JP7294637B2 - Dialogue adaptation device, dialogue adaptation method, program, and recording medium - Google Patents

Description

The present invention relates to a dialog adaptation device, dialog adaptation method, program, and recording medium.

Due to the problem of labor shortage, the use of interactive robots has been attempted in various fields. A plurality of dialogue models are stored in the dialogue robot, and a dialogue takes place between the user and the dialogue robot. Basically, it is only determined whether or not the dialogue has been established (Patent Documents 1 and 2).

2018-173456 publication 2018-147288 publication

In the field of services, it is important for users to achieve their goals through continuous interaction. However, as described above, in the interactive robot, it is only judged whether the dialogue is established. It is not considered whether

SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide a system capable of adapting a dialogue that can promote the achievement of a user's goal and give the user a sense of satisfaction, for example, through dialogue with a dialogue robot.

To achieve the above object, the dialogue adaptation device of the present invention comprises:
A storage unit, a dialogue model selection unit, an utterance output unit, a user information input unit, an effect amount calculation unit, a dialogue model evaluation unit, and a dialogue model re-selection unit,
The storage unit
Storing a dialogue model group including a plurality of dialogue models that encourage a user to achieve a goal;
The dialogue model selection unit
selecting an arbitrary dialogue model as an initial dialogue model from the dialogue model group;
The speech output unit is
outputting utterance information to the user based on the selected interaction model;
The user information input unit
Enter your user information,
The user information includes utterance information from the user,
The effect amount calculation unit is
calculating a user's goal achievement rate based on the user information as an effect amount indicating the effect of the interaction model;
The dialogue model evaluation unit
Based on the effect size and evaluation criteria,
evaluating the dialogue model as inappropriate if the effect size of the selected dialogue model does not meet the evaluation criteria;
evaluating the dialogue model as appropriate if the effect size of the selected dialogue model satisfies the evaluation criteria;
The dialogue model reselection unit
if the selected dialogue model is evaluated as inappropriate, selecting another dialogue model from the dialogue model group;
When the selected dialogue model is evaluated as appropriate, the selected dialogue model is continued.

The dialogue adaptation method of the present invention comprises:
using the memory
including a dialog model selection step, an utterance output step, a user information input step, an effect amount calculation step, a dialog model evaluation step, and a dialog model re-selection step;
The storage unit
A dialogue model group including a plurality of dialogue models that encourage the user to achieve a goal is stored,
The interaction model selection step includes:
selecting an arbitrary dialogue model as an initial dialogue model from the dialogue model group;
The speech output step includes:
outputting utterance information to the user based on the selected interaction model;
The user information input step includes:
Enter your user information,
The user information includes utterance information from the user,
The effect amount calculation step includes:
calculating a user's goal achievement rate based on the user information as an effect amount indicating the effect of the interaction model;
The dialogue model evaluation step includes:
Based on the effect size and evaluation criteria,
evaluating the dialogue model as inappropriate if the effect size of the selected dialogue model does not meet the evaluation criteria;
evaluating the dialogue model as appropriate if the effect size of the selected dialogue model satisfies the evaluation criteria;
The interaction model reselection step includes:
if the selected dialogue model is evaluated as inappropriate, selecting another dialogue model from the dialogue model group;
When the selected dialogue model is evaluated as appropriate, the selected dialogue model is continued.

A program of the present invention causes a computer to execute the dialogue adaptation method of the present invention.

A recording medium of the present invention is a computer-readable recording medium recording the program of the present invention.

A dialog system of the present invention is characterized by including the adaptation device of the present invention, and a specific example thereof is a dialog robot.

According to the present invention, the interaction model selected for the user is evaluated by the amount of effect based on the information from the user. A dialogue model can be selected again, and utterance information to the user based on it can be output to the user. For this reason, for example, even in a field where there is a shortage of manpower, it is possible to have a dialogue based on an optimum dialogue model that leads to the achievement of the user's goal. In addition, even for users who have difficulty going out, it is possible to have conversations suitable for achieving their goals.

FIG. 1 is a block diagram showing an example of a dialogue adaptation device according to Embodiment 1. FIG. FIG. 2 is a block diagram showing an example of the hardware configuration of the dialogue adaptation device according to the first embodiment; FIG. 3 is a flow chart showing the procedure of the dialogue adaptation method in the first embodiment. FIG. 4 is a flowchart showing a procedure for further repeating the output of utterance information to a user and the input of utterance information from the user to calculate the effect amount in the first embodiment.

The dialogue adaptation apparatus and dialogue adaptation method of the present invention perform dialogue adaptation suitable for achieving a user's goal. The present invention can be used, for example, in various services, and the field of service is not limited at all. According to the present invention, for example, using the selected interaction model, as a result of interacting with the user, it is evaluated whether or not the user is getting an effect in achieving the goal, and if the effect is insufficient, When determined, a dialog model suitable for the user is reselected and an output is provided to the user, so that a more adaptive dialog for the user is possible in order to achieve the goal. Thus, the present invention can be an apparatus and method for achieving user-set goals. The types of the services are not limited at all, and examples thereof include exercise support such as fitness and running, learning support such as self-study, dietary restrictions, and health management support such as continuation of smoking cessation.

