| 摘要 |
PURPOSE:To accurately KANA (Japanese syllabary)/KANJI (Chinese character)-convert characters consisting of more than three words by preparing cooccurrence information showing the connection of the words apart from regular cooccurrence information showing the connection of more than two words and using cooccurrence information in accordance with the word constitution of a conversion word candidate. CONSTITUTION:A KANA-KANJI conversion dictionary 14 storing a conversion word corresponding to reading, a two words cooccurrence dictionary 17a storing first cooccurrence information showing the connection of the two words and a three words cooccurrence dictionary 17b storing second cooccurrence information showing the connection of three words which are more than the two words are provided. The word constitution of an input character string which is generated by referring to the KANA-KANJI conversion dictionary 14 is checked. When it is two words, the priority of the conversion word candidates is decided based on first cooccurrence information in the two words cooccurrence dictionary 17a. When it is not two words, the priority of the conversion word candidates is decided based on first cooccurrence information and second cooccurrence information of the three words cooccurrence dictionary 17b, and the conversion word candidates are outputted to a display part 18 in accordance with the priority. |
