JP2023001577A - Avatar providing device, avatar providing method, and program - Google Patents

Abstract

An avatar providing device that provides an avatar capable of realizing at least one of appropriate interaction and automatic processing is realized.
The present invention includes two or more avatars having at least one of a dialogue function and an automatic processing function, a set of the dialogue function and a learning function related to dialogue, and a set of the automatic processing function and an automatic processing additional function. an avatar storage unit 101 storing two or more avatar information for outputting two or more avatars including one or more avatars having at least one set of functions of; and selecting one of the two or more avatars An instruction receiving unit 111 receives a selection instruction from the terminal device 2; an avatar acquisition unit 113 acquires avatar information of an avatar corresponding to the selection instruction from the avatar storage unit 101; and an avatar transmission unit 114 that transmits to the device 2 .
[Selection drawing] Fig. 1

Description

The present invention relates to an apparatus for providing avatars, and the like.

As a conventional technique, a plurality of avatars assigned to each user are made to appear on a conversation screen on a network, and messages sent from users are displayed and output through the conversation screen to allow users to converse with each other. A network chat service that does this has been known (see, for example, Patent Document 1).

Japanese Patent Application Laid-Open No. 2009-223419 (page 1, FIG. 1, etc.)

However, in conventional avatars that have a function of interacting with a user or avatars that have a function of performing automatic processing, the dialogue model that the avatar uses for dialogue and the automatic processing that the avatar can execute are predetermined. Therefore, there is a problem that it is difficult for the avatar to realize appropriate dialogue and automatic processing in accordance with the user's request.

SUMMARY OF THE INVENTION The present invention has been made to solve the problems described above, and an object of the present invention is to provide an avatar providing apparatus and the like capable of providing an avatar capable of realizing at least one of appropriate dialogue and automatic processing.

The avatar providing device of the present invention is two or more avatars having at least one function of a dialogue function and an automatic processing function, the combination of the dialogue function and the learning function related to the dialogue, and the addition of the automatic processing function and the automatic processing. an avatar storage unit storing two or more pieces of avatar information for outputting two or more avatars each including one or more avatars having at least one set of functions, and one of the two or more avatars an instruction receiving unit for receiving a selection instruction to select from a terminal device; an avatar acquisition unit for acquiring avatar information of an avatar corresponding to the selection instruction from an avatar storage unit; and an avatar transmitting unit for transmitting.

With such a configuration, it is possible to provide an avatar capable of realizing at least one of appropriate interaction and automatic processing.

Further, in the avatar providing apparatus of the present invention, in the avatar providing apparatus, the avatar storage unit stores avatar information of one or more avatars having a dialogue function and a learning function related to dialogue, and one avatar having a learning function is stored in the avatar storage unit. may be further provided with a learning processing unit that performs learning processing related to the dialogue according to the learning function of the avatar when a predetermined condition for is satisfied.

With such a configuration, it is possible to provide an avatar capable of realizing appropriate dialogues by learning about dialogues.

Further, in the avatar providing device of the present invention, in the avatar providing device, the learning processing unit acquires all or part of teacher data or interaction models for interaction functions possessed by other avatars, and acquires the acquired teacher data or A dialog model may be used to perform learning processing related to the dialog.

With such a configuration, it is possible to provide an avatar capable of appropriately learning about dialogue and realizing appropriate dialogue.

Further, in the avatar providing device of the present invention, in the avatar providing device, the avatar storage unit stores avatar information of one or more avatars having an automatic processing function and an additional function for automatic processing, and one having the additional function. If the avatar satisfies a predetermined condition, it may further include an additional processing unit that performs additional processing related to automatic processing in accordance with the additional function of the avatar.

Such a configuration enables the avatar to achieve appropriate automatic processing.

Further, the avatar providing device of the present invention uses an input information receiving unit for receiving a request for one avatar from a terminal device, and an automatic processing function in the avatar providing device to perform automatic processing corresponding to the request. A processing unit may be further provided, and the predetermined condition may be that the request processing unit cannot perform automatic processing corresponding to the request.

With such a configuration, it is possible to add an automatic process to be executed by the avatar when the automatic process corresponding to the request cannot be executed.

Further, in the avatar providing apparatus of the present invention, in the avatar providing apparatus, an input information receiving unit for receiving dialogue input information for dialogue with one avatar from a terminal device, and a dialogue function to receive the dialogue input information further comprising a response information acquisition unit that acquires response information in response, a response information transmission unit that transmits the response information to the terminal device, and an evaluation acquisition unit that acquires an evaluation result for the response information, wherein the predetermined condition is It may be an avatar providing device in which the evaluation result is low enough to satisfy the evaluation condition.

With such a configuration, it is possible for the avatar to implement appropriate dialogue by learning about the dialogue when appropriate response information cannot be obtained for the dialogue input information.

Further, in the avatar providing device of the present invention, the instruction receiving unit in the avatar providing device receives a selection instruction associated with a user identifier corresponding to the terminal device, and the user identifier corresponds to the avatar used by the user. a user identifier storage unit that stores the user identifier associated with the selection instruction received by the instruction receiving unit; The avatar providing apparatus may further include a charging unit that charges a user identified by the user identifier corresponding to the selection instruction for use of the avatar corresponding to the selection instruction.

With such a configuration, it is possible to appropriately charge the user for the use of the avatar.

Further, in the avatar providing device of the present invention, in the avatar providing device, the avatar information stored in the avatar storage unit is associated with one or more attribute values related to the avatar output by the avatar information, and the charging unit Alternatively, the avatar providing apparatus may charge a user associated with the selection instruction with different amount of money according to one or more attribute values associated with the avatar corresponding to the selection instruction.

With this configuration, it is possible to charge an appropriate amount of money according to the attribute value of the avatar used by the user.

Further, in the avatar providing apparatus of the present invention, in the avatar providing apparatus, the user identifier storage unit further stores information indicating whether or not the avatar is temporarily used in association with the user identifier. The billing unit may be an avatar providing device that bills the user for different amounts of money depending on the value of information indicating whether or not the avatar is to be used temporarily.

With this configuration, it is possible to perform appropriate billing for temporary use and non-temporary use.

Further, in the avatar providing device of the present invention, the selection instruction has one or more user attribute values, and the avatar acquisition unit determines the avatar to be acquired using the one or more user attribute values. However, the avatar information of the avatar may be obtained from the avatar storage unit.

With such a configuration, it is possible to acquire avatar information suitable for the user.

Further, the avatar providing device of the present invention performs machine learning learning processing on two or more teacher data having one or more user attribute values and avatar identifiers for identifying avatars in the avatar providing device. further comprising a learning device storage unit that stores the acquired learning device, the avatar acquisition unit performs machine learning prediction processing using one or more user attribute values included in the selection instruction and the learning device, The avatar provision device may acquire an avatar identifier and acquire avatar information of the avatar identified by the avatar identifier from the avatar storage unit.

With such a configuration, avatar information suitable for the user can be acquired using machine learning.

According to the avatar providing device and the like according to the present invention, it is possible to provide an avatar capable of realizing at least one of appropriate dialogue and automatic processing.

1 is a block diagram of an avatar utilization system according to an embodiment of the present invention; FIG. Flowchart explaining the operation of the same avatar providing device A diagram showing an example of an avatar management table of the same avatar providing device A diagram showing a user identifier management table of the same avatar providing device (FIGS. 4(a) to 4(c)) FIG. 11 is a diagram showing a display example of an avatar by the same terminal device; FIG. A diagram showing an example of a dialogue model management table of the same avatar providing device (Fig. 6(a)) and a diagram showing an example of an automatic processing model management table (Fig. 6(b)) A diagram showing an example of a billing information management table of the same avatar providing device. A diagram showing an example of a second avatar management table of the same avatar providing device A diagram showing an example of a second user identifier management table of the same avatar providing device Fig. 10(a) showing an example of a dialogue training data management table of the same avatar providing apparatus and Fig. 10(b) showing an example of an automatic processing training data management table A diagram showing an example of the appearance of a computer system according to the same embodiment. A diagram showing an example of a configuration of a computer system according to the same embodiment.

Hereinafter, embodiments of an avatar providing device and the like will be described with reference to the drawings. It should be noted that, since components denoted by the same reference numerals in the embodiments perform similar operations, repetitive description may be omitted.

(Embodiment)
FIG. 1 is a block diagram of an avatar utilization system 1000 according to this embodiment.

The avatar utilization system 1000 includes an avatar provision device 1 and one or more terminal devices 2 . The avatar providing device 1 and each terminal device 2 are connected so as to be able to transmit and receive information via a wired or wireless network or the like. Here, as an example, a case where the avatar utilization system 1000 has three or more terminal devices 2 will be described.

The avatar providing device 1 includes an avatar storage unit 101, a user identifier storage unit 102, a learning device storage unit 103, an instruction reception unit 111, an accumulation unit 112, an avatar acquisition unit 113, an avatar transmission unit 114, an input information reception unit 115, and request processing. It includes a unit 120 , a response information acquisition unit 121 , a response information transmission unit 122 , an evaluation acquisition unit 123 , a learning processing unit 124 , an addition processing unit 125 and a charging unit 126 .

The terminal device 2 includes a terminal reception section 201 , a terminal output section 202 , a terminal reception section 203 and a terminal transmission section 204 .

The avatar providing device 1 is, for example, a server device such as an ASP server or a cloud server. The avatar providing device 1 can be implemented by, for example, a computer.

The avatar storage unit 101 stores two or more pieces of avatar information. Each avatar information is information for outputting an avatar. Avatar information is, for example, information used to display an avatar on a monitor or the like.

An avatar is an alter ego of a person represented by a computer or the like. An avatar is, for example, a virtual object representing a person. An avatar may be considered as a character or the like representing a person. The person corresponding to the avatar here may be, for example, the user of the terminal device 2, an assistant of the user of the terminal device 2, a conversation partner, a concierge, a customer of the user of the terminal device 2, or a , a business entity such as a company or corporation that provides products and services to the user of the terminal device 2 . The appearance of the avatar does not necessarily have to be that of a person, and may have the appearance of an animal, robot, or the like, for example.

Avatar information has an image of an avatar, for example. The avatar image may be a still image or a moving image such as an animation image. Also, the avatar image may be a two-dimensional image or a three-dimensional image. The avatar images included in the avatar information may be a plurality of still images in different poses. Also, the avatar image included in the avatar information may be a plurality of moving images such as animation images respectively showing different movements of the avatar. Also, the avatar information may be three-dimensional model data, vector image data, or the like that defines the outer shape of an avatar that can output an image of the avatar by performing rendering or the like. In addition, the avatar information may include 3D model data and 1 or 2 or more pieces of information specifying 1 or 2 or more movements of the 3D model. In addition, the avatar information may further include texture information, color information, and the like that are used when rendering a normal three-dimensional model. Also, the avatar image has, for example, model information. The model information is information for configuring the display of the avatar. Model information includes, for example, mesh information, bone information, and material information. The data structure of the model information is, for example, glTF (GL Transmission Format), VRM, OBJ, FBX, STL, GLB, COLLADA, etc., and the type is not limited.

Avatar information is stored in the avatar storage unit 101 in association with, for example, an avatar identifier. An avatar identifier is information that can identify an avatar, such as a code assigned to the avatar or the name of the avatar. For example, avatar information associated with an avatar identifier may be avatar information having an avatar identifier.

The avatar storage unit 101 stores two or more pieces of avatar information for outputting two or more avatars each having at least one of an interactive function and an automatic processing function. An avatar having at least one function of an interactive function and an automatic processing function may be an avatar having only an interactive function, an avatar having only an automatic processing function, or an avatar having only an automatic processing function. may be an avatar with In addition, the avatar storage unit 101 stores one or more avatars for outputting one or more avatars having at least one of a set of a dialogue function and a learning function related to dialogue, and a set of an automatic processing function and an additional function for automatic processing. Avatar information is stored. For example, one or more pieces of avatar information outputting an avatar having a dialogue function stored in the avatar storage unit 101 may further have a learning function related to dialogue, and the automatic processing function stored in the avatar storage unit 101 may be used. One or more pieces of avatar information for outputting avatars may further have an additional function of automatic processing, and avatar information for outputting avatars having interactive functions and automatic processing functions stored in avatar storage unit 101 One or more may further have learning capabilities for interaction, additional functionality for automatic processing, or both learning functionality for interaction and additional functionality for automatic processing.

A dialogue function is, for example, a function in which an avatar interacts with users of one or more terminal devices 2 . The dialogue function is, for example, a function of outputting a response according to input information for dialogue input by the user. The dialogue here is performed using, for example, one or more languages that the user can understand. The interaction here may be an interaction through the actions of the avatar. The dialogue here may be dialogue via voice, dialogue via character strings, dialogue via sign language, gestures, or the like.

The avatar's interactive function is implemented using, for example, an interactive model. A dialogue model is, for example, information that defines data to be acquired in response to a user's utterance and responses to the user's utterance.

