JPH07325597A - Information input method and device for executing its method - Google Patents

Abstract

PURPOSE:To surely realize the high input efficiency by deciding whether confirming directly when respective inputs are performed or confirming after two pieces or above of the inputs are united according to the certainty of the inputted information when the information is inputted to a machine and confirming in the decided time. CONSTITUTION:An information inputter inputs the information to an information input means 10. The information related to the input such as the certainty, etc., of the inputted information and the input is transmitted to a conversation condition control means 20. The conversation condition control means 20 presents the information necessary for stimulation of next input and confirmation of the inputted information, etc., to the information inputter through an information output device 90 by commanding an output information generation means 80. Whenever the information is inputted from the information inputter to an above-mentioned information input device, a confirmation time decision means 30 operates and decides the time to be confirmed, and the conversation condition control means 20 revises an information input and its confirmation flow according to the confirmation time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for inputting information while confirming it in order to correctly input information to a machine, and in particular, information that shortens the time required for confirmation work and does not make the information inputter feel annoying. The present invention relates to an input method and an apparatus for implementing the method.

[0002]

2. Description of the Related Art When inputting information to a machine, it may be erroneously input. Therefore, it is indispensable to have a mechanism that allows the input person to confirm the input information and detect an error. However, in certain types of input work such as input work using voice recognition, this confirmation work prolongs the input time and causes the information input person to feel annoyed. Therefore, there has been a need for a confirmation work method in a short time and without making the input person feel annoyed. In the case of inputting information by voice recognition, a method of preparing a plurality of predetermined input work flows according to the proficiency of the input person and changing the confirmation timing according to the flow is also performed. Kaihei 3-160868
Issue). However, since the input work flow decided for each input person is applied, it is not possible to adapt to the change of the input person's state such as the voice being faint due to a cold or the like and the false recognition being increased in the voice input. In addition, it cannot adapt to changes in input items such as input items that are easy to input and input items that are prone to input errors.

As another example of information input by voice recognition, the recognition accuracy is estimated by using a recognition parameter such as a matching distance at the time of voice recognition, and when the accuracy is sufficiently high, the confirmation work is omitted and the accuracy is improved. There is a confirmation method in which confirmation work is performed only when the temperature is low (Japanese Patent Laid-Open No. 63-123098).
JP-A-63-169699, JP-A-3-
248199). However, even when the accuracy is high enough, it cannot be guaranteed that there is no error. At this time, if the confirmation is omitted, the error detection fails.
If the criterion of the recognition accuracy is tightened due to fear of failure in detecting this error, the number of inputs satisfying the criterion is reduced and the number of times the confirmation is omitted is reduced, which has a drawback that the feature of this method cannot be utilized.

[0004]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and in the information inputting work, it corresponds to the change of the state of the input person or the input item, and the confirmation is omitted. Therefore, it is an object to improve the efficiency of confirmation work by eliminating the risk that an error cannot be detected.

[0005]

According to the information input method of the present invention, evaluation information (for example,
Information that evaluates the necessity of confirming the input such as the distance from the standard pattern that is the basis of the recognition rate, the degree of similarity, or the importance of the input information). As a result, when it is the confirmation time, the confirmation process of whether or not the information is correctly input is executed. Further, the information input device of the present invention comprises a determination means (30) for determining the confirmation timing of the input information based on the above-mentioned evaluation information of the input information, each time the information is input.
Judgment criterion storage means (41) for storing the judgment timing judgment reference to be referred to in the judgment, and confirmation processing means (20) for executing confirmation processing as to whether or not information is correctly input at the confirmation time judged by the judgment means. ), A storage means (43) for storing the evaluation information and the confirmation processing result, and an updating means (50) for optimizing and updating the judgment criterion based on the contents of the storage means.

