摘要 |
PURPOSE:To decrease the number of templates and to raise the processing efficiency of a natural language processing by classifying the variable of a template into the necessary one and the omittable one, and calculating a matching degree with use of the classified result. CONSTITUTION:When a template is retrieved by a template retrieving part 2, a structure coincident deciding part 1 executes the matching of inputted data and retrieved templates as the fact that all variables can be omitted. The information that the result is a success or a failure, the template and the matching pattern to the variable is outputted. An omitting deciding part 4 decides, from the classification of the output and the variable from the structure coincident deciding part 1, that the result of the structure coincident deciding part 1 is successful and that the success is obtained when the opponent of the necessary variable exists. The template and the matching pattern to the variable are conveyed to a matching degree comparing part 6.
|