| 摘要 |
PURPOSE:To ensure a natural man-machine interaction without embarassing users by changing the expression of an answer sentence or outputting again an answer sentence containing other information when the time during which a user starts an input after production of an answer exceeds the expected threshold value. CONSTITUTION:An input interpreting part 1 analyzes a sentence supplied from a user into a command train to be given to a task processing part 3. An answer producing part 2 produces a data train through the part 3 as an answer sentence. A control part 4 sends a measurement start command and the threshold value to the part 4 itself to start measurement of the time elapsed. When this measuring time exceeds the set threshold value, a next candidate output command 13 to the part 2 to first produced an answer sentence i1 and then an answer sentence i2 in case no input is received from the user before the threshold value i1 and hereafter produced answer sentence ij every time the threshold value ij is exceeded. In such a way, it is possible to give in reply the system- based expression or the additional information in case the user is embarrassed since he/she is unable to understand a system answer sentence. |