An embodiment of the present invention will be described. In addition, this invention is not limited to the following embodiment. In addition, in the following figures, the same reference numerals are given to the same parts. Also, the descriptions of the respective embodiments can be referred to each other's description unless otherwise specified. Furthermore, the configuration of each embodiment can be combined unless otherwise specified.

[Embodiment 1]
FIG. 1 is a block diagram showing an example configuration of a dialogue adaptation device 10 of this embodiment. As shown in FIG. 1A, the dialogue adaptation device 10 includes a storage unit 11, a dialogue model selection unit 12, an utterance output unit 13, a user information input unit 14, an effect amount calculation unit 15, a dialogue model evaluation unit 16, and a dialogue model re-selection unit 17 . The dialogue adaptation device 10 is also called, for example, a dialogue adaptation system. The dialogue adaptation device 10 may be, for example, one device including the above units, or may be a device in which the above units can be connected via a communication network.

The communication network is not particularly limited, and a known communication network can be used, and may be wired or wireless. The communication network includes, for example, the Internet line, telephone line, LAN (Local Area Network), WiFi (Wireless Fidelity), and the like. The processing of each part of the dialogue adaptation device 10 may be performed on the cloud.

The dialogue adaptation device 10 itself may be, for example, in the form of a dialogue robot, or as shown in FIG. may

Next, FIG. 2 illustrates a block diagram of the hardware configuration of the dialogue adaptation device 10. As shown in FIG. The dialog adaptation device 10 includes, for example, a CPU (Central Processing Unit) 101, a memory 102, a bus 103, an input device 104, a display 105, a communication device 106, a storage device 107, and also, for example, a speaker 109, a microphone 110, camera 111, sensor 112, and the like. Each part of the dialogue adaptation device 10 is interconnected via a bus 103 by respective interfaces (I/F).

The CPU 101 is a processor responsible for overall control of the dialogue adaptation device 10, and is not limited to a CPU, and may be another processor. In the dialogue adaptation device 10, the CPU 101 executes, for example, the program of the present invention and other programs, and reads and writes various information.

The dialogue adaptation device 10 can be connected to a communication network, for example, by means of a communication device 106 connected to the bus 103, and can also be connected to external equipment via said communication network. The external device is not particularly limited, and may be, for example, a terminal, a PC (personal computer), or the like, and the terminal may be, for example, a tablet, a smart phone, a mobile phone, a robot, or the like. The connection method between the dialogue adaptation device 10 and the external device is not particularly limited, and may be wired connection or wireless connection, for example. The wired connection may be, for example, a cord connection or a cable connection for using a communication network. The wireless connection may be, for example, a connection using a communication network or a connection using wireless communication. The communication network is not particularly limited, and for example, a known communication network can be used, which is the same as described above.

The memory 102 includes, for example, a main memory, which is also referred to as main memory. When the CPU 101 performs processing, the memory 102 reads various operation programs 108 such as the program of the present invention stored in an auxiliary storage device described later, and the CPU 101 receives data from the memory 102. , executes the program 108 . The main memory is, for example, RAM (random access memory). Memory 102, for example, further includes a ROM (Read Only Memory).

The storage device 107 is also called a so-called auxiliary storage device, for example, in contrast to the main memory (main storage device). The storage device 107 includes, for example, a storage medium and a drive that reads from and writes to the storage medium. The storage medium is not particularly limited, and may be, for example, a built-in type or an external type, and includes HD (hard disk), CD-ROM, CD-R, CD-RW, MO, DVD, flash memory, memory card, and the like. and the drive is not particularly limited. The storage device 107 can also be exemplified by a hard disk drive (HDD) in which a storage medium and a drive are integrated, for example. For example, the storage device 107 stores the program 108 as described above, and the memory 102 reads the operation program 108 from the storage device 107 when the CPU 101 is executed as described above. Also, the storage device 107 includes, for example, the storage unit 11 .

The input device 104 is, for example, a scanner, touch panel, keyboard, or the like. The display 105 may be, for example, an LED display, a liquid crystal display, or the like, and the dialogue model may be output to the display 105 as characters, for example.

The dialogue model may be output from the speaker 109 as sound, for example. The microphone 110 acquires, for example, the user's dialogue, and the camera 111 acquires, for example, the user's face, actions, etc. as an image. The sensor 112 includes, for example, a temperature sensor, an acceleration sensor, a vital sensor such as a pulse meter, a heart rate meter, a blood pressure meter, and a thermometer, and acquires user's behavior as sensing data.

The storage unit 11 stores a dialogue model group including a plurality of dialogue models that encourage the user to achieve a goal. In this embodiment, a "user" is a user who interacts with the dialogue adaptation device 10, and a "dialogue model" is, for example, a form of dialogue having characteristics in coaching. The type of the dialogue model is not particularly limited, and examples include a praise and encouragement dialogue model, a logic-based explanation dialogue model, a severely pointed out dialogue model, a joke-returning dialogue model, a parable-returning dialogue model, and the like. is. The dialogue model may be, for example, an existing model, or may be created by machine learning using voice data of utterances as learning data.