The user's utterance is, for example, a sentence, voice, or the like indicating the user's utterance. The user's utterance is, for example, character string information or information such as voice information from which the character string can be obtained. The user's utterance is, for example, information spoken by the user to the avatar. The user's utterance is, for example, input information for dialogue (hereinafter referred to as dialogue input information) received by the input information receiving unit 115, which will be described later. For example, the dialogue input information is a character string indicating a sentence obtained by speech recognition from an utterance that is the voice uttered by the user, or a sentence that is the user's utterance input by the user via an input device such as a keyboard or a touch panel. is a string.

A dialogue model is, for example, one or more sets of sample dialogue input information and response information corresponding to the dialogue input information (hereinafter referred to as response information). Further, the dialogue model is, for example, information indicating the intention or purpose of the dialogue corresponding to the dialogue input information, or information indicating the intention or purpose of the dialogue and information defining one or more variables in the dialogue input information. and response information. Further, the dialogue model may be, for example, a learner obtained by performing machine learning using a plurality of sets of a plurality of sample dialogue input information and corresponding responses. It may be one or more sets of dialogue input information as a sample obtained from the dialogue input information and response information corresponding to the dialogue input information. The response information is, for example, information used for responding to dialogue input information, such as information such as a character string to be responded, and voice data. One or more of the dialogue input information and the response information may have information indicating variables. Further, the dialogue model may have response information that is used as a response when a response corresponding to the dialogue input information cannot be obtained or when information indicating the purpose cannot be obtained from the dialogue input information.

The interaction model used here may be a so-called generative-based model or a retrieval-based model. Also, the dialogue model is not limited to the dialogue model having the structure described above. For example, an interaction model may comprise a dialog model. Also, the dialogue model may have information indicating the sequence. Since the dialogue model is well known as a technology for dialogue systems such as avatars, detailed description thereof is omitted here. Also, the avatar's interactive function is not limited to the interactive function using the above-described dialogue model. The avatar's interactive function may be realized using information available for interaction other than the interaction model.

Response information is, for example, information that can be output by an avatar. The response information may be, for example, voice information that can be output by the avatar as voice on the terminal device 2, or character string information that can be converted into voice, and the avatar can display using a so-called balloon, dialog box, or the like. It is string information. Also, the response information may be image information.

The information that the avatar uses for interaction may be stored in the avatar storage unit 101 or may be stored in a storage unit of a server device (not shown) accessible by the avatar providing device 1 .

Information such as a plurality of different dialogue models may be prepared as the information such as the dialogue model that the avatar uses for dialogue, for example. Different information that can be used for these interactions may be stored, for example, in the avatar storage unit 101 or a storage unit (not shown). Information such as different dialogue models may be information that differs depending on the field or topic of dialogue, or where the field or topic of dialogue is the same, and the plurality of sets of dialogue input information and response information are different. It may be information.

Different identifiers or the like may be associated with information such as a plurality of interaction models that can be used by avatars for interaction. For example, a dialogue model that can be used by an avatar for dialogue may be associated with a dialogue model identifier such as a code that can identify the dialogue model.

Further, the information such as the dialogue model that can be used by the avatar for dialogue may be associated with information indicating the fields and topics of dialogue that can be realized by the information. The information indicating the field of dialogue is, for example, a field name and a field identifier such as a code assigned to the field. The information indicating the conversation topic is, for example, a topic name, a topic keyword, a topic identifier such as a code assigned to the topic, and the like.

Information such as a plurality of dialogue models that can be used by an avatar for dialogue may be associated with information indicating usage charges for using each piece of information. The usage fee may be, for example, a usage fee per time, or may be a usage fee per unit time. Also, the usage fee may be a combination of a usage fee for temporary use and a usage fee for non-temporary use different from this usage fee. The usage fee for non-temporary usage may be a usage fee for a so-called subscription, which allows unlimited usage within a predetermined period.

The interactive function of each avatar is implemented using, for example, one or more pieces of information such as the interactive model used for the above-described interactive functions associated with each avatar. Information associated with an avatar is, for example, avatar information or information associated with an avatar identifier. Avatar information may include information such as a dialogue model used for the avatar's dialogue function. Also, for example, the avatar information may include a program or the like for realizing dialogue using information such as a dialogue model used for the dialogue function.

The interactive function of one avatar used by one user is, for example, one or more pieces of information ( For example, it may be realized using one or more dialogue models). For example, the interactive function of one avatar used by one user is associated with a set of a user identifier of one user and an avatar identifier of one avatar stored in the user identifier storage unit 102, which will be described later. It may be realized using one or more pieces of information (for example, one or more dialogue models) used for the dialogue function.

A user identifier is information that can identify a user, such as a user name, a user's email address, a user's mobile phone number, a code assigned to the user, and the like. In cases such as when the terminal device 2 used by one user is fixed, the identifier of the terminal device 2 used by the user may be used as the user identifier of the user. The identifier of the terminal device 2 is, for example, the serial number of the terminal device 2, the MAC address or IP address of the terminal device 2, a code assigned to the terminal device 2, or the like. This also applies to the following.

One or more pieces of information such as a dialogue model that each avatar uses for the dialogue function may be predetermined as a default or the like. Further, one or more pieces of information such as a dialogue model used by each avatar for the dialogue function may be specified by the user using the avatar. One or more pieces of information, such as the interaction model that each avatar uses for interactivity, may be updatable. Updatable means, for example, that one or more of addition, change, and deletion is possible.

An automatic processing function is, for example, a function in which an avatar automatically performs one or more processes. For example, an avatar is a function that automatically executes processing in response to a request from a user. The request here is, for example, information input to the avatar, requesting that the user automatically perform one or more processes on the avatar. The request here is, for example, information that can detect the process to be executed. The request here may have one or more parameters required for execution of processing. The request here is information obtained in the same manner as the interaction input information described above. Note that the request here may or may not be considered to be included in the interactive input information described above. In the following, as an example, a case where the above-described interaction input information does not include a request will be described.

The one or more processes automatically executed by the avatar using the automatic processing function may be, for example, a process of searching on demand or a process of collecting information on demand. In addition, the processing that is automatically executed by the automatic processing function is, for example, the processing of performing a purchase procedure on an electronic commerce site, or calculating the cumulative purchase amount when performing a purchase procedure and transmitting it to the terminal device. It is a process to In addition, the processing automatically executed by the automatic processing function may be the processing of making a reservation for a store or service, etc., in response to a request. It may be processing such as transmission. The one or more processes executed by the automatic processing function are not limited to the processes described above. The one or more processes executed by the automatic processing function are preferably processes performed on an external server device (not shown) other than the terminal device 2, for example. In addition, it is preferable that this process be a process other than the process whose purpose is to output the information obtained by the process to the user.

Automated processing functions are implemented, for example, using an automated processing model for performing automated processing. An automatic processing model is, for example, one or more pieces of information that define automatic processing to be executed in response to requests from users. An automatic processing model is information having a request from a sample user and information defining an automatic processing to be executed in response to this request. A sample request is, for example, information having information indicating the intent or purpose of the request. A sample request may also have information defining one or more variables. The request from the user is, for example, an instruction to purchase a product or service, an instruction to make a reservation for the product or service, or an instruction to create and send an e-mail or the like. The information defining automatic processing includes information indicating automatic processing to be executed, information instructing execution of automatic processing, and the like. Information defining automatic processing may be, for example, one or more programs or identifiers of one or more programs (for example, executable file names, etc.), one or more functions, method names, "if", It may be composed of a control syntax or the like indicated by "then" or the like. The automatic process defined by the information defining the automatic process may be an automatic process that utilizes one or more variables, etc. within the corresponding request. The automatic processing model may be a learner obtained by performing machine learning using a plurality of sets of user requests and information defining automatic processing to be executed in response to the requests as teacher data. It may also be an automatic processing model having the user's request obtained by learning as described above and information defining automatic processing to be executed in response to this request. The learning device here may be, for example, a learning device that acquires information defining processing corresponding to a user's request, or a learning device that automatically executes processing corresponding to the request. There may be.

Note that the automatic processing function of the avatar is not limited to the automatic processing function executed using the above automatic processing model. Avatar automation functions may be performed using information available to the automation functions other than the automation model.

Information used by the avatar for the automatic processing function may be stored in the avatar storage unit 101, or may be stored in a storage unit of a server device (not shown) accessible by the avatar providing device 1.

For example, information such as a plurality of different automatic processing models may be prepared as the information such as the automatic processing model that can be used by the avatar to perform the automatic processing function. Different information that can be used for executing these automatic processing functions may be stored, for example, in the avatar storage unit 101 or a storage unit (not shown). Information such as an automatic processing model that can be used to execute different automatic processing functions may be information in which the processing to be executed differs, such as product purchases and reservations, or the processing to be executed may be the same. , information used for automatic processing (for example, a search site used for searching, an electronic commerce site for purchasing a product, etc.) may be different.

Different identifiers and the like may be associated with the information, such as multiple automation models that the avatar can use to perform the automatic processing functions. For example, an automatic processing model that can be used by an avatar to execute an automatic processing function may be associated with an automatic processing model identifier such as a code that can identify the automatic processing model.

In addition, the information that the avatar can use to execute the automatic processing function includes information indicating the field of automatic processing (for example, field name, identifier indicating field, etc.) and information used for automatic processing (for example, reservation site , an identifier that identifies an electronic commerce site, etc.) may be associated. The field of automatic processing includes information indicating the classification of processing content of automatic processing, information indicating the classification of processing targets to be processed by automatic processing, a combination of these, and the like. The field of automated processing may be a field in which automated processing is used. For example, it may be determined that the automatic processing is in a different field depending on whether the content of the processing performed by the automatic processing is reservation or purchase. Also, if the fields of products or services purchased by automatic processing are different, it may be determined that the automatic processing is for different fields.

For information such as multiple automatic processing models that can be used by avatars to execute automatic processing functions, similar to information such as multiple dialogue models that can be used for dialogue above, usage fees for using each information will be charged. The information shown may be associated.

The automatic processing function of each avatar is implemented using, for example, one or more pieces of information such as the automatic processing model used for the above-described automatic processing function associated with each avatar. The information associated with the avatar is, for example, avatar information corresponding to the avatar or information associated with the avatar identifier. Information such as the automatic processing model used for the automatic processing function of the avatar may be information included in the avatar information.

The automatic processing function of one avatar used by one user is, for example, one or more automatic processing functions associated with a set of a user identifier of one user and an avatar identifier of one avatar. It may be implemented using information (eg, one or more automated processing models) or the like. For example, the automatic processing function of one avatar used by one user is associated with a set of a user identifier of one user and an avatar identifier of one avatar stored in the user identifier storage unit 102 described later. It may also be realized using one or more pieces of information (for example, one or more pieces of automatic processing model) used for the automatic processing function.

One or more pieces of information such as an automatic processing model that each avatar uses for the automatic processing function may be predetermined as a default or the like. Also, one or more pieces of information such as an automatic processing model that each avatar uses for the automatic processing function may be specified by the user using the avatar. One or more pieces of information, such as the automation model that each avatar uses for its automation functions, may be updatable.

The learning function related to dialogue is, for example, learning of information used in dialogue. The learning of information used in dialogue is, for example, updating of information such as a dialogue model used in dialogue. Here, updating means addition, change, deletion, or the like of information. The update of the dialogue model etc. may be, for example, the addition of a learner used for the dialogue function.

Also, the learning of information used in dialogue may be the updating of teacher data used when obtaining a dialogue model used in dialogue by machine learning or the like, and machine learning is performed using this teacher data. It is also possible to add a learner or dialogue model acquired by the method, or to replace an existing learner or dialogue model with these learners or dialogue models.

An additional function of the automatic processing function is the learning of information used for the automatic processing. The learning of information used for automatic processing is, for example, updating of information such as an automatic processing model used for automatic processing. Here, updating means addition, change, deletion, or the like of information. It may also be an update of information used for processing used in automatic processing (for example, an identifier identifying a search site or an electronic commerce site).

Further, the learning of information used for automatic processing may be the updating of teacher data used when acquiring an automatic processing model used for automatic processing by machine learning or the like. It may be addition of a learning device or an automatic processing model acquired through learning, or replacement of an existing learning device or automatic processing model with these learning devices or automatic processing model.

The avatar information may be associated with a plurality of teacher data as described above, which are used when acquiring information such as a dialogue model used for executing the dialogue function by machine learning. A plurality of teacher data may be stored in the avatar storage unit 101, for example.

In addition, the avatar information may be associated with a plurality of teacher data as described above, which are used when acquiring information such as an automatic processing model used for executing the automatic processing function by machine learning. . A plurality of teacher data may be stored in the avatar storage unit 101, for example.