[0006]

In general, the following two kinds of time can be considered when performing the confirmation work. There are two methods, one is to perform the information immediately after the user inputs one item of information, and the other is to perform the information after two or more items are collected. The method of performing confirmation work immediately is effective when the input information is likely to be incorrect, and the method of performing confirmation work after collecting two or more items is effective when the input information is likely to be correct. (International Man-Machine Studies, Volume 28, (198
8), Einsworth, “Optimization of Input Length in Speech and Numeric Input with Error Correction” p. 573-581 [I
nt. J. Man-Machine Studies,
Vol. 28, (1988), W.W. A. Ainswor
th: “Optimization of Strin
g length for Spoken Digit
Input with Error Correct
")"). Especially for input such as voice recognition in which erroneous recognition is inevitable, the input time is divided into two methods, one that is performed immediately after inputting one item and one that is performed after two or more items of information are collected. The relationship with the voice recognition rate is as shown in Fig. 5. For input methods other than voice, when the output from the device uses a voice message, or when the recognition process or data communication takes time. Therefore, the same relationship as described above is established for an input requiring a long reaction cycle of the input device, and thus the information input method of the present invention determines whether the input information can be accepted or not accepted. Not only functional criteria, input accuracy such as whether the input information is likely or not correct,
Evaluated information that evaluates the necessity of input confirmation such as the importance of information such as whether it is an input that is greatly affected by incorrect input, and the confirmation work is performed each time information is input. By properly changing the timing of performing, it is possible to efficiently input correct information. It can be said that this is a technique for inputting information in a natural manner so that it is difficult to hear even in conversation between humans, and when it is important to listen back quickly for important contents. As described above, according to the present invention, when the information is input by the input person, the accuracy or importance of the information is estimated, the time for performing the confirmation work is determined according to the estimation result, and the determined confirmation is performed. I decided to do confirmation work when the time came for work,
It is possible to obtain an optimal time for confirmation according to the input information.

In the information input device of the present invention,
Every time the information is input, the determination means determines the timing of confirmation of the input information by referring to the determination timing determination criteria stored in the determination criteria storage means based on the above-described evaluation information of the input information. . Then, the confirmation processing means executes the confirmation processing as to whether or not the information is correctly input at the confirmation timing determined by the determination means. The updating unit stores the evaluation information and the confirmation processing result in the storage unit, and optimizes and updates the determination standard based on the contents of the storage unit.
According to the present invention, since the confirmation work of the input information is performed at the optimum time, the time required for the confirmation work and the troublesomeness of the input person can be minimized. Further, according to the present invention, the criterion is learned by updating the criterion according to the confirmation processing result such as whether the recognition result for the input is correct or incorrect, and thus the criterion is learned by the user. The input speed can be improved according to the optimization.

[0008]

FIG. 6 is a block diagram of the first embodiment, and FIG. 1 is an overall flow chart. In this example,
This is an example in which a recognition score output from a voice recognition device, which is one of the parameters indicating the accuracy of recognition, is used as a criterion for determining a confirmation timing for a task input by voice as a criterion for determining a confirmation timing.

First, the user speaks into the microphone 11 an item to be input. The voice input from the microphone 11 is recognized by the voice recognition device 12.
The output data of the voice recognition device 12 is the recognition result of the first candidate and its score, and the difference between the recognition scores of the first candidate and the second candidate. These output data of the voice recognition device 12 are passed to the dialogue state management means 20. The dialogue state management means 20 collects data for deciding whether or not to take confirmation immediately, and passes it to the confirmation time determination means 30. That is, referring to the dialogue state storage means 42 together with the output data of the voice recognition device 12 such as the recognition result of the first candidate, the score of the first candidate, the difference between the scores of the first and second candidates, and so on. The data necessary for the confirmation occasion determination such as the number of already input items that have been stored without confirmation and the dialogue state such as the break of the dialogue is passed to the confirmation occasion determination means 30.