The storage unit 11 stores user information in association with user identification information. The user information is user's utterance information, specifically, utterance information from the user to the dialogue adaptation device 10 . The user's utterance information is, for example, utterance information from the user to the dialogue adaptation device 10 when the dialogue adaptation device 10 itself is a dialogue robot, and when the external device 40 is a dialogue robot, for example, It is utterance information from the user to the external device 40 and is the utterance information input from the external device 40 . The identification information is, for example, information that can uniquely identify the user, such as name, ID, and ID of the external device. The user information is input by the user information input unit 14, for example. The user information is preferably associated with, for example, the date and time when the user information was acquired. In the following description, the utterances from the user are exemplified as utterances to the dialogue adaptation device 10, but are not limited to this, and the same applies to the external device 40, which is the dialogue robot, for example.

The user information may include, for example, behavior information of the user as described later. The user's behavior information is, for example, acquired and input in parallel with the utterance information from the user. Examples of the behavior information include image information obtained by imaging the user's expression, posture, movement, etc., sensing information obtained by sensing the user's body temperature, movement, etc., and voice information such as the pitch and level of voice related to the user's utterance. .

The user information may, for example, further include questionnaire information answered by the user. The questionnaire information may be, for example, questionnaire information obtained before the initial selection of the dialogue model, questionnaire information obtained when the selected dialogue model is evaluated as inappropriate, or both. . The content of the questionnaire is not particularly limited, and preferably includes, for example, a dialogue model preferred by the user, a goal of the user, results perceived by the user, an opinion of the user on the dialogue model, and the like. If the questionnaire information is obtained before the initial selection of the dialogue model, for example, the dialogue model more suitable for the user can be selected in the initial selection of the dialogue model. Further, if the questionnaire information is questionnaire information obtained when the selected dialogue model is evaluated as inappropriate, for example, when the dialogue model is reselected, a dialogue model more suitable for the user can be selected.

The dialogue model selection unit 12 selects an arbitrary dialogue model from the dialogue model group in the storage unit 11 as an initial dialogue model. The initial dialogue model selected first is not particularly limited, and any dialogue model can be selected. Further, when the questionnaire information is input in advance, for example, a dialogue model more suitable for the user can be selected based on the questionnaire information. As will be described later, when the selected dialogue model is evaluated as inappropriate by the dialogue model evaluation unit 16, a new dialogue model is reselected by the dialogue model reselection unit 17, which will be described later.

The speech output unit 13 outputs speech information to the user based on the selected dialogue model. Specifically, the speech output unit 13 generates speech information for the user based on the selected dialogue model, and outputs the information. The utterance information may be in a form that allows the user to recognize the content, and may be, for example, voice or text. In the case of voice, for example, it is output from the speaker 109, and in the case of text, it is output as a display on the display 105, for example.

The user information input unit 14 inputs the user information. The user information includes utterance information from the user, as described above. The utterance information is, for example, utterance information spoken by the user to the dialogue adaptation device 10, as described above. If the dialogue adaptation device 10 has the microphone 110 , for example, the microphone 110 may acquire speech information from the user and the user information input unit 14 may input the information. In addition, when the microphone of the external device 40 acquires the speech information from the user, for example, it may be received by the communication device 106 via the communication network 20 and input by the user information input unit 14 . The utterance information from the user may be, for example, voice information, information obtained by language-processing voice, or character information.

The user information input unit 14 may input the behavior information of the user as described above in addition to the utterance information from the user, and may input the questionnaire information as the user information. .

The behavior information is, for example, information about user behavior. Examples of the behavior information include image information obtained by imaging the user's expression, posture, movement, etc., sensing information obtained by sensing the user's body temperature, movement, etc., and voice information such as the pitch and level of voice related to the user's utterance. . The behavior information may be acquired, for example, in parallel with the speech information from the user, and more specifically, it is also information about the user's behavior while speaking to the dialogue adaptation device 10 . If the dialogue adaptation device 10 has, for example, a microphone 110 , a camera 111 and a sensor 112 , information acquired by these may be input by the user information input unit 14 . Further, when the user's behavior information is acquired by the external device 40 , for example, it may be received by the communication device 106 via the communication network 20 and input by the user information input unit 14 .

Since the utterance information from the user and the behavior information of the user, which are the user information, are used to calculate the amount of effect of the selected dialogue model, during the dialogue period with the user using the selected dialogue model, User information is preferably stored in the storage unit 11 . Then, for example, after a predetermined period of time has passed, it is preferable that the effect amount is calculated by the effect amount calculation unit 15 based on the accumulated user information. The predetermined period is not particularly limited, and the lower limit is, for example, one day, one week, two weeks, etc., or in units of days or weeks, and the upper limit is, for example, in units of months, such as one month or three months. .