The different avatar information stored in the avatar storage unit 101 is normally information for outputting different avatars. Different avatars are, for example, avatars with different appearances and movements. For example, different avatar information may have different avatar images, or may have different three-dimensional model data capable of outputting different avatar images. Also, different avatars or avatars capable of outputting different sounds may be used. Also, the different avatars may be avatars with different interactive functions or avatars with different automatic processing functions. Avatars having different interactive functions are, for example, avatars with different information such as dialogue models used for the interactive functions. A different dialogue model is a dialogue model that contains different information (for example, multiple sets of dialogue input information and responses, or multiple sets of information indicating the purpose of dialogue and response information). It is also possible to use dialogue models with different target fields and topics, or with different teacher data used to acquire the dialogue models. Avatars having different automatic processing functions are, for example, avatars with different information such as an automatic processing model used for executing the automatic processing functions. Different automatic processing models may be automatic processing models in which automatic processing to be executed differs, or automatic processing models in which information such as search sites used for automatic processing differs.

The avatar information may include, for example, a program that causes a computer or the like to perform one or more functions as described above, and may be associated with such a program. The programs here may be considered to include so-called macros and the like. It is associated with data (for example, dialogue model, automatic processing model, etc.). The avatar information is, for example, a program for executing one or more interactive functions associated with the avatar indicated by the avatar information, a program for executing one or more automatic processing functions, or an identifier that identifies these programs. etc. This program, for example, uses information used to execute interactive functions and automatic processing functions, such as interactive models and automatic processing models associated with avatar information, to perform one or more functions as described above. is a program that executes

The avatar storage unit 101 is preferably a non-volatile recording medium, but can also be realized with a volatile recording medium. The same applies to other storage units described below.

The user identifier storage unit 102 stores user identifiers in association with avatars used by users identified by the user identifiers. Storing in association with an avatar means, for example, storing in association with an avatar identifier. The user identifier may be stored in association with date and time information such as accumulated date and time.

The user identifier storage unit 102 may further store information indicating whether or not the avatar is temporarily used in association with the user identifier. For example, the user identifier storage unit 102 may further store information indicating whether or not the avatar is temporarily used in association with the user identifier and the avatar. Information indicating whether or not to temporarily use is, for example, so-called flag information.

User identifier storage section 102 may store a user identifier and an avatar having a learning function in association with information for executing a dialogue function added by the learning function. The information for executing the dialogue function added by the learning function is, for example, information such as a dialogue model that the avatar added by the learning function uses for the dialogue function. The fact that the information used by the avatar for the interactive function is associated and stored means that, for example, the avatar identifier is associated with the identifier of the information used for the interactive function (for example, the identifier of one or more interaction models). It may be stored. Further, the information for executing the dialogue function added by the learning function may be, for example, teacher data used for acquiring information such as a dialogue model that the avatar uses for the dialogue function by machine learning. . The fact that the avatar is associated with the teacher data and stored means, for example, that the avatar identifier is associated with the identifier of the teacher data and stored.

Further, the user identifier storage unit 102 may store a user identifier and an avatar having an additional function of the automatic processing function in association with information for executing the dialogue function added by the additional function. . The information used for executing the dialogue function added by the additional function is, for example, information such as an automatic processing model that the avatar added by the additional function uses for the automatic processing function. The fact that the information used by the avatar for the automatic processing function is associated and stored means that, for example, the avatar identifier corresponds to the identifier of the information used for the automatic processing function (for example, the identifier of one or more automatic processing models). It may be attached and stored. In addition, the information used to execute the automatic processing function added by the learning function is, for example, the teacher used to acquire information such as the automatic processing model that the avatar uses for the automatic processing function by machine learning. It may be data. The fact that the avatar is associated with the teacher data and stored means, for example, that the avatar identifier is associated with the identifier of the teacher data and stored.

The learning device storage unit 103 stores learning devices obtained by performing machine learning learning processing on two or more teacher data having one or more user attribute values and avatar identifiers for identifying avatars. be. A user's attribute value is a value indicating a user's age, sex, preference, and the like. The preferences here are, for example, information about the user's favorite characters, animals, and the like. The learner here is, for example, a learner used to acquire an avatar that matches the user.

For example, the learning device stored in the learning device storage unit 103 uses, as training data, a set of one or more user attribute values and an avatar identifier of an avatar used by each user, which are acquired for each of a plurality of users. It is a learner generated by performing learning. As machine learning, for example, neural network learning, SVM, SVR, deep learning, decision tree, random forest, etc. can be used.

Note that storing a learning device in the learning device storage unit 103 means, for example, that the learning device itself (for example, a function that outputs a value in response to an input, a learning result model, etc.) is stored. Alternatively, information such as parameters necessary for constructing the learning device may be stored. This is because, even in the latter case, the learning device can be configured using information such as the parameters, so it can be considered that the learning device is substantially stored in a storage unit (not shown).

The instruction receiving unit 111 receives from the terminal device 2 a selection instruction for selecting one of the two or more avatars stored in the avatar storage unit 101 . The instruction receiving unit 111 receives, for example, a selection instruction associated with a user identifier corresponding to the terminal device 2 (for example, a user identifier of a user using the terminal device 2). The selection instruction may be, for example, an instruction that allows one avatar to be selected as a result. For example, the selection instruction may have an avatar identifier of one avatar selected by the user corresponding to the terminal device 2 . Also, the selection instruction may have one or more user attribute values for the user of the terminal device 2 . The one or more user attribute values are user attribute values that can be used for selecting an avatar, and are, for example, information that can be given as input to a learning device stored in the learning device storage unit 103 . The one or more user attribute values here are, for example, one or more user attribute values having the same attribute as one or more user attribute values possessed by the teacher data used to acquire the learning device stored in the learning device storage unit 103. be. The avatar used by the user is determined. For example, the user identifier and the avatar identifier of the avatar used by the user indicated by the user identifier are stored in the user identifier storage unit 102 or the like in association with each other. In that case, the selection indication may comprise at least the user identifier.

Also, the selection instruction may have information indicating whether to temporarily use the avatar.

The instruction receiving unit 111 is usually realized by wireless or wired communication means.

For example, the accumulation unit 112 associates the user identifier associated with the selection instruction received by the instruction receiving unit 111 with the avatar corresponding to the selection instruction and accumulates the user identifier in the user identifier storage unit 102 . The avatar corresponding to the selection instruction is an avatar acquired by the avatar acquisition unit 113, which will be described later, in accordance with the selection instruction. Storing the user identifier in association with the avatar means, for example, accumulating the user identifier in association with the avatar identifier.

Avatar acquisition section 113 acquires avatar information of the avatar corresponding to the selection instruction from avatar storage section 101 . For example, the avatar acquisition unit 113 may acquire avatar information associated with the avatar identifier included in the selection instruction from the avatar storage unit 101 . Avatar acquisition section 113 may acquire from avatar storage section 101 avatar information associated with the avatar identifier stored in user identifier storage section 102 in association with the user identifier included in the selection instruction. Avatar acquisition unit 113 obtains avatar information associated with the avatar identifier associated with the user identifier included in the selection instruction and stored in user identifier storage unit 102 in association with information that is not temporarily used. It may be acquired from the avatar storage unit 101 .

Also, the avatar acquisition unit 113 may determine an avatar to acquire using one or more user attribute values included in the selection instruction, and acquire avatar information of the avatar from the avatar storage unit 101 . For example, the avatar acquisition unit 113 performs machine learning prediction processing using one or more user attribute values included in the selection instruction and the learner stored in the learner storage unit 103, acquires an avatar identifier, Avatar information of an avatar identified by an avatar identifier may be obtained from the avatar storage unit 101 . For example, the avatar acquisition unit 113 may acquire avatar information identified by an avatar identifier obtained by inputting user attribute values into a learning device.

Further, the avatar storage unit 101 stores a plurality of pieces of avatar information in association with one or more combinations of user attribute values. may be acquired from the avatar storage unit 101, the avatar information associated with the same attribute value as one or a combination of two or more of

Note that the learning device storage unit 103 may be omitted when the learning device is not used to acquire avatar information.

The avatar transmission unit 114 transmits the avatar information acquired by the avatar acquisition unit 113 to the terminal device 2 that transmitted the selection instruction. Avatar transmission unit 114 is usually realized by wireless or wired communication means.

The input information receiving unit 115 receives information transmitted from the terminal device 2 . The input information receiving unit 115 receives, for example, dialogue input information for dialogue with one avatar from the terminal device 2 . For example, the input information receiving unit 115 receives interactive input information associated with the user identifier of the user who uses the terminal device 2 . The dialogue input information received by the input information receiving unit 115 may be associated with the avatar identifier of the avatar to be dialogued. However, input information receiving section 115 may receive interactive input information only for one avatar having an interactive function.

Also, the input information receiving unit 115 receives, for example, a request for one avatar from the terminal device 2 . For example, the input information receiving unit 115 receives a request associated with the user identifier of the user using the terminal device 2 . The request received by the input information receiving unit 115 may be associated with the avatar identifier of the requested avatar. However, the input information receiving unit 115 may receive requests only for avatars having an automatic processing function.

The request processing unit 120 and the response information acquiring unit 121 may determine whether the information received by the input information receiving unit 115 is interactive input information or a request. For example, interactive input information input by a user and a request may be information that can be determined as respective information. For example, only the request includes one or more specific phrases that can be determined to be a request (for example, phrases such as "purchase", "reserve", etc.), and if this phrase is included, The input information receiving unit 115 may determine that the input information is a request, and may determine that it is corresponding input information if it does not contain the request. Also, at the beginning of the correspondence input information, a rule is set that a specific word of appeal (for example, words such as "question", "tell me", etc.) is required, and information in accordance with this rule is used as correspondence input information. Further, when inputting dialogue input information, a specific operation or input is required for the terminal device 2, and when there is a specific operation or input, the terminal device 2 Information such as a flag indicating that it is interactive input information may be transmitted in association with the interactive input information, and it may be determined that the input information is interactive input information when there is information such as this flag. When inputting a request, a specific operation or input is required in the same manner as described above, and information such as a flag indicating a request is sent to the terminal device 2 in response to this operation or input. may be transmitted in association with the request.

Further, the input information receiving unit 115 performs language processing on the received information to determine whether or not execution of some processing is requested. If it is not a request, it may be determined as interaction input information. Machine learning may also be used to determine whether received information is interaction input information or a request. For example, a learning device acquired by learning sentences that are dialog input information and sentences that are requests as teacher data is stored in advance in a storage unit (not shown), and this learning device is used. may be used to determine whether the received information is interactive input information or a request.

It should be noted that instead of input information receiving section 115 determining whether the received information is interactive input information or request information, request processing section 120 determines the information received by input information receiving section 115 as described above. , and the response information acquisition unit 121 performs the same processing on the information received by the input information receiving unit 115 to determine whether it is interactive input information. You may make it

Information such as requests and dialogue input information received by the input information receiving unit 115 may be character string information or voice information. In the case of voice information, the input information receiving unit 115 preferably acquires character string information corresponding to the voice information. For example, the input information receiving unit 115 preferably acquires character string information corresponding to the voice information by performing voice recognition processing or the like on the voice information. The information used for speech recognition processing may be held by the avatar provision device 1 or may be held by another server device (not shown) or the like accessible by the avatar provision device 1 . In addition, the input information receiving unit 115 transmits the received voice information to a server device (not shown) or the like that performs voice recognition or the like to acquire information of a character string corresponding to the voice information, and then transmits the character string. Information on the character string corresponding to the voice information may be obtained by executing processing for obtaining the string and receiving the character string obtained thereby.

The request processing unit 120 uses the automatic processing function to execute automatic processing corresponding to the request. For example, when the input information receiving unit 115 receives a request for one avatar (for example, a request associated with an avatar identifier), the request processing unit 120 receives this one avatar (for example, a request associated with the request). avatar (identified by the avatar identifier) to perform automated processing in response to the request. For example, when a request for one avatar is received, the request processing unit 120 automatically responds to the received request using an automatic processing model associated with this avatar and stored in the avatar storage unit 101 or the like. Acquire information that defines a process and perform an automated process defined by the acquired information. The request processing unit 120 uses, for example, the automatic processing model stored in the user identifier storage unit 102 in correspondence with the user identifier of the user of the terminal device 2 that transmitted the request and the avatar associated with the request. , may obtain information that defines automated processing in response to the received request. For example, the request processing unit 120 retrieves information defining automatic processing associated with a request corresponding to the received request in an automatic processing model associated with the avatar. For example, if the information defining the automatic process to be acquired is the executable file name of the program for executing the automatic process, the request processing unit 120 executes the executable file indicated by this executable file name to respond to the request. Perform automatic processing accordingly. If the automatic process to be executed is an automatic process that acquires information indicating that the process has ended, a processing result, or the like, the request processing unit 120 may acquire such information. The request processing unit 120 may use one or more variables included in the request to perform automatic processing corresponding to the request. For example, automatic processing including search processing using one or more keywords included in the request may be executed as the automatic processing corresponding to the request. Note that the request processing unit 120 may execute automatic processing corresponding to a request without using the automatic processing model.

