| 摘要 |
PURPOSE:To improve processing efficiency, by progressing a dividing process handling a non-registered word as if it is a registered word even when the non-registered word is present, and constituting a system so that the part of speech of the non-registered word is decided, and is registered on a dictionary during the above operation. CONSTITUTION:A dictionary retrieval part 5 divides the character of an input text in a word unit (or in a clause unit), retrieving and checking the content of a dictionary 1. An adjacent relation checking part 6 performs a division correctly by re-selecting, for example, a unit so that a divided unit is arranged correctly from a grammatical standpoint. The longest dividing list 8 is set so as to save a divided result until then while a dividing process is performed, and when it is necessary to perform a back track (to retry to divide ex post facto) at the time of finding the non-registered word, and it is constituted so that the result of the dividing process until then is kept, and it is utilized fro the process of the non-registered word. At a non-registered word processing part 7, the dividing process is performed in such a way that the non- registered word is considered as a word registered already temporarily, however, the part of speech obtained at that time is attached on the word, then being registered in the dictionary 1.
|