The confirmation timing determining means 30 determines the timing for confirmation. FIG. 7 shows a flowchart of this determination procedure. The confirmation time determination means 30 receives the data from the dialogue state management means 20 (step S201), and determines whether or not the dialogue states are separated by the received input item (step S211). It is considered that, for example, in a task of registering a schedule, four items of month, date, time, and requirement are separated from each other when the state of dialogue is separated from each other. It can be said that there is a break even when all the input is completed. If there is a break, it can be concluded that it is necessary to immediately confirm the input up to that point, so the confirmation timing determination process is performed (step S282).
If there is no break, it is checked whether or not there are less than three items that have not been confirmed among the items that have been input so far (step S221). When there are three or more items, it is concluded that it is necessary to immediately check all the items that have not been confirmed, because there are too many accumulated items, and the confirmation timing determination process is performed (step S282). If three words are not accumulated, it is determined whether the next input item is important, so the recognition result is important by referring to the list of whether the recognition result is an important word in the state at that time. It is determined whether the word is a word (step S2
51). Next, the determination criterion is referred to in order to determine the confirmation timing (step S252). In this embodiment, the criterion is referred to based on the importance of the recognized word, the recognition score of the first candidate, and the difference between the scores of the first candidate and the second candidate. The time of confirmation is set as the determination result depending on which region of the determination standard is divided (step S281). The determination result is sent to the dialogue state management means 20 (step S290).

FIG. 9 shows a conceptual diagram of a specific example of the confirmation timing judgment criteria. In the example of FIG. 9, the confirmation timing is determined by the recognition score of the first candidate and the difference between the recognition scores of the first candidate and the second candidate. The horizontal axis is the recognition score of the first candidate, and the score goes to the right. The vertical axis represents the difference between the recognition scores of the first candidate and the second candidate, and the higher the score, the larger the difference in scores. Using the determination criteria of FIG. 9, it is determined according to the recognition score that confirmation is taken immediately when the area is classified into the area 2 and collective confirmation is performed later when classified into the area 3. The area 5 is a portion that changes depending on the importance of the word, and it is determined that the immediate confirmation is made when the important word is made, and the collective confirmation is made later when the important word is not made. Whether or not it is an important word is determined by referring to the important word list stored in the confirmation timing determination criterion storage means 41. An example of the important word list is shown in FIG. As described above, the timing of the determined confirmation is sent to the dialogue state management means 20, and the procedure for determining the confirmation timing is completed. In addition, the place colored in gray on the upper left of FIG. 9 indicates a portion where the recognition score of the second candidate has a negative value. In this example, since the recognition score is considered to be a positive value, there is no data in this gray area.

Depending on the confirmation timing determined by the confirmation timing determination means 30, the dialogue state management means 20 may or may not immediately confirm. When confirming immediately, the dialogue state storage means 42 is referred to, and if there is an input item that has not been confirmed so far, the input item is presented to the input person together with the input item, Perform confirmation work. If an error is detected as a result of the confirmation, the error will be corrected, but the method is omitted here. If it is confirmed as correct as a result of the confirmation, the dialogue state storage means 42 records that the confirmed item has been confirmed together with the result of this input, and proceeds to the next input. If the confirmation is not performed immediately, the fact that the confirmation has not been completed is recorded together with the result input this time, and the process proceeds to the next input.

The message presented from the device to the user such as confirmation of the input items and prompting for the next input is output information generation means 8 according to the instruction of the dialogue state management means 20.
0 reads out the message stored in the output information storage means 43, synthesizes a voice signal by the voice synthesizer 81, and presents it to the input person via the amplifier 91 and the speaker 92. FIG. 10 shows a state of the dialogue between the input person and the apparatus in the information input which progresses in this way.

As described above, in this embodiment, when the information is input by the input person, the accuracy and importance of the information are estimated, and the timing of the confirmation work is determined according to the estimation result. Since the confirmation work is performed when the time for the determined confirmation work comes, it is possible to obtain the optimum confirmation time according to the input information.