The response information acquisition unit 121 acquires response information according to interactive input information using the interactive function. For example, when the input information receiving unit 115 receives dialogue input information for one avatar (for example, dialogue input information associated with an avatar identifier), the response information acquiring unit 121 obtains this one avatar (for example, Acquires response information corresponding to the interactive input information using the interactive function of the avatar identified by the avatar identifier associated with the interactive input information. For example, when dialogue input information for one avatar is received, the response information acquisition unit 121 uses the dialogue model associated with this avatar and stored in the avatar storage unit 101 or the like to obtain the received dialogue input information. Get the response information corresponding to .

For example, the response information acquisition unit 121 stores in the user identifier storage unit 102 the user identifier of the user of the terminal device 2 who transmitted the dialogue input information and the avatar associated with this dialogue input information in association with the user identifier. Acquire response information corresponding to the received dialogue input information using the dialogue model. For example, the response information acquisition unit 121 acquires response information associated with the dialogue input information corresponding to the received dialogue input information in the dialogue model associated with the avatar. The response information acquisition unit 121 may acquire response information using one or more variables included in the dialogue input information. For example, response information including variables may be acquired by replacing a character string or the like indicating a predetermined variable in response information corresponding to the dialog input information with one or more variables included in the dialog input information. . Note that the response information acquisition unit 121 may acquire response information corresponding to dialogue input information without using the dialogue model.

The response information transmission unit 122 transmits the response information acquired by the response information acquisition unit 121 to the terminal device 2 . The response information transmission unit 122 transmits, for example, the response information acquired by the response information acquisition unit 121 according to the dialogue input information to the terminal device 2 that transmitted the dialogue input information. The response information transmission unit 122 is implemented by, for example, wireless or wired communication means.

The evaluation acquisition unit 123 acquires evaluation results for the response information acquired by the response information acquisition unit 121 . The evaluation result is, for example, information indicating an evaluation as to whether or not appropriate response information has been obtained with respect to the dialogue input information, or information as to whether or not response information has been obtained with respect to the dialogue input information. Evaluation information indicating that appropriate response information could not be acquired is, for example, evaluation information indicating that the response corresponding to the dialogue input information is broken. The evaluation result is, for example, a binary value representing whether or not the response information is highly evaluated, or a value representing the evaluation in multiple stages of 3 or more. For example, the evaluation acquisition unit 123 acquires evaluation information indicating a high evaluation when the response information acquisition unit 121 acquires response information for the dialogue input information. Evaluation information indicating that the evaluation is low may be obtained when response information indicating that appropriate response information corresponding to the input information could not be obtained is obtained. Also, the evaluation acquisition unit 123 may acquire evaluation information according to the number of pieces of response information acquired for the dialogue input information. For example, the evaluation acquisition unit 123 may acquire evaluation information in which the evaluation decreases continuously or in stages as the number of response information acquired in response to dialogue input information increases.

Also, the evaluation acquisition unit 123 may acquire evaluation information according to dialogue input information received by the input information reception unit 115 in response to response information transmitted by the response information transmission unit 122 . For example, when the dialogue input information received for one piece of response information transmitted by the response information transmission unit 122 is information indicating that the one piece of response information was not appropriate response information, the evaluation acquisition unit 123 obtains an evaluation You may acquire the evaluation information which shows low.

Note that the evaluation acquisition unit 123 may acquire evaluation information using questionnaire response information indicating an evaluation of response information received from the terminal device 2 by a receiving unit (not shown). For example, the questionnaire answer information corresponding to the response information is, for example, information transmitted by the response information transmission unit 122 together with the response information when the response information is transmitted, and is information asking the evaluation of the response information to be transmitted. This is information on answers to questionnaire information. The questionnaire response information may be information indicating whether or not the response information corresponding to the questionnaire information is appropriate, or may be information obtained by evaluating the response information in three or more stages. The questionnaire information is questionnaire information for obtaining such answers. The evaluation acquisition unit 123 may acquire the evaluation possessed by one questionnaire answer information as the evaluation information, and obtain the values obtained by statistically processing the evaluations indicated by the plurality of questionnaire answer information corresponding to the same dialogue input information as the evaluation information. may be obtained.

The evaluation acquisition unit 123 stores the acquired evaluation information in a storage unit (not shown), the avatar storage unit 101, or the like in association with the corresponding dialogue input information or a sample of the dialogue input information, for example. It may be transmitted to one or more specified destinations.

When a predetermined condition is satisfied for one avatar having a learning function, the learning processing unit 124 performs a learning process related to dialogue according to the learning function of this avatar. For example, the learning processing unit 124 is one avatar having a learning function, and when a predetermined condition is satisfied for an avatar used by one user, the learning processing unit 124 conducts a dialogue according to the learning function of this avatar. Perform learning processing for An avatar used by one user is, for example, an avatar with which one user is interacting. Performing the learning process according to the learning function may be considered as executing the learning function and performing the learning process, for example.

The predetermined condition is, for example, that the evaluation result obtained by the evaluation obtaining unit 123 for one avatar is so low as to satisfy the evaluation condition. The evaluation condition is, for example, a condition for evaluating whether the evaluation indicated by the evaluation result is high or low. For example, if the evaluation result is information that expresses the evaluation numerically, the evaluation condition may be, for example, the condition that the evaluation result is lower than the threshold (the higher the evaluation result is, the higher the evaluation is) or the condition that the evaluation result is higher than the threshold (evaluation result The lower the numerical value, the higher the evaluation). An evaluation condition may be considered as a condition that any one of the values of one or more evaluation results indicating a low evaluation is matched. It should be noted that the evaluation conditions are not limited to the above information as long as it is information that allows determination of whether or not the evaluation is high based on these evaluation result values.

The learning process here may be to add, delete, or update information such as a dialogue model that one avatar uses for dialogue, and one avatar uses for dialogue with one user. It may be to add, delete or change information such as a dialogue model. For example, the interaction model stored in the user identifier storage unit 102 in association with the avatar identifier and the user identifier may be added, deleted, or changed. Also, the learning process here may be acquisition of all or part of teacher data for interactive functions possessed by other avatars. In addition, the learning process here may be to add, delete, change, or the like a dialogue model or the like used for the dialogue function by using machine learning or the like.

For example, when a predetermined condition is satisfied, the learning processing unit 124 uses dialogue input information as a determination target for determining whether or not the condition is satisfied, and information used by one avatar to realize the dialogue function. Learning processing may be performed using a dialogue model or the like. For example, the learning processing unit 124 determines that the evaluation of the acquired response information for the dialogue input information to be judged is low, and uses part or all of this dialogue input information to respond to this dialogue input information. Learning may be performed so as to obtain highly evaluated response information as response information. For example, the learning processing unit 124 determines that the evaluation of the acquired response information for the dialogue input information to be judged is low, and uses one or more dialogue models used for the dialogue function to determine the one or more Learning may be performed to obtain a dialogue model other than the dialogue model, which provides highly evaluated response information.

The learning processing unit 124 acquires, for example, all or part of teacher data for dialog functions possessed by other avatars other than one avatar, and uses the acquired teacher data to perform learning related to dialogue for the one avatar. process. The teacher data for the interactive function is, for example, teacher data used for acquiring or updating the dialogue model used for the interactive function. The teacher data here is, for example, a plurality of data used as teacher data. The learning process here is machine learning using acquired teacher data and the like.

For example, the learning processing unit 124 uses teacher data stored in the avatar storage unit 101 or the like to be used by other avatars for the dialog function, and the dialogue corresponding to the response information determined to satisfy a predetermined condition. Acquire all or part of the teacher data related to the input information, perform machine learning using the acquired teacher data to acquire a new dialogue model, or acquire the acquired teacher data and the already acquired teacher data and may be used to perform machine learning to update the dialogue model used for dialogue. Teacher data related to dialogue input information is, for example, data having dialogue input information of the same field or topic as the dialogue input information corresponding to the response information determined to satisfy the conditions, and response information therefor. For example, dialogue input information of the same field, topic, purpose, etc. can be detected using a known topic extraction technique or the like. The teacher data related to dialogue input information is obtained from the dialogue input information corresponding to the response information for which the attribute value indicating the dialogue field, topic, etc. associated with the teacher data satisfies the conditions. It may be teacher data that matches or is similar to the topic or the like. In this case, for example, such attribute values are preferably associated with the teacher data and stored in the avatar storage unit 101 or the like.

The learning processing unit 124 acquires, for example, all or part of one or more dialogue models for dialogue functions possessed by other avatars other than one avatar, and uses the acquired dialogue model to obtain the one avatar performs learning processing related to the dialogue. Acquisition of the dialogue model here may be acquisition of an identifier of the dialogue model, a link to the dialogue model, or the like. A dialogue model for the dialogue function is a dialogue model used for dialogue by an avatar having the dialogue function. The part of the dialogue model may be one or more of two or more dialogue models, and part of the data constituting one dialogue model (for example, samples of dialogue input information possessed by the dialogue model). one or more of a plurality of sets with response information). The learning process here is to add a dialogue model that one avatar uses for dialogue or a dialogue model that one avatar uses for dialogue with one user. The addition of the dialogue model here is performed by associating the obtained dialogue model with one avatar or a pair of one avatar and one user as a newly added dialogue model, or by adding in correspondence. Alternatively, it may be merging the acquired dialogue model with the dialogue model already used by one avatar.

For example, the learning processing unit 124 uses dialogue models stored in the avatar storage unit 101 or the like, which are used for dialogue functions by other avatars, for dialogue corresponding to response information determined to satisfy a predetermined condition. All or part of a dialogue model related to input information is acquired, and the acquired dialogue model is added as a dialogue model used by one avatar. A dialogue model related to the dialogue input information is, for example, a dialogue model for the same field or topic as the dialogue input information corresponding to the response information determined to satisfy the conditions. It is a dialogue model in which the topic is the same as the dialogue input information corresponding to the response information judged to satisfy the conditions. For example, dialogue input information on the same field or topic can be detected using a known topic extraction technique or the like. The dialog model related to the dialog input information is obtained from the dialog input information corresponding to the response information for which the attribute value indicating the field or topic of the dialog associated with the dialog model satisfies the conditions. It may be a dialogue model or the like that matches or is similar to the topic or the like. In this case, for example, it is preferable to store such attribute values in the avatar storage unit 101 or the like in association with the dialogue model or the like.

The additional processing unit 125 performs additional processing related to automatic processing according to the additional function of the avatar when a predetermined condition is satisfied for one avatar having the additional function. Predetermined conditions are, for example, conditions relating to automatic processing that is executed in response to a request. The predetermined condition is, for example, that the request processing unit 120 cannot execute automatic processing corresponding to the request received by the input information receiving unit 115 . The inability to perform automatic processing may be considered to be the inability to find an automatic processing model with information defining the automatic processing corresponding to the request. Also, the predetermined condition may be that the user's evaluation result for the result of the automatic processing is low enough to satisfy the evaluation condition, as in the case of the evaluation result for the dialogue. The evaluation result is, for example, information indicating the result of the automatic processing and questionnaire information for the automatic processing similar to the above is sent to the terminal device 2 that sent the request, and is sent from the terminal device 2 in response to this. You may make it acquire from the questionnaire result information. Performing additional processing related to automatic processing according to the additional function of the avatar means, for example, performing additional processing of automatic processing by executing the additional function.

Additional processing related to automatic processing performed by the additional processing unit 125 is, for example, adding automatic processing that can be executed by one avatar. Automatic processing that can be executed by one avatar may be considered automatic processing that can be executed by one avatar for one user, for example. This also applies to the following. Further, the additional processing related to automatic processing performed by the additional processing unit 125 is, for example, adding information such as an automatic processing model used when one avatar executes automatic processing. Adding an automatic processing model may be, for example, adding a new automatic processing model in addition to the automatic processing model already used by the avatar. and the data of the new automatic processing model so as not to cause duplication. The additional processing unit 125 may add information such as one or more automatic processing models that other avatars use for automatic processing, and may add information such as one or more automatic processing models stored in a storage unit (not shown). Information such as an automatic processing model may be added. If the predetermined condition is that the automatic processing corresponding to the request cannot be executed, the additional processing unit 125 detects and adds an automatic processing model that enables execution of the automatic processing corresponding to the request, for example. do. For example, the additional processing unit 125 converts the request received by the input information receiving unit 115 into a sample request having the same intention and purpose, or a sample having the same intention and purpose and having the same variables as the request. Add an automatic processing model with the following requirements: Also, information such as an automatic processing model associated with the same or similar fields as the received request may be added. Techniques for acquiring information such as the field of a request from a request or a sample request of an automatic processing model are well known, so detailed description thereof will be omitted. Further, in this case, for example, as in the case of the dialogue model described above, the information such as the automatic processing model is associated with the information such as the field in advance so that the additional processing unit 125 receives the input information receiving unit 115 may detect and add automated processing models associated with fields, etc. that are the same or similar to the fields, etc. obtained from the received request.