Based on the user information, the effect amount calculation unit 15 calculates a user's target achievement rate as an effect amount indicating the effect of the selected interaction model. For example, the effect amount calculation unit 15 may further calculate the user's stress numerical value as the effect amount based on the user information. The achievement of the goal includes, for example, subjective achievement and objective achievement. The subjective achievement is, for example, an achievement subjectively perceived as fun, happy, or feeling good, and the objective achievement is, for example, a target value included in a goal such as distance, time, or weight. It is an objective achievement such as approaching the target value or clearing the target value. The unit of the target achievement rate can be arbitrarily set, and its numerical value can be represented by, for example, a relative numerical value. The method for calculating the target achievement rate is not particularly limited, and can be set as appropriate according to the type of target, the type of achievement (for example, subjective, objective), and the like. The stress numerical value is a numerical value indicating the degree of stress that the user feels consciously or unconsciously. The unit of the stress numerical value can be arbitrarily set, and the numerical value can be represented by, for example, a relative numerical value. Also, the target achievement rate and the stress numerical value can be represented by, for example, "high, medium, low", "good, acceptable, unacceptable" or the like according to the numerical range. The "low" goal achievement rate means, for example, that the goal has not been achieved much, that is, "impossible", and the "low" stress value means, for example, that there is not much stress, that is, " means good.

A dialogue model evaluation unit 16 evaluates the effect of the output dialogue model based on the effect amount and evaluation criteria. Specifically, if the effect size of the selected interaction model does not satisfy the evaluation criteria, the interaction model is evaluated as unsuitable for the user, that is, the interaction model is inappropriate, and the effect size of the selected interaction model is evaluated as If the evaluation criteria are met, the interaction model is evaluated as suitable for the user, that is, as appropriate. For the target achievement rate and the stress numerical value, which are the effect amounts, for example, an evaluation criterion may be provided for each, or an evaluation criterion combining both may be provided. Moreover, it is also possible to use an evaluation criterion that gives priority to one of them, and for example, it is preferable to give priority to the target achievement rate.

A dialogue model re-selection unit 17 selects another dialogue model from the dialogue model group when the selected dialogue model is evaluated as inappropriate, and when the selected dialogue model is evaluated as appropriate, the selected dialogue continue the model. The selection method of the other dialogue model is not particularly limited. Based on the effect amount, another interaction model may be arbitrarily selected, and in addition to the effect amount, the previous questionnaire information, the questionnaire information for the output interaction model, or both questionnaire information Other interaction models may be selected based on

Further, when the selected dialogue model is evaluated as inappropriate and the effect amount exceeds zero, for example, another dialogue model similar to the selected dialogue model is reselected, and the selected dialogue model is is inappropriate and the effect size is less than or equal to zero, another interaction model dissimilar to the selected interaction model may be reselected. The similarity determination method is not particularly limited, and for example, feature items for each interaction model may be used. That is, in the storage unit 11, each of the plurality of dialogue models of the dialogue model group may be stored in association with characteristic items, and the dialogue model re-selecting unit 17 may select other similar models based on the characteristic items. Interaction models and dissimilar other interaction models can also be reselected.

Next, the dialogue adaptation method of this embodiment will be described using the flowchart of FIG. The flowchart of FIG. 3 is an example for explaining the selection of a dialogue model in the dialogue adaptation method. The dialog adaptation method of the present embodiment can be implemented using, for example, the dialog adaptation device 10 shown in FIGS. 1 and 2. FIG. Note that the dialogue adaptation method of this embodiment is not limited to the use of the dialogue adaptation device 10 .

The dialogue adaptation method of this embodiment uses a storage unit and includes a dialogue model selection step, an utterance output step, a user information input step, an effect size calculation step, a dialogue model evaluation step, and a dialogue model reselection step. For the storage unit, for example, the storage unit in the dialogue adaptation device of this embodiment can be used.

The storage unit stores the dialogue model group as described above, and further stores preliminary questionnaire information for each user, for example (S100).

The dialogue model selection step selects an arbitrary dialogue model as an initial dialogue model from the dialogue model group (S101). The initial dialogue model can be selected, for example, based on the user's tendency according to the preliminary questionnaire. This step can be executed, for example, by the dialogue model selection unit 12 of the dialogue adaptation device 10 .

The speech output step outputs speech information D to the user based on the selected dialogue model. Specifically, based on the selected dialogue model, utterance information D for the user is generated and output to the user (S102). This step can be executed by the speech output unit 13 of the dialogue adaptation device 10, for example.

The user information input step acquires and inputs utterance information U from the user as user information (S103). This step can be performed, for example, by the user information input unit 14 of the dialogue adaptation device 10 .

The output of the speech information D to the user (S102) and the input of the speech information U from the user (S103) are repeated, so that a dialogue between the user and the dialogue adaptation device 10 is performed.

In parallel with the input of the utterance information U (S103), the user's action information is acquired and input (S104). This step can be performed, for example, by the user information input unit 14 of the dialogue adaptation device 10 .

In the effect amount calculation step, the user's goal achievement rate is calculated based on the user information as an effect amount indicating the effect of the dialogue model (S105). In the effect amount calculation step, for example, the user's stress numerical value may be further calculated as the effect amount based on the user information. This step can be executed by the effect amount calculation unit 15 of the dialogue adaptation device 10, for example.

The dialogue model evaluation step evaluates the effect of the output dialogue model based on the effect amount and evaluation criteria. Specifically, depending on whether or not the effect amount satisfies the evaluation criteria (S106), if the effect amount of the selected dialogue model does not satisfy the evaluation criteria, the dialogue model is evaluated as inappropriate, and the If the effect size of the selected interaction model satisfies the evaluation criteria, the interaction model is evaluated as appropriate. This step can be performed, for example, by the dialogue model evaluation unit 16 of the dialogue adaptation device 10 .