(Second Embodiment) The second embodiment is the same as the first embodiment except that the area 1 having a very low recognition accuracy and the area 4 having a very high recognition accuracy are used as a criterion for confirmation timing.
Is added, and in area 1, a function for prompting re-input immediately without adopting the recognition result and for omitting confirmation work in area 4 is added. That is, as shown in FIG. 11, when confirmation time judgment criteria are classified into areas 1 to 4 and an input classified into areas 2 and 3 is input, the same as in the first embodiment. To work. When an input classified into the area 1 is input, the recognition result is not adopted and the input user is prompted again to make the same input. When an input classified into the area 4 is input, it is considered that there is no erroneous recognition and confirmation is not performed. This can be realized by entering the contents of the items entered in the dialogue state in the dialogue state storage means 42 and the fact that there is no need for confirmation. According to the present embodiment, when the recognition accuracy is extremely high, the confirmation work is omitted, and even when the recognition accuracy is extremely low, the confirmation work is not performed and the re-input is performed immediately, so that the input speed is increased. There is.

(Third Embodiment) In the case where an interface which combines a touch panel to be touched with a finger and input guidance by voice output is used so as to be easily understandable, it is compared with the case of only visual output. Takes some time. The third embodiment is an embodiment that enables efficient input in such a case. FIG. 12 is a block diagram of the third embodiment, and the entire flow chart is common to FIG. FIG. 14 shows a display example of the display 95 when a numeral is input. The input person touches the touch panel 13 in accordance with the display on the display 93 to perform input.
The touch panel 13 outputs the position information touched by the input person. The input processing means 14 determines the input of the input person from the output of the touch panel 13 and the display content of the display, and informs the dialogue state management means 20 of the input content and the distance to the edge of the area touched in addition to the determination result. FIG. 15 is an enlarged view of a part of the touch panel display of FIG. 14, where the position touched by the finger is x.
An area in which the button is touched is indicated by a broken line. When the dialogue state management means 20 receives the information from the input processing means 14, it refers to the timing means 21 constituting one of the functions of the dialogue state management means 20, and refers to the time when the input person touches the touch panel to input the information ( Time stamp). Next, with reference to the dialogue state storage means 42, it is determined whether or not the input this time is a good input for confirmation. Then, data necessary for determining the time of confirmation such as the button input by the input person, the time stamp thereof, the distance to the edge of the area, and whether or not the input is well separated are input to the confirmation time determination means 30. The confirmation time determination means 30 is notified.

Confirmation timing determining means 30 that has received the data
Determines the timing of confirmation based on the data. FIG. 13 shows a flowchart of the determination. First, it is checked whether or not the input has a good break (step S211). If the input is well-separated, it is determined that confirmation is to be performed immediately including items that have not been confirmed so far (step S283). If the input is not well separated, the time difference from the previous input is calculated from the time stamp (step S261). Further, the time difference is compared with the average and variance of the input intervals so far to determine whether or not the deviation is large (step S262), and is used for the determination of the confirmation timing to be described later. Next, the distance between the touched position and the end of the button area is calculated (step S263). Then, reference is made to the determination criteria of the confirmation occasion shown in FIG. 16 stored in the confirmation occasion determination reference storage means 41. From the time difference from the previous input and the distance between the touched position and the edge of the button area, it is possible to know whether to enter the area 2 or the area 3. Time difference is T
When it is in the range between min and Tmax and is smaller than the distance Dmin between the touched position and the end of the button area, the area 3 is set, and the other areas are set as the area 2. Tmin and Tmax in FIG. 16, which are the criteria for determining the region,
Dmin and the like are different depending on the input state, the position and size of the button, and the like. FIG. 16 shows that Tmin and Tmax are changed, and Dmin is changed depending on the position of the button, depending on whether the input state is numeric input or command input. As a result of referring to this determination criterion, when entering the area 2, the button was pressed very quickly and intended to be pressed once, resulting in a double touch, or was it too late and hesitated to press? Since the pressed position is near the edge, it is highly likely that the input is incorrect, such as touching the button even if you intend to press the next button, or pressing due to the parallax between the touch panel and the display, so it is determined to confirm immediately. (Step S284). When entering the area 3, the button is pressed in a stable manner and there is a high possibility that the input is correct, so it is determined later that confirmation will be taken collectively (step S).
285). The contents thus determined are transmitted to the dialogue state management means 20 (step S290).