The billing unit 126 bills the user identified by the user identifier corresponding to the selection instruction for use of the avatar corresponding to the selection instruction. The billing process uses information that can be used for payment associated with the user, such as user credit card information (for example, credit card number) and electronic money information (for example, account). It may be a process of collecting a fee for the use of , and an instruction to perform such charging is output to a credit card company, a company that performs electronic money settlement, a server device operated by these companies, etc. It may be a process of issuing a bill for the usage of the avatar to the user. The billing process is not limited to the process described above, as long as the user can be billed for the use of the avatar. Information on the user's credit card or electronic money required for billing may be stored, for example, in a storage unit (not shown) in association with the user identifier.

The billing unit 126 performs billing processing of different amounts for the user associated with the selection instruction, for example, according to one or more attribute values associated with the avatar corresponding to the selection instruction. Avatar attribute values include, for example, the functions that can be used by the avatar, the number of functions that can be used by the avatar, the dialogue model that the avatar uses to execute the function, the number of dialogue models that are used, and the data volume of the dialogue model that is used. , the automatic processing model used by the avatar to execute the function, the number of automatic processing models to be used, the amount of data of the automatic processing model to be used, the amount of teacher data used by the avatar to acquire the dialogue model, etc. The amount of data here may be the number of data, or the capacity of data (for example, the number of bits, etc.).

A different amount of money is, for example, an amount that varies according to the attribute value (in other words, is determined according to the attribute value). For example, the billing unit 126 provides an amount of money that increases in proportion to the number of functions used by the avatar, the number of dialogue models, or the number of automatic processing models, which are attribute values of the avatar (or correspondingly increases continuously or stepwise). (amount to be paid) is charged. Also, for example, the billing unit 126 may increase the amount of money (or continuously or stepwise billing process for the amount that increases to Also, the billing unit 126 may perform a process of billing an amount associated with, for example, a dialog model used by an avatar, an automatic processing model, or teacher data.

The charging unit 126 may charge the user with different amounts of money depending on the value of the information indicating whether or not the avatar is to be used temporarily. For example, when receiving a selection instruction or the like, the instruction receiving unit 111 receives information indicating whether or not the avatar indicated by the selection instruction is to be used temporarily. The unit 126 may determine whether the use of the avatar by the user is temporary use. For example, as described above, information indicating whether to temporarily use (for example, flag information) is stored in the user identifier storage unit 102 in association with the user identifier and the avatar identifier included in the selection instruction. In accordance with the stored information indicating whether or not to use temporarily, the billing unit 126 determines whether the use of the one avatar to be billed by the one user is temporary use. You can judge whether In this case, the information indicating whether the user temporarily uses the avatar stored in the user identifier storage unit 102 may be the information received when receiving the selection instruction as described above. It may be information accumulated at the time of contract for the user to use the avatar. In addition, how the avatar providing apparatus 1 manages whether or not one user temporarily uses one avatar is not limited to the above.

For example, the billing unit 126 sets the unit price associated in advance with the attribute value of the avatar used when determining the billing amount to different amounts depending on whether the avatar is used temporarily or not. By using the avatar, different amounts may be charged depending on whether the avatar is used temporarily or not. Depending on whether it is temporary use or not, this amount is corrected by adding or multiplying a predetermined value, so that it differs depending on whether it is temporary use or not. You may make it charge the amount of money. However, how the different amounts are determined is not limited to the above.

For example, when the instruction receiving unit 111 receives a selection instruction, the billing unit 126 may perform billing processing for use of the avatar corresponding to the selection instruction to the user corresponding to the selection instruction. . The user corresponding to the selection instruction is, for example, the user of the terminal device 2 that transmitted the selection instruction. Also, when the user has finished using the avatar, the billing unit 126 may perform billing processing for the use of the avatar. Billing unit 126 uses information such as user identifiers stored in user identifier storage unit 102 in association with avatar identifiers at a predetermined time (for example, the end of each month or the 20th of each month). The billing process or the like may be performed on the same date as the date when the information was accumulated, for example, every year. Note that the timing at which the charging unit 126 performs the charging process is not limited to the above.

The terminal device 2 is, for example, a device used by a user. The terminal device 2 can be realized by, for example, a computer, a so-called smart phone, a mobile phone, a tablet terminal, a wearable terminal, or the like.

The terminal receiving unit 201 receives information transmitted from the avatar providing device 1 . For example, the terminal receiving unit 201 receives avatar information transmitted from the avatar providing device 1 . Also, for example, the terminal receiving unit 201 receives response information transmitted from the avatar providing device 1 . Also, for example, the terminal receiving unit 201 receives information indicating the result of automatic processing transmitted from the avatar providing device 1 . Also, for example, the terminal receiving unit 201 receives information indicating that automatic processing corresponding to the request transmitted from the avatar providing device 1 cannot be executed. Note that the information indicating that the automatic processing corresponding to the request cannot be executed may or may not be considered as one aspect of the information indicating the result of the automatic processing. The response information received by the terminal reception unit 201 and the information indicating the result of automatic processing are, for example, information associated with the avatar identifier. The information received by the terminal reception unit 201 is not limited to the above information. For example, along with response information and information indicating the results of automatic processing, questionnaire information about these may be received.
The terminal reception unit 201 is implemented by, for example, wireless or wired communication means.

The terminal output section 202 outputs using the information received by the terminal reception section 201 . For example, the terminal output unit 202 uses the avatar information received by the terminal reception unit 201 to output an avatar. For example, when the received avatar information is three-dimensional data, vector data, or the like, the terminal output unit 202 may display images such as still images and moving images obtained by rendering these data. Also, the terminal output unit 202 outputs, for example, response information received by the terminal reception unit 201, information indicating the result of automatic processing, and the like. Also, the terminal output unit 202 may output questionnaire information received by the terminal reception unit 201 . For example, the terminal output unit 202 may output the response information received by the terminal reception unit 201, the information indicating the result of the automatic processing, etc. as the output of the avatar. Outputting as the avatar's output means outputting information that looks like the avatar is outputting it. It may be displayed using, for example. For example, when the response information received by the terminal receiving unit 201, the information indicating the result of automatic processing, or the like is audio information, the audio using this audio information may be output as the output of the avatar. If the received response information, information indicating the result of automatic processing, or the like is character string information, the character string information may be converted into voice, and the converted voice may be output as the output of the avatar. In addition, when the response information received by the terminal receiving unit 201, the information indicating the result of automatic processing, or the like is character string information, a balloon containing the character string or a dialog box can be used as an output of the avatar, such as an image of the avatar. may be appropriately associated with and output.

Note that the output from terminal output section 202 is not limited to output using information received by terminal reception section 201 . For example, the terminal output unit 202 uses information of an operation screen for receiving an operation for outputting an avatar selection instruction stored in advance in a storage unit (not shown) to display such an operation screen. may be displayed. Also, the output from the terminal output unit 202 is not limited to the output as described above.

The output here is usually display using a display device such as a monitor or a projector, but may be audio output or a combination of display and audio output. The terminal output unit 202 can be realized by, for example, a display device driver or a combination of the display device and its driver.

The terminal reception unit 203 receives operations and inputs from the user. For example, the terminal reception unit 203 receives an operation of selecting an avatar from the user. The operation of selecting an avatar here may be considered to include acceptance of user attribute values used for selecting an avatar. As the user attribute value, a user attribute value stored in advance in a storage unit (not shown) may be read. Also, the terminal reception unit 203 may receive, for example, an input of a request for an avatar from the user. Also, the terminal reception unit 203 receives, for example, input of interactive input information for the avatar from the user. The avatar here is, for example, the avatar output by the terminal output unit 202 . For example, the terminal reception unit 203 receives a character string input from the user via an input device (not shown) such as a keyboard or touch panel. Further, for example, the terminal reception unit 203 may receive input by voice through an input device (not shown) such as a microphone. Inputs accepted by the terminal accepting unit 203 are not limited to the above inputs. For example, the terminal reception unit 203 may receive from the user questionnaire response information for the questionnaire information output by the terminal output unit 202 .
The terminal reception unit 203 can be realized by, for example, an input device driver or a combination of an input device and its driver.

The terminal transmission unit 204 transmits information according to the input received by the terminal reception unit 203 to the avatar providing device 1 . For example, the terminal transmission unit 204 may associate the information according to the input with the user identifier of the user of the device itself, which is stored in advance in the storage unit (not shown), and transmit the information to the avatar providing device 1 . . Further, the terminal transmission unit 204 may transmit, for example, information corresponding to the input received by the terminal reception unit 203 regarding one avatar to the avatar providing device 1 in association with the avatar identifier of this one avatar.

For example, the terminal transmission unit 204 transmits to the avatar providing device 1 a selection instruction corresponding to the operation of selecting an avatar received by the terminal reception unit 203 . This selection instruction may be information having user attribute values. Also, the terminal transmission unit 204 transmits, for example, the request received by the terminal reception unit 203 to the avatar providing device 1 . Also, the terminal transmission unit 204 transmits, for example, the dialogue input information received by the terminal reception unit 203 to the avatar providing device 1 . Further, the terminal transmission unit 204 may transmit the questionnaire answer information received by the terminal reception unit 203 to the avatar providing device 1 . Here, when the request or dialogue input information received by the terminal transmission unit 204 is voice information, the terminal transmission unit 204 may transmit the received information to the avatar providing device 1 as voice information. A character string corresponding to the information may be acquired and the acquired character string may be transmitted to the avatar providing device 1 .

For example, the terminal transmission unit 204 acquires character string information corresponding to the voice information by performing voice recognition processing or the like on the voice information. The information used for speech recognition processing may be held by the terminal device 2 or may be held by another server device (not shown) or the like accessible by the terminal device 2 . In addition, the terminal transmission unit 204 transmits the received voice information to a server device (not shown) or the like that performs voice recognition or the like to acquire information of a character string corresponding to the voice information, and transmits the character string. may be executed to obtain the character string information corresponding to the voice information.
The terminal transmission unit 204 can be realized by, for example, wireless or wired communication means.

Next, an example of the operation of the avatar provision device 1 of the avatar utilization system 1000 will be described using the flowchart of FIG.

(Step S101) The instruction receiving unit 111 determines whether or not a selection instruction has been received. Here, as an example, it is assumed that the selection instruction received by instruction receiving section 111 is a selection instruction having a user identifier. Also, as an example here, it is assumed that a selection instruction for newly using an avatar has one or more user attribute values and information indicating whether or not the avatar is to be used temporarily. Note that the selection instruction for using the avatar for the first time may have information indicating that the avatar is for new use. If the selection instruction has been received, the process proceeds to step S102, and if not, the process proceeds to step S107.

(Step S102) The avatar acquisition unit 113 determines whether the selection instruction received in step S101 has a user attribute value of 1 or more. If so, go to step S103; if not, go to step S105. It is also possible to determine whether the selection instruction received in step S101 has information indicating new use, and if so, proceed to step S103, and if not, proceed to step S105.

(Step S103) The avatar acquisition unit 113 acquires avatar information from the avatar storage unit 101 using one or more user attribute values included in the selection instruction received in step S101. For example, avatar information associated with the same one or two or more user attribute values as the selection instruction may be acquired from the avatar storage unit 101, and the one or two or more user attribute values included in the selection instruction may be stored in the learner. The avatar information identified by the avatar identifier obtained as an output from the learner stored in the unit 103 may be acquired.

(Step S104) The storage unit 112 stores the user identifier included in the selection instruction received in step S101, the avatar identifier included in the selection instruction received in step S103, the avatar identifier included in the avatar acquisition unit 113 in step S103, the user identifier included in the selection instruction received in step S101, and the temporary It is stored in the user identifier storage unit 102 in association with information indicating whether or not it is used.

(Step S105) Avatar acquisition unit 113 acquires the avatar identifier stored in user identifier storage unit 102 in association with the user identifier of the selection instruction received in step S101, and acquires the avatar identified by this avatar identifier. Information is obtained from the avatar storage unit 101 . Here, in particular, the user identifier included in the selection instruction received in step S101 and the avatar identifier stored in the user identifier storage unit 102 in association with the information indicating non-temporary use are acquired, and the avatar identifier is used for identification. The avatar information to be used is acquired from the avatar storage unit 101 .

(Step S106) The avatar transmission unit 114 transmits the avatar information acquired by the avatar acquisition unit 113 in the previous step S103 or step S105 to the terminal device 2 that transmitted the selection instruction. Then, the process returns to step S101.

(Step S107) The input information receiving unit 115 determines whether or not interactive input information has been received. Here, as an example, it is assumed that the received dialogue input information is associated with the avatar identifier of the avatar output by the terminal device 2 and the user identifier of the user who uses the terminal device 2 . If so, go to step S108; if not, go to step S116.

(Step S108) The response information acquiring unit 121 determines whether the avatar corresponding to the interactive input information received in step S107 has an interactive function. For example, it is determined whether the avatar information corresponding to this avatar has information indicating that it has an interactive function. If the interactive function is available, the process proceeds to step S109; otherwise, the process returns to step S101.