If the selected dialogue model does not satisfy the evaluation criteria (NO) and is evaluated as inappropriate, the dialogue model reselection step reselects another dialogue model from the dialogue model group (S107). , if the selected dialogue model satisfies the evaluation criteria (YES) and is evaluated as appropriate, the dialogue model reselection step continues to use the selected dialogue model (S108). The dialogue model reselecting step, for example, reselects another dialogue model similar to the selected dialogue model if the effect size exceeds zero, and reselects the selected dialogue model if the effect size is less than or equal to zero. Other interaction models dissimilar to the model may be reselected. The similar other interaction model and the dissimilar other interaction model may be reselected, for example, based on the features of the interaction model. This step can be executed, for example, by the dialogue model reselection unit 17 of the dialogue adaptation device 10 .

If the selected dialogue model is evaluated as inappropriate, for example, a questionnaire is further conducted for the user, and the questionnaire information is further stored in the storage unit 11 and used for reselection of another model (S107). is preferred. At this time, it is preferable that the questionnaire has contents for extracting points of dissatisfaction with respect to the dialogue model evaluated as inappropriate, for example.

Satisfying the user by continuing the interaction between the dialogue adaptation device 10 and the user, evaluating the effect of the selected dialogue model and reselecting a more suitable dialogue model as described above, and , it enables an ongoing dialogue with the user, which also effectively assists the user in achieving his or her goals.

Here, the steps from S100 to S108 in the dialog adaptation method of the present embodiment will be described in more detail, using the flowchart of FIG. 4 as an example. In this example, the output of speech information D to the user (S102) and the input of speech information U from the user (S103) are repeated until the effect amount is calculated (S105).

It is determined whether or not the dialogue has ended in the dialogue in which the input of the utterance information U from the user (S103) and the output of the utterance information D to the user (S102) are repeated (S200). The end of the dialogue here means the end of continuous dialogue between the user and the dialogue adaptation device 10 . If it is determined that the dialogue has not ended (NO), it is assumed that the dialogue is continuing. Continue the input (S103). On the other hand, if it is determined that the dialogue has ended (YES), it is determined whether or not a predetermined period has passed since the selected dialogue model was started to be used (S201).

If the predetermined period has not passed (NO), for example, it is determined that sufficient time has not passed to confirm the effect of the selected dialogue model, and the use of the dialogue model is continued in the dialogue. On the other hand, if the predetermined period has passed (YES), for example, it is assumed that sufficient time has passed to confirm the effect of the selected dialogue model. (S202), and the effect amount is calculated (S105). After calculating the effect amount, for example, it is as described above.

[Embodiment 2]
Next, the dialogue adaptation method using the dialogue adaptation apparatus 10 of the present invention will be specifically described by taking fitness as an example. In this embodiment, the dialogue adaptation device 10 is installed as a dialogue robot at the reception desk, and the user has an environment in which he interacts with the dialogue adaptation device 10 before and after training.

First, a user is given a questionnaire in advance, and the questionnaire information is stored in the storage unit 11 of the dialogue adaptation device 10 . Then, if the preliminary questionnaire information indicates that the user is "a type that encourages praise", the user selects "a dialogue model that encourages praise" from the dialogue model group in the storage unit 11, and selects this as the initial dialogue model. and

Next, it generates utterance information D for the user based on the initial dialogue model, outputs it to the user, conducts dialogue with the user, and inputs utterance information U from the user. In the first dialogue, a goal set by the user is extracted from the utterance information U, a recommendation for achieving the goal is set, and the speech is given to the user who inserts this into the "praise and stretch dialogue model". Information D is output to the user. The extraction of the target can be performed, for example, by language processing of the utterance information U from the user. In addition, for example, recommendation candidates for various goals can be stored in the storage unit 11 in advance, and can be called from the storage unit 11 according to the extracted goals. Further, the generation of the utterance information D for the user by inserting the recommendation can utilize, for example, an existing program for generating sentences.

As a specific example, when the utterance information U from the user is "My target weight is 60 kg and I want to run 5 km every day", the goal "Weight is 60 kg and I want to run 5 km every day" can be extracted from the utterance information U. Then, as a recommendation to achieve this, "2km in the morning and 3km at night" is set, and this is inserted into the "praise and stretch dialogue model", and "great! Good luck! 2km in the morning and 3km at night". Speech information D for the user saying "Why don't you try 3km as a goal?" is generated and output to the user.

Before and after the user's training, dialogue with the user is performed to acquire utterance information U from the user including the content and results of the training. The utterance information D to the user is generated by, for example, extracting the contents and results from the utterance information U from the user, inserting recommendations for the extracted information, and generating the utterance information D based on the initial dialogue model. Output for

Here, a specific example of dialogue is shown below.
User: “I ran 3km .
Apparatus 10: "You've been doing great this past month. You've been running less distance this past week ."
User: "I'm fine with this. I haven't lost any weight recently, so I'll continue. "

Regarding such a dialogue, based on the utterance information U from the user, the target achievement rate and stress value of the user are calculated as the effect amount of the initial dialogue model "praise and stretch dialogue model". The effect amount is calculated, for example, after a predetermined period of time has passed (for example, after one week has passed).