The dialogue state management means 20 confirms the input or accepts the next input according to the confirmation time determined by the confirmation time determination means 30. When it is determined that the confirmation should be performed immediately, the confirmation content is sent to the output information generating means 80. Output information generating means 80
Reads the information to be output when confirming the items to be confirmed from the output information storage means 43, changes the display on the display 92 according to the type of information, and presents a voice message through the amplifier 91 speaker 92. hand,
Send to the input person. Then, the input person can confirm the entered information, and further confirm whether the information is correct or incorrect.
Enter from 3. At this time, it is possible to assign nothing when it is correct. If there is an error, correct it. The dialogue state management means 20 responds to an input from the input person as to whether the dialogue state storage means 4 is correct or incorrect.
For items that can be confirmed in 2, the confirmation is recorded, and if there is an error, the dialog state is updated by making a correction and the next input is accepted.

When it is later determined that the items should be collectively checked, the input item of this time is recorded in the dialogue state storage means 42 as an item which has been input but not yet confirmed,
The next input is accepted. When the state of accepting the next input is entered, the output information generating unit 80 is instructed to display a display prompting the input or to output a voice message in order to display the prompt for the next input as necessary. The output information generation means 80 is the output information storage means 43.
To change the display on the display 92 or output a voice message through the voice synthesizer 81, the amplifier 91 and the speaker 92. The information input work is performed by repeating the above flow until the input is completed.

According to the present embodiment, in inputting information on the touch panel, when a message of an appropriate input guidance by voice output is transmitted to the input person through each process of determination of confirmation timing and confirmation processing, It is possible to minimize the time consumption by outputting the message and input without error.

(Fourth Embodiment) In the fourth embodiment, the confirmation timing judgment criterion in the first embodiment is updated based on the input accuracy information and the confirmation processing result. FIG. 17 is a block diagram of the configuration of the fourth embodiment. In the configuration of the first embodiment of FIG.
It has a configuration in which 0 is added. FIG. 18 is a flow chart of the overall processing operation of the apparatus of the fourth embodiment. First, the input person speaks into the microphone 11 for an item to be input. The voice input from the microphone 11 is recognized by the voice recognition device 12. The output information of the voice recognition device 12 is the recognition result of the first candidate and its score, and
It is the difference in recognition score between the first and second candidates. The output information of the voice recognition device 12 is passed to the dialogue state management means 20. The dialogue state management means 20 refers to the dialogue state storage means 42, counts the number of items that have been input so far but is not confirmed, and transmits the output data of the voice recognition together with the number to the confirmation time determination means 30. To do.

The confirmation time determination means 30 determines the confirmation time based on the data passed from the dialogue state management means 20. FIG. 19 shows a flowchart of the determination procedure.
First, data is received from the dialogue state management means 20 (step S201), it is determined whether or not the input item is the last input (step S212), and if it is the last input, a confirmation is made to finish the input work. Since it is necessary to perform the above, it is decided to confirm immediately (step S284). If it is not the last input, the number of unconfirmed input items is checked (step S221). If the number is 3 or more, it is determined that the confirmation will be performed immediately (step S2).
84). When the number is two or less, the confirmation timing determination reference shown in FIG. 20 is referred to (step S252), and it is determined which area to divide (step S272). 20, the space defined by the recognition score of the first candidate of voice recognition and the difference between the recognition scores of the first candidate and the second candidate is divided into small regions. It is a thing. Each small area is divided into an area 2 and an area 3. From which recognition score of the first candidate and the difference between the scores of the first and second candidates, it is determined which region the input this time belongs to. Area 2
In the case of 1, there is a high possibility that the first candidate is correct, but since it is not so high, it is determined to perform confirmation immediately (step S284). In the case of region 3, the first candidate has a relatively high probability of being correct, so it is determined to collectively confirm later (step S285). The determined result is transmitted to the dialogue state management means 20 (step S290).