(Step S109) The response information acquisition unit 121 acquires response information corresponding to the dialogue input information received in step S107. For example, among the dialogue models stored in the avatar storage unit 101, the dialogue model associated by default with the avatar identifier associated with the received dialogue input information, or the dialogue model associated with the received dialogue input information Acquire response information using an interaction model associated with another avatar identified by one or more interaction model identifiers associated with a set of an avatar identifier and a user identifier. If response information cannot be obtained from the dialogue model, response information indicating that appropriate response information cannot be obtained is obtained. This response information may be response information indicating that the interaction input information is not understood.

(Step S110) The response information transmission unit 122 transmits the response information acquired in step S108 to the terminal device 2 that transmitted the dialogue input information.

(Step S111) The evaluation acquisition unit 123 determines whether the avatar corresponding to the dialogue input information received in step S107 has a learning function. For example, it is determined whether the avatar information corresponding to this avatar has information indicating that it has a learning function. If the learning function is provided, the process proceeds to step S112, and if not, the process returns to step S101.

(Step S112) The evaluation acquisition unit 123 acquires an evaluation result of the response information acquired in step S109. This evaluation result may be considered as an evaluation result for the interactive function.

(Step S113) The learning processing unit 124 determines whether the evaluation result obtained in step S110 is so low as to satisfy evaluation conditions prepared in advance. For example, this determination may be considered as determining whether or not the value of the evaluation result matches one or more values indicating that the evaluation indicated by the evaluation conditions prepared in advance is low. One or more values herein may be a range of values. If the evaluation result is low, the process proceeds to step S114; otherwise, the process returns to step S101.

(Step S114) The learning processing unit 124 selects an interaction model associated with an avatar other than the avatar indicated by the avatar identifier associated with the interaction input information received in step S107, Get one or more corresponding interaction models. For example, a dialogue model is acquired in which the field, topic, etc. corresponding to the dialogue model match the field and topic of the dialogue input information. Obtaining a dialogue model may be considered as obtaining a dialogue model identifier that identifies the dialogue model. Instead of acquiring the dialogue model, teacher data associated with other avatars and used for the dialogue function corresponding to the received dialogue input information may be acquired. Also, using this teacher data, machine learning may be performed to acquire or update the dialogue model used by the avatar indicated by the avatar identifier of the dialogue input information.

(Step S115) The accumulation unit 112 associates the dialogue model identifier of the dialogue model acquired in step S112 with the avatar identifier and the user identifier associated with the dialogue input information received in step S107, and stores the dialogue model identifier in the user identifier storage unit. 102 is stored (for example, added). Then, the process returns to step S101.

(Step S116) The input information receiving unit 115 determines whether a request has been received. Here, as an example, the request to be received is associated with the avatar identifier of the avatar output by the terminal device 2 and the user identifier of the user using the terminal device 2 . If so, go to step S117; if not, go to step S125.

(Step S117) The request processing unit 120 determines whether the avatar corresponding to the request received in step S116 has an automatic processing function. For example, it is determined whether the avatar information corresponding to this avatar has information indicating that it has an automatic processing function. If the automatic processing function is provided, the process proceeds to step S118; otherwise, the process returns to step S101.

(Step S118) The request processing unit 120 determines automatic processing corresponding to the request received in step S116. For example, among the automatic processing models stored in the avatar storage unit 101, the automatic processing model associated by default with the avatar identifier associated with the received request, or the avatar identifier associated with the received request Information defining an automatic process is obtained using an automatic process model associated with another avatar identified by one or more automatic process model identifiers associated with a set of a user identifier and a user identifier. Obtaining information that defines automatic processing can be considered substantially determining the automatic processing defined by this information. If information defining automatic processing cannot be obtained from any automatic processing model, it is determined that appropriate automatic processing cannot be determined. In this case, information may be acquired indicating that an appropriate automatic process cannot be determined.

(Step S119) The request processing unit 120 executes the automatic processing obtained in step S108. Note that the request processing unit 120 may transmit information indicating the result of automatic processing to the terminal device 2 that transmitted the request. If the automatic processing could not be determined in step S118, the automatic processing is not executed in step S119, and information indicating that the automatic processing could not be executed is transmitted as information indicating the result of the automatic processing. You may transmit to the terminal device 2 which carried out.

(Step S120) The evaluation acquisition unit 123 determines whether the avatar corresponding to the request received in step S116 has additional functions. For example, it is determined whether the avatar information corresponding to this avatar has information indicating that it has additional functions. If there is an additional function, the process proceeds to step S121; otherwise, the process returns to step S101.

(Step S121) The evaluation acquisition unit 123 acquires an evaluation result of the automatic processing executed in step S119. It should be noted that, in step S119, if the automatic processing could not be executed, for example, an evaluation result about the failure is obtained. The evaluation result here may be considered as the evaluation result for the automatic processing function.

(Step S122) The additional processing unit 125 determines whether the evaluation result obtained in step S121 is so low as to satisfy evaluation conditions prepared in advance. For example, this determination may be considered as determining whether or not the value of the evaluation result matches one or more values indicating that the evaluation indicated by the evaluation conditions prepared in advance is low. One or more values herein may be a range of values. If the evaluation result is low, the process proceeds to step S123; otherwise, the process returns to step S101.

(Step S123) The additional processing unit 125 selects an automatic processing model associated with an avatar other than the avatar indicated by the avatar identifier with which the request received in step S116 is associated, and which corresponds to the request received in step S116. Get the above automatic processing model. For example, an automatic processing model is obtained in which the field corresponding to the automatic processing model matches the requested field. Obtaining an automatic processing model may be considered as obtaining an automatic processing model identifier that identifies the automatic processing model. Note that instead of acquiring the automatic processing model, teacher data associated with other avatars that are used for the automatic processing function corresponding to the received request may be acquired. Also, using this training data, machine learning may be performed to acquire or update the automatic processing model used by the avatar indicated by the avatar identifier associated with the request.

(Step S124) The accumulation unit 112 associates the automatic processing model identifier of the automatic processing model acquired in step S123 with the avatar identifier and user identifier associated with the request received in step S116, 102 is stored (for example, added). Then, the process returns to step S101.

(Step S125) The billing unit 126 determines whether it is time to perform billing processing for using the avatar. For example, it is determined whether or not a predetermined date (for example, a predetermined date of every month) has come for performing billing processing. Further, when a predetermined process or the like is performed, it may be determined that it is time to execute the billing process. If it is the time to execute the billing process, the process proceeds to step S126, and if not, the process returns to step S101.

(Step S126) The billing unit 126 performs billing processing for using the avatar. For example, the billing unit 126, according to the avatar used by the user indicated by the information stored in association with the user identifier in the user identifier storage unit 102, and the dialogue model and automatic processing model used by this avatar, Billing processing is performed for the user. For example, the billing unit 126 performs billing processing of different amounts for one user depending on whether the use of one avatar by one user is temporary use or not. In addition, the number of dialogue models and automatic processing models associated with the avatar used by one user (that is, the dialogue model used by the avatar for the dialogue function and the automatic processing model used for the automatic processing function), and the number of each dialogue Billing processing of an amount that varies according to the amount associated with the model or the automatic processing model is performed. Then, the process returns to step S101.

The billing unit 126 stores teacher data used for the avatar interactive function indicated by information stored in association with the user identifier in the user identifier storage unit 102, and teacher data used for the avatar automatic processing function. An amount of money may be charged according to the number of data, the amount of data, or the like.

Note that in the flowchart of FIG. 2, the process ends when the power is turned off or when the process ends.

Next, the operation of the terminal device 2 of the avatar utilization system 1000 will be briefly described.

When the terminal reception unit 203 receives an instruction to select an avatar from the user, the terminal transmission unit 204 transmits a selection instruction corresponding to this instruction to the avatar providing device 1 in association with the user identifier. When the terminal receiving unit 201 receives avatar information from the avatar providing device 1 in response to the transmitted selection instruction, the terminal output unit 202 outputs an avatar using the received avatar information.

When the terminal reception unit 203 receives a request from the user to perform one automatic process on the output avatar, the terminal transmission unit 204 transmits the request with the user identifier and the avatar identifier of the output avatar. , and transmitted to the avatar providing device 1. When terminal receiving section 201 receives information indicating the result of automatic processing in response to the transmitted request, terminal output section 202 outputs the received information as an output from the avatar currently being output.

When the terminal reception unit 203 receives input of one piece of dialogue input information for the output avatar from the user, the terminal transmission unit 204 converts this dialogue input information into the user identifier and the avatar identifier of the output avatar. , and transmitted to the avatar providing device 1. When terminal receiving section 201 receives response information in accordance with the transmitted dialogue input information, terminal output section 202 outputs the received response information as an output from the avatar currently being output.

When the terminal receiving unit 201 receives information indicating the results of automatic processing and questionnaire information associated with response information from the avatar providing device 1, the terminal output unit 202 outputs this questionnaire information to, for example, Output as an avatar output or the like, the terminal reception unit 203 receives the user's response to the output questionnaire information, and the questionnaire response information indicating the received response is associated with the user identifier and the avatar identifier (or received may be transmitted to the avatar providing device 1 in association with the questionnaire information obtained.

(Specific example 1)
A specific operation of avatar utilization system 1000 according to the present embodiment will be described below. Here, it is assumed that the avatar providing device 1 is a server device, and the terminal device 2 is a smart phone used by a user. Further, here, information (for example, a credit card number, an identifier of electronic money, etc.) necessary for settlement of a user using the terminal device 2 is stored in advance in a storage unit such as a server device (not shown) that performs settlement. (not shown) is stored in association with the user identifier.

In order for user A to newly use an avatar, the terminal output unit 202 of the terminal device 2 displays an input screen for applying for the use of the avatar. User identifier "U0001" and user attribute values for a plurality of predetermined attributes such as one's own gender "male" and age "25" are entered. When information designating usage by subscription is input, the terminal reception unit 203 receives these inputs. Then, if user A presses a send button (not shown) for sending the input information, the terminal reception unit 203 accepts this operation, and the terminal transmission unit 204 responds to this operation by transmitting the above A selection instruction having the received information is transmitted to the avatar providing device 1 .

When the instruction receiving unit 111 of the avatar provision device 1 receives the selection instruction from the terminal device 2 of the user A, the selection instruction has the user attribute value, so the avatar acquisition unit 113 acquires a new avatar. To do so, machine learning prediction processing is performed using a learning device stored in advance in the learning device storage unit 103 and the plurality of user attribute values described above, and an avatar identifier is acquired. For example, the avatar acquisition unit 113 inputs user attribute values for a plurality of attributes to a learning device and acquires an avatar identifier output by the learning device. The learning device here indicates, for example, the user attribute values of the same plurality of attributes as above obtained for each of the plurality of users, the avatar identifier of the avatar used by each user, and the evaluation of the avatar used by each user. It may be a learning device obtained by performing machine learning using information having information as teacher data, and user identifiers of the same multiple attributes as above for multiple users who have high evaluations of the avatars they use. and the avatar identifier of the avatar used by each user as training data. Note that the learning device is not limited to the above learning device as long as it can identify an avatar suitable for the user by performing machine learning prediction processing using user attribute values for a plurality of attributes. Assume that the avatar identifier acquired using the learning device is "A101".

FIG. 3 is an avatar management table for managing avatar information stored in the avatar storage unit 101. As shown in FIG. The avatar management table has attributes of "avatar ID", "avatar information", "interaction model", "automatic processing model", "learning", and "addition". "Avatar ID" is an avatar identifier. "Avatar information" is avatar information, and indicates the file name of the avatar information here. Here, for convenience of explanation, the case where the avatar information of one avatar is composed of one file is shown as an example, but the avatar information of one avatar is composed of a plurality of files. Avatar information of one avatar may be information obtained by packaging a plurality of files into one. This also applies to the following "dialogue model" and "automatic processing model". "Dialogue model" is the dialog model identifier of the dialog model used by the avatar by default. "Automatic processing model" is the automatic processing model identifier of the automatic processing model that the avatar uses by default. For avatars that do not have a dialogue function, the value of "dialogue model" is blank or NULL. For avatars that do not have an automatic processing function, the value of "automatic processing model" is blank or NULL. "Learning" is information indicating whether or not the avatar has a learning function related to dialogue, with a value of "1" having a learning function and a value of "0" not having a learning function. . A non-interactive avatar typically does not have learning capabilities. "Additional" is information indicating whether or not there is an additional function for automatic processing, the value "1" indicates that there is an additional function for automatic processing, and the value "0" indicates that there is no additional function. indicates that Generally, avatars that do not have automation do not have the additional functionality of automation. In the avatar management table, one record (row) corresponds to avatar information of one avatar.

FIG. 4A is a user identifier management table for managing user identifiers stored in the user identifier storage unit 102. FIG. The user identifier management table has attributes of "user ID", "avatar ID", "temporary use", "additional interaction model", and "additional automatic processing model". "User ID" is a user identifier. "Avatar ID" is the avatar identifier of the avatar used by the user, "temporary use" is information indicating whether or not the avatar is used temporarily, the value "1" indicates temporary use, and the value "0" indicates temporary use. ” indicates non-temporary use. Here, non-temporary use is assumed to be use by subscription in which billing occurs on a monthly basis. "Additional dialogue model" is assumed to be a dialogue model identifier of one or more dialogue models added by the learning function related to avatar dialogue. "Additional automatic processing model" is assumed to be an automatic processing model identifier of one or more automatic processing models added by the addition function of automatic processing of the avatar.

