| 摘要 |
PURPOSE:To improve the testing capacity for a candidate character by segmenting a temporary sentence clause containing the candidate character, and deciding whether the candidate character is proper or not, when the character kind of the first candidate character is different from the character kind of characters before and after said character. CONSTITUTION:A KANJI (Chinese character)-string testing part 9 segments two characters each from the left of the KANJI-string, gives a priority in order of a KANJI two-character word, two-character prefix and suffix, one-character prefix and suffix, and a KANJI one-character word, and searches each dictionary 15-18. If a matching word exists, the connection of parts of speech to a word immediately before said word is checked by using a connection weight matrix table 19. Also, as for other grammatical check, a KANJI-string having the unsuitable array of parts of speech such as a surfix + a prefix, etc., and a KANJI-string having the word array of a low frequency are decided to be unsuitable. These processings are repeated until other KANJI candidate character comes not to exist, and the character-string is outputted to the maximum likelihood candidate character selecting part 12. The maximum likelihood candidate character selecting part 12 selects the maximum likelihood candidate character by utilizing such information as a candidate order, the degree of similarity, a connection weight, an appearance frequency, etc.
|