Then, the dialogue based on the initial dialogue model is continued, and when an arbitrary duration has passed, it is evaluated whether or not the effect amount satisfies the evaluation criteria, and the initial dialogue model is continuously used. or re-select another interaction model. For example, regarding the effect amount, it is assumed that the target achievement rate is "low" and the stress value is "medium" during the duration period. The target achievement rate is a priority evaluation target, the evaluation standard for the target achievement rate is "medium or higher (evaluated as appropriate)", and the evaluation standard for the stress numerical value is "medium or lower (evaluated as appropriate)" , there is no need to newly reselect the initial dialogue model as it is appropriate for the stress numerical value, but the target achievement rate, which is a priority evaluation target, does not meet the evaluation criteria and needs to be further improved. Therefore, it is determined that the initial dialogue model needs to be reselected for a new dialogue model.In addition, when the target achievement rate is "medium" and the stress value is "high", the same is true and when the evaluation criteria for the target achievement rate is "middle or higher (evaluated as appropriate)" and the evaluation criteria for the stress numerical value is "middle or lower (evaluated as appropriate)", the target achievement rate is appropriate However, since the stress numerical value to be evaluated does not satisfy the evaluation criteria and needs to be further improved, the initial dialogue model is It is determined that re-selection of the appropriate interaction model is necessary.

When reselection is required, another dialogue model different from the initial dialogue model is selected from the storage unit 11, and dialogue based on the newly selected dialogue model is started. On the other hand, if reselection is unnecessary, the initial dialogue model is reselected as it is, and the dialogue based on the initial dialogue model is continued. The newly selected dialogue model is similarly evaluated based on the input user information, and it is appropriately determined whether or not the new dialogue model is reselected.

In the case of the specific example of the dialogue, depending on the initial response model "praise and extend dialogue model", the user is not able to achieve the goal and is not able to maintain the motivation to achieve the goal. As a new dialogue model, it is also possible to select "a dialogue model that treats severely when the goal cannot be achieved".

For example, the evaluation criteria can be set for each of the target achievement rate and the stress numerical value. Further, the evaluation criteria may preferentially use the result of either the target achievement rate or the stress numerical value in the evaluation. For example, the goal achievement rate may be preferentially used. preferable. For example, if the stress numerical value satisfies the evaluation criteria but the target achievement rate does not satisfy the evaluation criteria, a new dialogue model is selected from the dialogue model group based on the fact that the target achievement rate does not satisfy the evaluation criteria. may be determined by selecting

[Embodiment 3]
In the present embodiment, the target achievement rate, the stress numerical value, the method of calculating the effect amount, the method of calculating the evaluation of the dialogue model, the method of calculating the reselection of the dialogue model, and the like will be exemplified. Note that these calculation methods are examples, and the present invention is not limited to these.

(1) Goal Achievement Rate As a method of calculating the goal achievement rate, for example, the following formula can be used.

(2) Stress Numerical Value A method for calculating the stress numeric value is, for example, the following formula. The method of calculating the stress value is not limited to this, and for example, a method of calculating the stress value by calculating the pulse of the user from image information, or estimating the emotion from voice information and calculating the stress value. It's okay.

(3) Effect Amount A method for calculating the effect amount is, for example, the following formula.

(4) Evaluation of Dialogue Model The evaluation of whether the dialogue model is inappropriate or appropriate, that is, the evaluation of whether to continue or reselect can be evaluated by the following equation, for example.

(5) Reselection of Dialogue Model Reselection of the dialogue model can be performed, for example, by using the effect amount, the questionnaire, and the information of the dialogue model group according to the following formula.

The function f _r reselects the model _c based on the user's request of the questionnaire quest _c when the effect size ef _st or ef _g is 0≦0. At this time, in order not to lose the goodness of the current dialogue model, for example, a new dialogue model is reselected in which items with effect amounts of 0 or more are similar to the current dialogue model. As a specific example, when ef _st =−10 and ^∞ ef _g =10, the questionnaire quest _c (st ₁ , st ₂ , . . . st _n ) items and the dialogue model items (st ₁ , st ₂ , . st _n ) is high, and the items of the questionnaire quest _c (g ₁ , g ₂ , . . . g _n ) and the items of the dialogue model (g ₁ , g ₂ , _. It is preferable to select an interaction model with a high degree of similarity. “st _n ” may be, for example, the score of a question item in a questionnaire or the factor loading amount.

[Embodiment 4]
A program of the present embodiment is a program capable of executing the dialogue adaptation method of the present invention on a computer. Alternatively, the program of this embodiment may be recorded on a computer-readable recording medium, for example. The recording medium is not particularly limited, and includes, for example, the above-described storage medium.