Here, among the records in the user identifier management table, a record indicating temporary use is deleted by the billing unit 126 or the like when the billing process for that record is completed, and the record is not temporarily used. After the use of the avatar is completed, the record indicating the A record shall be deleted. However, the management of the user identifier management table is not limited to the above mechanism, as long as it is possible to properly charge according to the use of avatars.

The accumulation unit 112 associates the avatar identifier “A101” acquired by the avatar acquisition unit 113 with the user identifier “U0001” included in the selection instruction, and information indicating that the avatar usage included in the selection instruction is not temporary use. , and newly stored in the user identifier storage unit 102 . The user identifier management table shown in FIG. 4A shows a user identifier management table in which records (rows) corresponding to this new information are accumulated. The newly accumulated record is the bottom row in FIG. 0”. At the time when new information is accumulated, there is no dialogue model added by the learning function or no automatic processing model added by the additional function. U0001”, the values of “additional interaction model” and “additional automatic processing model” are blank or NULL.

The avatar acquisition unit 113 acquires the value "D101" of the "avatar information" of the record whose "avatar ID" is "A101" in the avatar management table shown in FIG. and the avatar identifier “A101” are read from the avatar storage unit 101 . Then, the avatar transmission unit 114 transmits the read avatar information to the terminal device 2 of the user A who transmitted the selection instruction.

When the terminal receiving unit 201 of the terminal device 2 of the user A receives the avatar information with the file name "D101", the terminal output unit 202 outputs the avatar using the received avatar information. Show your avatar here.

FIG. 5 is a diagram showing a display example of an avatar displayed on the monitor 202a of the user A's terminal device 2. As shown in FIG.

Next, assuming that user A uses the microphone of terminal device 2 to input dialogue input information "tell me how to watch videos on a smartphone" to the displayed avatar, this voice The input is received by the terminal reception unit 203, transmitted to a server device (not shown) that provides a service of converting the voice information of the received voice into corresponding character string information, and the above voice information converted by this server device Receives the character string information corresponding to "Tell me how to watch videos on a smartphone". Then, the information of the received character string is associated with the user identifier "U0001" of the user A and the avatar identifier "A101" of the currently output avatar, and is transmitted to the avatar providing device 1 as interactive input information. .

The input information receiving unit 115 of the avatar providing device 1 receives interactive input information of a character string "tell me how to watch videos on a smartphone" associated with the user identifier "U0001" and the avatar identifier "A101". Upon reception, the response information acquisition unit 121 sets one or more dialogues as the value of the “dialogue model” of the record corresponding to the avatar identifier “A101” associated with the dialogue input information in the avatar management table shown in FIG. Determine if the model identifier is stored. Here, assuming that the dialog model identifier "C101" is stored by default, the response information acquiring unit 121 determines that the avatar whose avatar identifier is "A101" is an avatar having a dialog function. For this reason, the response information acquiring unit 121 selects the dialog model identified by the default dialog model identifier “C101” from among the dialog models stored in the avatar storage unit 101 in association with the dialog model identifier, Search for sample interaction input information corresponding to the received interaction input information, and obtain response information corresponding to the detected sample interaction input information. Note that the process of detecting sample dialogue input information corresponding to the received dialogue input information is a known technique in dialogue processing using a dialogue model, so detailed description thereof will be omitted.

Here, the response information acquisition unit 121 identifies the response information of the character string "When you tap the video application ABC and select the video you want to watch, the video will be played." Suppose that it is obtained using a dialogue model.

The response information transmission unit 122 transmits the acquired character string response information to the terminal device 2 of the user A who transmitted the dialogue input information.

When the terminal receiving unit 201 of the terminal device 2 of the user A receives the response information, the terminal output unit 202 converts the character string response information into voice information, and the converted voice information is currently being output. The avatar output is output from a speaker or the like (not shown). Specifically, the audio "Tap the video application ABC and select the video you want to watch, and the video will be played back" is output as the output of the avatar. The conversion of the character string into the response information may be performed by the terminal device 2, and the response information of the character string is transmitted to a server device (not shown) or the like that provides a service for converting the character string into voice information, Information of the converted character string may be received from the server device. Instead of converting the response information into voice information, the character string response information may be displayed by arranging it in a balloon corresponding to the currently output avatar, a dialog box, or the like.

Note that the response information transmitting unit 122 may convert the character string response information into voice information and transmit the voice information to the terminal device 2 . In this case, the conversion of character strings into voice information may be performed by the avatar providing device 1, or a server device (not shown) or the like that provides a service for converting character strings into voice information may be used in the same manner as described above. You can do it. The same applies to output of response information and the like, output of information indicating the result of automatic processing, and the like.

Here, it is assumed that the dialogue input information transmitted by the terminal device 2 in response to the input by the user A is information of a character string "I want to know how to obtain a passport", and the dialogue model identifier is "C101". Suppose that sample dialogue input information corresponding to dialogue input information could not be detected in the above default dialogue model, and response information corresponding to the detected sample dialogue input information could not be acquired.

In this case, in the user identifier management table shown in FIG. ID" matches the avatar identifier "A101" associated with the dialogue input information. If the detected record has one or more dialogue model identifiers as the value of "additional dialogue model", the response information acquisition unit 121 uses the dialogue model identified by the one or more dialogue model identifiers. Then, the response information is obtained in the same manner as described above, and the response information obtained by the response information transmitting unit 122 is transmitted to the terminal device 2 .

However, here, it is assumed that the value of the "additional interaction model" of the record detected in the user identifier management table shown in FIG. 4(a) is absent or NULL. In this case, the response information acquisition unit 121 determines that appropriate response information could not be acquired. Then, the response information acquiring unit 121 transmits to the terminal device 2 information such as a predetermined character string indicating that the response information could not be acquired (for example, “I did not understand”). The terminal device 2 that has received this information outputs the voice corresponding to this information as the output of the avatar in the same manner as described above.

When the response information acquiring unit 121 performs the process of acquiring the response information using the dialogue model as described above and, as a result, transmits the response information or information indicating that the response information could not be acquired to the terminal device 2. , the learning processing unit 124 determines that the value of "learning" of the record (row) in the avatar management table shown in FIG. , is a value “1” indicating that it has a learning function related to dialogue. Here, it is assumed that the value is "1" and the avatar is determined to have a learning function.

Therefore, the evaluation acquisition unit 123 acquires the evaluation result for the response information. Here, if the response information can be acquired, the value "1" is acquired as the evaluation result, and if the response information cannot be acquired, the value "0" is acquired as the evaluation result. Here, it is assumed that the value "0" is obtained as the evaluation result.

Then, the learning processing unit 124 determines whether or not the evaluation result obtained by the evaluation obtaining unit 123 is so low as to satisfy a predetermined evaluation condition. Here, if the evaluation condition is that the value of the evaluation result is "0", the learning processing unit 124 determines that the obtained evaluation result is so low that it satisfies the evaluation condition.

Therefore, the learning processing unit 124 acquires a dialogue model in the same field as the field of the dialogue input information received by the input information receiving unit 115 from among the dialogue models associated with avatar identifiers other than the avatar identifier "A101". do.

6A is a dialogue model management table for managing dialogue models stored in the avatar storage unit 101. FIG. The dialogue model management table has attributes of "dialogue model" and "field". "Interaction model" is an interaction model identifier. “Field” is the field name of the field corresponding to the dialogue model, for example, the field name indicating the field of sample dialogue input information included in the dialogue model.

The learning processing unit 124 acquires the field name of the content of the interactive input information from the interactive input information "I want to know how to obtain a passport." For example, from the dialog input information, morphological analysis or the like is performed to obtain noun phrases, and one or more noun phrases prepared for each of a plurality of field names pre-stored in a storage unit (not shown), Alternatively, using information such as a combination of a plurality of noun phrases, the field name corresponding to the noun phrase that matches the noun phrase obtained from the dialogue input information is obtained as the field name of the dialogue input information. For example, it is assumed here that the field name "travel" is obtained from the noun phrase "passport".

The morphological analysis and the like may be performed by the avatar providing device 1, or may be performed using a server device (not shown) or the like that provides a service for performing morphological analysis. Information used for morphological analysis may be possessed by the avatar providing device 1, or information possessed by another server device (not shown) or the like may be used.

The learning processing unit 124 detects, in the dialogue model management table shown in FIG. A dialogue model identifier that is the value of the "dialogue model" of the record and is associated with the avatar identifier "A101" in the avatar management table shown in FIG. 3, and the user identifier management table shown in FIG. acquires the dialogue model identifiers except for the dialogue model identifiers associated with the user identifier “U0001” and the avatar identifier “A101”. Here, it is assumed that "C103" is acquired as the dialogue model identifier. Then, accumulation unit 112 associates the dialogue model identifier acquired by learning processing unit 124 with user identifier “U0001” and avatar identifier “A101” associated with the dialogue input information, and accumulates it in user identifier storage unit 102. . Specifically, in the user identifier management table shown in FIG. , add the dialog model identifier “C103” obtained above. If multiple dialog model identifiers with the same corresponding field name are detected, all of them may be acquired and stored in the user identifier management table, or only some of them may be acquired and stored. You may When only part of the data is to be acquired, the target to be acquired may be determined in any way, for example, the target to be acquired may be determined in descending order of the amount of data or in ascending order of the amount of data. , may randomly determine the target to be acquired.

FIG. 4(b) is a diagram showing a dialogue model management table obtained by adding the dialogue model identifier obtained above to the dialogue model management table shown in FIG. 4(a).

As a result, when the dialogue input information associated with the user identifier "U0001" and the avatar identifier "A101" is received, the dialogue model prepared by default whose dialogue model identifier is "C101" is used to perform the above When the response information corresponding to the dialogue input information cannot be acquired, the dialogue model identifier is "C103", which is associated with the user identifier "U0001" and the avatar identifier "A101", according to the user identifier management table. By performing the process of acquiring the response information corresponding to the interaction input information, it can be expected that appropriate response information can be acquired, and the interaction function of the avatar used by the user can be improved. In particular, the dialog model whose dialog model identifier is "C103" corresponds to the field of "travel". It is possible to increase the possibility of acquiring appropriate response information and to make appropriate responses. Note that response information corresponding to the above-described dialogue input information may be acquired using the added dialogue model and transmitted to the terminal device 2 . In this case, as described above, the information indicating that the response information could not be acquired may not be transmitted to the terminal device 2 .

In the above, in the avatar management table shown in FIG. 3, the value of "learning" of the record (row) whose "avatar ID" matches the avatar identifier "A101" associated with the corresponding input information is the learning about dialogue. When the value is “0” indicating that the function is not provided, or when the evaluation result of the interactive function obtained by the evaluation obtaining unit 123 is not low enough to satisfy the evaluation condition, the learning processing unit 124 Learning processing for adding a dialogue model such as is not performed.

Next, assuming that the user A uses the microphone of the terminal device 2 to input a voice request to the displayed avatar, "I want you to purchase the magazine ABC." The receiving unit 203 receives the voice information of the received voice and transmits it to a server device (not shown) that provides a service of converting it into corresponding character string information, and the server device responds to the converted voice information. Receive string information. Then, the information of the received character string is transmitted as a request to the avatar providing apparatus 1 in association with the user identifier "U0001" of the user A and the avatar identifier "A101" of the currently output avatar.

When the input information receiving unit 115 of the avatar provision device 1 receives a request for a character string "I want you to purchase the magazine ABC" associated with the user identifier "U0001" and the avatar identifier "A101", The request processing unit 120, in the avatar management table shown in FIG. Determine if is stored. Here, assuming that the automatic processing model identifier "M101" is stored by default, the request processing unit 120 determines that the avatar with the avatar identifier "A101" has the automatic processing function. Therefore, the request processing unit 120 selects the automatic processing model identified by the default automatic processing model identifier "M101" from among the automatic processing models stored in the avatar storage unit 101 in association with the automatic processing model identifier. In the model, look up the sample requests corresponding to the requests received above and obtain the information defining the automatic processing corresponding to the sample requests found. For example, the request processing unit 120 may substitute the character string "Magazine ABC" included in the request into the information defining the automatic process to be acquired as the value of the variable of the information defining the automatic process. Note that the process of detecting the sample dialogue input information corresponding to the received request is the same technique as the above-described process of detecting the sample dialogue input information corresponding to the received dialogue input information, and will not be described in detail here. omitted.

Here, the request processing unit 120 uses the automatic processing model identified by the automatic processing model identifier "M101" to process the information defining the process of purchasing "Magazine ABC" from a specific e-commerce site that sells magazines. obtained by

The request processing unit 120 executes the automatic processing using the information defining the above automatic processing, and performs the processing of purchasing "magazine ABC" for the user A from a specific electronic commerce site. Since the process of automatically purchasing a product is a well-known technique, detailed description thereof will be omitted here. Then, information indicating that the purchase has been completed is transmitted to the terminal device 2 of the user A as information indicating the result of the automatic processing. It is assumed here that the information indicating that the purchase has been completed is character string information. For example, assume that this character string is "Purchase of magazine ABC has been completed."

