| 摘要 |
PURPOSE:To continue highly efficient processing with a mechanical translation system even in a group unit containing plural continuous unregistered words, by adding a processing unit recognizing part to divide an input original sentence for each processing unit based on the retrieving result obtained by a dictionary retrieving part. CONSTITUTION:An input original sentence received from a input part 1 is inputted to a translating part 5 via an editing control part 4 and then translated. The part 5 includes a processing unit recognizing part 52 to decide the processing unit from the retrieving result obtained by a dictionary retrieving part 51. This decided processing unit is outputted to a translation processing part 53. In other words, a character pointer set at an unregistered part is set at the 2nd character from the head and this character pointed by the pointer is compared with its immediately preceding character in terms of their types as long as the pointed character is kept within the unregistered part. Based on the result of this comparison, the original sentence is divided as long as the adjacent characters have different types. While a process where the pointer is advanced by a step is repeated if the adjacent characters have the same types. When the pointer gets out of the unregistered area, the remaining part is divided under the single processing condition. Thus the highly efficient translation processing is continued even in the case of plural continuous unregistered words.
|