Although the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

<Appendix>
Some or all of the above-described embodiments and examples can be described as in the following appendices, but are not limited to the following.
(Appendix 1)
A storage unit, a dialogue model selection unit, an utterance output unit, a user information input unit, an effect amount calculation unit, a dialogue model evaluation unit, and a dialogue model re-selection unit,
The storage unit
Storing a dialogue model group including a plurality of dialogue models that encourage a user to achieve a goal;
The dialogue model selection unit
selecting an arbitrary dialogue model as an initial dialogue model from the dialogue model group;
The speech output unit is
outputting utterance information to the user based on the selected interaction model;
The user information input unit
Enter your user information,
The user information includes utterance information from the user,
The effect amount calculation unit is
calculating a user's goal achievement rate based on the user information as an effect amount indicating the effect of the interaction model;
The dialogue model evaluation unit
Based on the effect size and evaluation criteria,
evaluating the dialogue model as inappropriate if the effect size of the selected dialogue model does not meet the evaluation criteria;
evaluating the dialogue model as appropriate if the effect size of the selected dialogue model satisfies the evaluation criteria;
The dialogue model reselection unit
if the selected dialogue model is evaluated as inappropriate, selecting another dialogue model from the dialogue model group;
If the selected interaction model is evaluated as appropriate, continue with the selected interaction model;
A dialog adaptation device characterized by:
(Appendix 2)
2. The dialogue adaptation apparatus according to Supplementary Note 1, wherein the effect amount calculation unit calculates, as the effect amount, the target achievement rate and the user's stress numerical value based on the user information.
(Appendix 3)
3. A dialogue adaptation apparatus according to any one of the appendices 1 or 2, wherein the user information further comprises user behavior information.
(Appendix 4)
3. The dialogue adaptation apparatus according to appendix 3, wherein the user's behavior information is at least one of user's image information and sensing information.
(Appendix 5)
The dialogue model reselection unit
If the selected interaction model is evaluated as inappropriate and the effect size is greater than zero, reselecting another interaction model similar to the selected interaction model;
5. The method of claim 1, wherein if the selected interaction model is evaluated as inappropriate and the effect size is less than or equal to zero, reselecting another interaction model that is dissimilar to the selected interaction model. dialogue adaptor.
(Appendix 6)
each of the plurality of dialogue models in the dialogue model group is associated with a characteristic item and stored;
6. The dialogue adaptation apparatus according to appendix 5, wherein the dialogue model reselection unit reselects another similar dialogue model and another dissimilar dialogue model based on the feature item.
(Appendix 7)
The storage unit
7. A dialogue adaptation device according to any one of appendices 1 to 6, wherein said input user information is stored.
(Appendix 8)
The user information input unit
In addition, enter the user's questionnaire information,
The storage unit
storing the input user information;
The dialogue model reselection unit
8. The dialogue adaptation device according to any one of appendices 1 to 7, wherein the other dialogue model is reselected based on the effect amount and the questionnaire information.
(Appendix 9)
using the memory
including a dialog model selection step, an utterance output step, a user information input step, an effect amount calculation step, a dialog model evaluation step, and a dialog model re-selection step;
The storage unit
A dialogue model group including a plurality of dialogue models that encourage the user to achieve a goal is stored,
The interaction model selection step includes:
selecting an arbitrary dialogue model as an initial dialogue model from the dialogue model group;
The speech output step includes:
outputting utterance information to the user based on the selected interaction model;
The user information input step includes:
Enter your user information,
The user information includes utterance information from the user,
The effect amount calculation step includes:
calculating a user's goal achievement rate based on the user information as an effect amount indicating the effect of the interaction model;
The dialogue model evaluation step includes:
Based on the effect size and evaluation criteria,
evaluating the dialogue model as inappropriate if the effect size of the selected dialogue model does not meet the evaluation criteria;
evaluating the dialogue model as appropriate if the effect size of the selected dialogue model satisfies the evaluation criteria;
The interaction model reselection step includes:
if the selected dialogue model is evaluated as inappropriate, selecting another dialogue model from the dialogue model group;
If the selected interaction model is evaluated as appropriate, continue with the selected interaction model;
A dialogue adaptation method characterized by:
(Appendix 10)
10. The dialogue adaptation method according to appendix 9, wherein the effect amount calculating step calculates, as the effect amount, the target achievement rate and the user's stress numerical value based on the user information.
(Appendix 11)
11. A dialog adaptation method according to clause 9 or 10, wherein said user information further comprises user behavior information.
(Appendix 12)
12. The dialogue adaptation method according to appendix 11, wherein the user's behavior information is at least one of user's image information and sensing information.
(Appendix 13)
The interaction model reselection step includes:
If the selected interaction model is evaluated as inappropriate and the effect size is greater than zero, reselecting another interaction model similar to the selected interaction model;
13. Any of clauses 9 to 12, wherein if the selected interaction model is evaluated as inappropriate and the effect size is less than or equal to zero, then reselecting another interaction model that is dissimilar to the selected interaction model. Dialogue adaptation method.
(Appendix 14)
each of the plurality of dialogue models in the dialogue model group is associated with a characteristic item and stored;
14. The dialogue adaptation method according to appendix 13, wherein the dialogue model selection step selects another similar dialogue model and another dissimilar dialogue model based on the feature item.
(Appendix 15)
The storage unit
15. A dialogue adaptation method according to any one of appendices 9 to 14, wherein said input user information is stored.
(Appendix 16)
The user information input step includes:
In addition, enter the user's questionnaire information,
The storage unit
storing the input user information;
The interaction model reselection step includes:
16. The dialogue adaptation method according to any one of appendices 9 to 15, wherein the other dialogue model is reselected based on the effect size and the questionnaire information.
(Appendix 17)
17. A program for causing a computer to execute the dialogue adaptation method according to any one of appendices 10 to 16.
(Appendix 18)
A computer-readable recording medium recording the program according to appendix 17.
(Appendix 19)
9. An interactive robot comprising a dialogue adaptation device according to any one of appendices 1 to 8.

