihood single reading to a character string. <P>SOLUTION: A dictionary with notations and reading associated therein is retrieved, and reading corresponding to the notations of a selected whole character string is extracted (S402). The reading of the characters or words included in the selected character string is extracted (S406). The reading extracted by the step S406 is combined, and then the path of reading matched with the reading of the whole character string inversely converted by the step S402 is selected so that the lattice of possible reading can be constructed (S408). Scores are applied to each of paths included in the lattice (S410). In this case, the scores are based on the total of nodes dividing the reading included in the path. The path with the highest scores is selected (S412), and single reading including the reading included in the selected path is generated (S414). <P>COPYRIGHT: (C)2007,JPO&INPIT |