The dialogue state management means 20 carries out confirmation work or accepts the next input according to the confirmation time transmitted from the confirmation time determination means 30. FIG. 21 shows a flowchart of the confirmation work. First, the dialogue state management means 20 reads an item that has not been confirmed so far from the dialogue state storage means 42, and instructs the output information generation means 80 to output a confirmation message. The output information generation means 80 refers to the output information storage means 43 and outputs a message required for confirmation of information not confirmed through the amplifier 91 and the speaker 92 (step S501).

If there is no error in the input matter that the input person has heard the message necessary for confirmation, the input person inputs "Yes", which is a silent word having the affirmative meaning or the same meaning, into the microphone 11. The dialogue state management means 20 monitors the input from the input person (step S502), and when there is an input, checks whether or not the input is "yes" (step S5).
11, step S512), if there is no input, it is checked whether there is no input for a certain period of time (step S503). As a result, when the voice recognition device 12 outputs a recognition result of "yes" or when the speech recognition device 12 is silent for a certain period of time, the dialogue state management means 20 confirms the contents of the dialogue state storage means 42 for the items confirmed this time. Change to state (step S53)
2). Then, the next input is received (step S533).

On the other hand, when the input person finds an error in the input matter, he / she inputs "No" having a negative meaning to the microphone 11 or re-states the correct input content. When the voice recognition device 12 gives a recognition result of "No", the dialogue state management means 20 changes the contents of the dialogue state storage means 42 to the negated state of the item confirmed this time (steps S521, S532). . The denied item is re-input by the same procedure as the normal input (step S553). When rephrasing, the item recognized this time is changed to the denied state (step S5).
51, step S552), and then repeats from the confirmation time determination with the recognition result of the rewording (step S553). At the time of accepting the next input, the dialogue state management means 20 transmits a command for prompting the next input to the output information generation means 80. The output information generation means 80 refers to the output information storage means 43 and outputs a message prompting the next input to the amplifier 91 and the speaker 92. The input person makes the next input in the same manner as the previous input in response to the message. In this way, input is repeated until input of all items is completed.

In parallel with the above-mentioned information input procedure, the system of this embodiment changes the confirmation timing judgment criterion. There are two possible ways to make changes: immediately when a confirmation operation occurs, or when a certain amount of entered information has accumulated. Here's how to do it when you're away. The flowchart is shown in FIG. At the time when one set of input work is completed, the dialogue state management means 20 confirms the confirmation timing judgment criterion changing means 5
Command 0 to rewrite the confirmation criterion. The confirmation timing determination criterion changing unit 50 firstly detects, from the confirmation timing determination criterion storage unit 42, voice recognition results of scores classified into the small regions, which are stored for each small region as shown in FIG. , All the recognition results indicated by the number of inputs up to that point and the number of correct recognitions recognized as correct at the time of confirmation are read (step S801). Next, the difference between the recognition result of the confirmed input item, the recognition score of the first candidate, and the recognition score of the first candidate and the second candidate, which are input by the series of input operations at that time, from the dialogue state storage means 42. As a result of the confirmation, whether the result is affirmative or negative is read one by one (step S802). Based on the recognition score of the first candidate and the difference between the recognition scores of the first candidate and the second candidate for each item, it was determined which sub-region they corresponded to, and whether the recognition result for each item was affirmed in the subsequent confirmation work. The recognition result of the small area is changed depending on whether it is denied, that is, whether the recognition result is a correct answer or an incorrect answer. That is, when the answer is correct, the number of correct answers and the number of inputs are increased, and when the answer is incorrect, only the number of inputs is increased by one (step S803). When all input items have been changed, the recognition rate of each small area is calculated by dividing the number of correct answers by the number of inputs (step S805). As a result of the calculation, each small area is redivided into areas 1 to 4 according to a predetermined criterion as shown in FIG. 24 (step S806). By rewriting the result in the confirmation time criterion storage means 42 (step S
807), the work ends. For the following input,
Use new criteria.

According to the fourth embodiment, the same effect as that of the third embodiment can be obtained, and the judgment criterion is updated according to whether the recognition result for the input is correct or incorrect. Since the judgment criterion is learned by this, the judgment criterion is optimized according to the user, and the input speed can be improved.