According to the present invention, the interaction model selected for the user is evaluated by the amount of effect based on the information from the user. A dialogue model can be selected again, and utterance information to the user based on it can be output to the user. For this reason, for example, even in a field where there is a shortage of manpower, it is possible to have a dialogue based on an optimum dialogue model that leads to the achievement of the user's goal. In addition, even for users who have difficulty going out, it is possible to have conversations suitable for achieving their goals.

10 Dialogue adaptation device 11 Storage unit 12 Dialogue model selection unit 13 Speech output unit 14 User information input unit 15 Effect amount calculation unit 16 Dialogue model evaluation unit 17 Dialogue model reselection unit

Claims (10)

A storage unit, a dialogue model selection unit, an utterance output unit, a user information input unit, an effect amount calculation unit, a dialogue model evaluation unit, and a dialogue model re-selection unit,
The storage unit
Storing a dialogue model group including a plurality of dialogue models that encourage a user to achieve a goal;
The dialogue model selection unit
selecting an arbitrary dialogue model as an initial dialogue model from the dialogue model group;
The speech output unit is
outputting utterance information to the user based on the selected interaction model;
The user information input unit
Enter your user information,
The user information includes utterance information from the user,
The effect amount calculation unit is
calculating a user's goal achievement rate based on the user information as an effect amount indicating the effect of the interaction model;
The dialogue model evaluation unit
Based on the effect size and evaluation criteria,
evaluating the dialogue model as inappropriate if the effect size of the selected dialogue model does not meet the evaluation criteria;
evaluating the dialogue model as appropriate if the effect size of the selected dialogue model satisfies the evaluation criteria;
The dialogue model reselection unit
if the selected dialogue model is evaluated as inappropriate, selecting another dialogue model from the dialogue model group;
If the selected interaction model is evaluated as appropriate, continue with the selected interaction model;
A dialog adaptation device characterized by: 2. The dialogue adaptation apparatus according to claim 1, wherein said effect amount calculation unit calculates said target achievement rate and said user's stress numerical value as said effect amount based on said user information. 3. A dialogue adaptation device according to claim 1 or 2, wherein said user information further comprises user behavior information. 4. The dialogue adaptation apparatus according to claim 3, wherein said user's behavior information is at least one of user's image information and sensing information. The dialogue model reselection unit
If the selected interaction model is evaluated as inappropriate and the effect size is greater than zero, reselecting another interaction model similar to the selected interaction model;
5. If the selected interaction model is evaluated as inappropriate and the effect size is less than or equal to zero, reselecting another interaction model that is dissimilar to the selected interaction model. A dialog adaptation apparatus as described in . each of the plurality of dialogue models in the dialogue model group is associated with a characteristic item and stored;
6. The dialogue adaptation apparatus according to claim 5, wherein said dialogue model reselection unit reselects another similar dialogue model and another dissimilar dialogue model based on said feature item. The user information input unit
In addition, enter the user's questionnaire information,
The storage unit
storing the input user information;
The dialogue model reselection unit
7. The dialogue adaptation device according to any one of claims 1 to 6, wherein said other dialogue model is reselected based on said effect amount and said questionnaire information. using the memory
including a dialog model selection step, an utterance output step, a user information input step, an effect amount calculation step, a dialog model evaluation step, and a dialog model re-selection step;
The storage unit
A dialogue model group including a plurality of dialogue models that encourage the user to achieve a goal is stored,
The interaction model selection step includes:
selecting an arbitrary dialogue model as an initial dialogue model from the dialogue model group;
The speech output step includes:
outputting utterance information to the user based on the selected interaction model;
The user information input step includes:
Enter your user information,
The user information includes utterance information from the user,
The effect amount calculation step includes:
calculating a user's goal achievement rate based on the user information as an effect amount indicating the effect of the interaction model;
The dialogue model evaluation step includes:
Based on the effect size and evaluation criteria,
evaluating the dialogue model as inappropriate if the effect size of the selected dialogue model does not meet the evaluation criteria;
evaluating the dialogue model as appropriate if the effect size of the selected dialogue model satisfies the evaluation criteria;
The interaction model reselection step includes:
if the selected dialogue model is evaluated as inappropriate, selecting another dialogue model from the dialogue model group;
If the selected interaction model is evaluated as appropriate, continue with the selected interaction model;
A dialogue adaptation method characterized by: A program for causing a computer to execute the dialogue adaptation method according to claim 8. A computer-readable recording medium recording the program according to claim 9 .

Images