[0028]

According to the present invention, when information is input by an input person, the accuracy and importance of the information is estimated, and the timing of the confirmation work is determined according to the estimation result, and the determination is made. Since the confirmation work is performed when the time comes for the confirmation work, it is possible to obtain the optimum time for confirmation according to the input information. As described above, according to the present invention, the confirmation work of the input information is performed at an optimum time, so that the time required for the confirmation work and the troublesomeness of the input person can be minimized. Further, according to the present invention, since the criterion is learned by updating the criterion according to whether the recognition result for the input is correct or incorrect, the criterion is optimized according to the user. Therefore, the input speed can be improved.

[Brief description of drawings]

FIG. 1 is a flowchart showing a method according to claim 1 of the present invention.

FIG. 2 is a block configuration diagram showing a configuration of an apparatus for carrying out the present invention.

FIG. 3 is a flowchart showing a processing procedure of the apparatus according to claim 2 of the present invention.

FIG. 4 is a block diagram of the device of the invention.

FIG. 5 is an explanatory diagram of the effect of the present invention.

FIG. 6 is a block configuration diagram of a first embodiment in which the present invention is applied to voice recognition.

FIG. 7 is a flowchart showing a confirmation timing determination procedure of the first embodiment.

FIG. 8 is a diagram showing an example of an important word list of confirmation timing judgment criteria used in the first embodiment.

FIG. 9 is a conceptual diagram of confirmation timing judgment criteria used in the first embodiment.

FIG. 10 is a diagram showing a state of an information input dialogue according to the first embodiment.

FIG. 11 is a conceptual diagram of confirmation timing determination criteria used in the second embodiment.

FIG. 12 is a block diagram of a third embodiment in which the present invention is applied to touch panel input.

FIG. 13 is a flowchart showing a confirmation timing determination procedure of the third embodiment.

FIG. 14 is a diagram showing a display example of a touch panel according to a third embodiment.

FIG. 15 is an enlarged view of the touch panel of the third embodiment.

FIG. 16 is a conceptual diagram of confirmation timing determination criteria used in the second embodiment.

FIG. 17 is a block configuration diagram of a fourth embodiment.

FIG. 18 is an overall flowchart of the fourth embodiment.

FIG. 19 is a flow chart showing a confirmation timing determination procedure of the fourth embodiment.

FIG. 20 is a conceptual diagram of confirmation timing judgment criteria applied to the fourth embodiment.

FIG. 21 is a flowchart showing the procedure of the confirmation work of the fourth embodiment.

FIG. 22 is a flow chart showing the procedure for changing the judgment criteria for the confirmation time according to the fourth embodiment.

FIG. 23 is a conceptual diagram of data for changing the judgment criterion at the time of confirmation in the fourth embodiment.

FIG. 24 is an example of a confirmation timing change criterion change criterion in the fourth embodiment.

[Explanation of symbols]

10 ... Information input means, 20 ... Dialogue state management means, 3
0 ... Confirmation timing determination means, 40 ... Recording means, 41 ...
Confirmation timing determination reference storage means, 42 ... Dialogue state recording means, 43 ... Output information storage means, 50 ... Confirmation timing determination reference calculation means, 80 ... Output information generation means, 90 ... Information output means.

Claims (2)

[Claims] 1. Each time information is input, the timing of confirmation of the input information is determined based on the evaluation information of the input information, and as a result of the determination, the information is correctly input when the confirmation timing is reached. An information input method, characterized in that confirmation processing is performed. 2. A judgment means for judging the timing of confirmation of the inputted information based on the evaluation information of the inputted information every time the information is inputted, and a criterion storing memory for storing the criterion of the determination occasion referred to in the judgment. Means, confirmation processing means for performing confirmation processing as to whether or not information is correctly input at the confirmation time determined by the determination means, storage means for storing the input accuracy information and confirmation processing result, and the storage means An information input device comprising: an updating unit for optimizing and updating the determination criterion based on the content of