When the terminal receiving unit 201 of the terminal device 2 of the user A receives the information indicating that the purchase has been completed, the terminal output unit 202 converts the character string response information into voice information, and outputs the converted voice information. , is output from a speaker or the like (not shown) as the output of the currently output avatar. Specifically, a voice saying "Purchase of magazine ABC is completed" is output as the output of the avatar.

Here, it is assumed that the request transmitted by the terminal device 2 in response to the input by the user A is character string information "I want to make a reservation for a hotel in Tokyo on October 2", and the dialog model identifier is "M101 ”, the sample interactive input information corresponding to the request could not be detected, and the response information corresponding to the detected sample request could not be obtained.

In this case, the request processing unit 120 determines that in the user identifier management table shown in FIG. , find a record that matches the avatar identifier "A101" associated with the above request. If the detected record has one or more automatic processing model identifiers as the value of "additional automatic processing model", the request processing unit 120 performs the automatic processing identified by the one or more automatic processing model identifiers. Using the model, in the same manner as described above, information defining automatic processing corresponding to the request is acquired, automatic processing is performed, and the processing result is transmitted to the terminal device 2 .

However, here, it is assumed that the value of the "additional automatic processing model" of the record detected in the user identifier management table shown in FIG. 4B is absent or NULL. In this case, the request processing unit 120 determines that automatic processing could not be executed. Then, the request processing unit 120 transmits to the terminal device 2 information such as a predetermined character string indicating that the automatic processing has failed (for example, “I did not understand”). The terminal device 2 that has received this information outputs the voice corresponding to this information as the output of the avatar in the same manner as described above.

When the request processing unit 120 executes the automatic processing, or when the automatic processing cannot be executed and transmits information indicating that it could not be executed to the terminal device 2, the learning processing unit 124 executes the avatar shown in FIG. In the management table, the value of "addition" of the record (row) in which the "avatar ID" matches the avatar identifier "A101" associated with the corresponding input information is a value indicating that the automatic processing addition function is provided. It is determined whether it is a certain "1". Here, it is assumed that the value is "1" and the avatar is determined to have an additional function.

Therefore, the evaluation acquisition unit 123 acquires the evaluation result for the automatic processing. Here, when the automatic processing can be executed, the value "1" is obtained as the evaluation result, and when the automatic processing cannot be executed, the value "0" is obtained as the evaluation result. Here, it is assumed that the value "0" is obtained as the evaluation result.

Then, the additional processing unit 125 determines whether the evaluation result obtained by the evaluation obtaining unit 123 is so low as to satisfy a predetermined evaluation condition. Here, if the evaluation condition is that the value of the evaluation result is "0", the additional processing unit 125 determines that the evaluation result is so low that it satisfies the evaluation condition.

Therefore, the additional processing unit 125 acquires the automatic processing model of the same field as the requested field received by the input information receiving unit 115 from among the automatic processing models associated with avatar identifiers other than the avatar identifier "A101". do.

FIG. 6B is an automatic processing model management table for managing automatic processing models stored in the avatar storage unit 101. FIG. The automatic processing model management table has attributes of "automatic processing model" and "field". "Automatic processing model" is an automatic processing model identifier. "Field" is the field name of the field corresponding to the automatic processing model, for example, the field name indicating the field of sample requests included in the automatic processing model.

The additional processing unit 125 acquires the field name of the content of the request from the request "I would like to make a reservation for a hotel in Tokyo on October 2nd." For example, from the request, morphological analysis or the like is performed to acquire noun phrases, and one or more noun phrases prepared for each of a plurality of field names pre-stored in a storage unit (not shown), or a plurality of Using the information such as the combination of noun phrases in the request, the field name corresponding to the noun phrase that matches the noun phrase obtained from the request is obtained as the field name of the request. For example, here, it is assumed that the field name "hotel reservation" is acquired from the combination of the noun phrase "hotel" and the noun phrase "reservation".

The morphological analysis and the like may be performed by the avatar providing device 1, or may be performed using a server device (not shown) or the like that provides a service for performing morphological analysis. Information used for morphological analysis may be possessed by the avatar providing device 1, or information possessed by another server device (not shown) or the like may be used.

The additional processing unit 125 detects, in the automatic processing model management table shown in FIG. The automatic processing model identifier "M101" which is the value of the "automatic processing model" of the record and is associated with the avatar identifier "A101" in the avatar management table shown in FIG. Acquire automatic processing model identifiers excluding automatic processing model identifiers associated with user identifier "U0001" and avatar identifier "A101" in the shown user identifier management table (here, there is no corresponding automatic processing model identifier) do. Here, it is assumed that "M104" is acquired as the automatic processing model identifier. Then, the storage unit 112 stores the automatic processing model identifier acquired by the additional processing unit 125 in the user identifier storage unit 102 in association with the user identifier “U0001” and the avatar identifier “A101” associated with the request. Specifically, in the user identifier management table shown in FIG. , the automatic processing model identifier "M104" obtained above is added.

FIG. 4(c) is a diagram showing an automatic processing model management table obtained by adding the automatic processing model identifier obtained above to the automatic processing model management table shown in FIG. 4(b).

As a result, when a request associated with the user identifier "U0001" and the avatar identifier "A101" is received, the automatic processing model prepared by default whose automatic processing model identifier is "M101" is used to When the information specifying the automatic processing corresponding to the request cannot be obtained, the automatic processing model identifier "M104" associated with the user identifier "U0001" and the avatar identifier "A101" is used according to the user identifier management table. Therefore, it is expected that information specifying appropriate automatic processing can be acquired by performing processing to acquire information specifying automatic processing corresponding to the above-mentioned request, and automatic processing of avatars used by users can be expected. function can be improved. In particular, the automatic processing model whose automatic processing model identifier is "M104" corresponds to the field "hotel reservation". is received, the possibility of being able to acquire information specifying appropriate automatic processing can be increased, and an appropriate response can be made. It should be noted that the automatic processing according to the above request may be performed using the added automatic processing model. In this case, as described above, the information indicating that the automatic processing could not be performed may not be transmitted to the terminal device 2 .

In the above, in the avatar management table shown in FIG. 3, the value of "addition" of the record (row) whose "avatar ID" matches the avatar identifier "A101" associated with the corresponding input information is the addition related to the dialogue. If the value is “0” indicating that the function is not provided, or if the evaluation result of the automatic processing function acquired by the evaluation acquisition unit 123 is not low enough to satisfy the evaluation condition, the additional processing unit 125 Learning processing for adding an automatic processing model as described above is not performed.

Assume that it is now time to perform billing processing for the user's use of the avatar. The timing here is a predetermined date of every month, and when this date arrives, the billing unit 126 performs billing processing for each user regarding the use of the avatar in the previous month of the month. .

FIG. 7 is a billing information management table stored in the avatar storage unit 101. As shown in FIG. The billing information management table manages information for determining billing for use of avatars. The billing information management table has attributes of "condition" and "amount". "Condition" is a condition for determining charging. Also, the "amount" is a calculation formula for calculating the amount of money when the condition is satisfied or a variable related to the condition. Here, it is assumed that the user is charged the sum of the amounts corresponding to all the conditions that the user's use of the avatar satisfies among the conditions of the billing information management table. Note that V ₁ , V ₂ , etc. are arbitrary values that indicate the amount of money. Here, the "conditions" are expressed in natural language, but the "conditions" are, for example, one or more functions corresponding to these natural languages, method names, "if", "then", etc. It may be composed of the control syntax shown. Also, the "condition" may be an algorithm for performing judgment processing corresponding to these natural languages.

The billing unit 126 reads the information of each record (row) of the user identifier management table shown in FIG. 4(c), and performs billing processing using the read information. Here, first, the billing process for a user whose "user ID" in the first row from the top of the user identifier management table is "U2023" will be described. For example, the billing unit 126 acquires the value “A537” of the “avatar ID” in the record in the first row from the top, and the record (row ) is neither blank nor NULL, the condition "with interactive function" indicated by the value of "Condition" in the first row from the top of the billing information management table shown in FIG. It is determined that the condition is satisfied, and the corresponding "amount" value "V ₁ " is obtained. If any, it is determined that the condition is not satisfied, and the "amount" value is not obtained. If the value of "automatic processing model" of the same record is neither blank nor NULL, the value of "condition" in the second row from the top of the billing information management table shown in FIG. It is determined that the condition "Yes" is satisfied, and the corresponding "amount" value "V ₂ " is acquired. If any, it is determined that the condition is not satisfied, and the "amount" value is not acquired. Also, if the value of "learning" of the same record is "1", the value of "condition" in the second row from the top of the billing information management table shown in FIG. It is determined that the condition is satisfied, and the corresponding "amount" value "V ₃ " is obtained. If it is "0", it is determined that the condition is not satisfied and the "amount" value is not obtained. Also, if the value of "learning" of the same record is "1", it means that "there is an automatic processing addition function" indicated by the value of "condition" in the second row from the top of the billing information management table shown in FIG. It is determined that the condition is satisfied, and the corresponding "amount" value "V ₄ " is obtained. If it is "0", it is determined that the condition is not satisfied and the "amount" value is not obtained. Here, assuming that all the conditions from the top to the fourth row of the billing information management table shown in FIG. Assume that V ₁ to V ₄ are obtained as values.

Further, the billing unit 126 determines whether or not the value of "temporary use" in the record on the first row from the top of the user identifier management table shown in FIG. 4(c) is "1". Here, since the value is "0", which indicates non-temporary use, the billing unit 126 does not satisfy the condition "temporary use" on the fifth line from the top of the billing information management table shown in FIG. Assuming that the condition "not for temporary use" on the sixth line from the top is satisfied, the value "V ₆ " of the "amount" on the sixth line from the top is acquired. If the value is "1", the condition "temporary use" on the fifth line from the top of the billing information management table shown in FIG. Assuming that the condition is not satisfied, the value "V ₅ " of the "amount" on the fifth line from the top is acquired.

Also, the billing unit 126 determines whether the value of the “additional interaction model” in the record on the first line from the top is either blank or NULL. In this case, neither of these are the case. Therefore, it is determined that the condition "additional dialogue model exists" in the seventh row from the top of the billing information management table shown in FIG. 7 is satisfied, and the dialogue model identifier which is the value of "additional dialogue model" is satisfied. is counted, and the counted number “2” is substituted as the value of P of the “amount” to obtain the value of the amount to be charged “V ₇ ×2”. It should be noted that if the value of the "additional interaction model" is either blank or NULL, no billing amount is acquired for this condition.

Further, the billing unit 126 determines whether the value of "additional automatic processing model" in the record on the first line from the top is neither blank nor NULL. Here, since the column is either blank or NULL, it is determined that the condition "additional automatic processing model exists" in the eighth row from the top of the billing information management table shown in FIG. 7 is not satisfied. You don't get the amount to charge. If the field is neither blank nor NULL, the number of interaction model identifiers, which are the values of the "additional interaction model", is counted, and the counted number is substituted as the Q value of the "amount" to determine the amount of money to be charged. We get the value "V ₈ ×Q".

Then, the billing unit 126 acquires V ₁ +V ₂ +V ₃ +V ₄ +V ₆ +V ₇ ×2, which is the sum of the billing amounts obtained above, as the billing amount for the user whose “user ID” is “U2023”. do. The billing unit 126 then associates the acquired amount with the user identifier “U2023” and transmits it to a server device (not shown) that performs settlement. The server device that performs payment uses the received user identifier and amount of money to perform payment processing for the user to pay the usage fee for the avatar.

The billing unit 126 indicates that the record in the first row for which billing processing has been performed in the user identifier management table shown in FIG. Therefore, this record is determined to be a record that requires billing processing even at the timing of the next billing, and is left without being deleted. If the value of "temporary use" is "0", this record is deleted because the information in this record will not be used for the next billing. The same applies to other records in the user identifier management table. Note that the deletion here may be to add information such as a flag indicating that the record has been deleted.

The billing unit 126 similarly performs the same billing process as described above for the records in the second and subsequent rows from the top of the user identifier management table shown in FIG. 3(c). As a result, appropriate billing can be performed according to the status of the avatar used by the user (for example, whether or not a dialogue model has been added).

In the above, when the user who has already used the avatar for purposes other than temporary use outputs (for example, displays) the avatar on the terminal device 2, the selection instruction having the user identifier is given via the terminal device 2. When the selection instruction is received by the instruction reception unit 111, the avatar acquisition unit 113 determines that the value of "user ID" matches the user identifier of the selection instruction from the user identifier management table. The avatar information indicated by the avatar identifier, which is the value of "avatar ID" of the record whose "temporary use" value is "0", is obtained from the avatar management table shown in FIG. The avatar information may be transmitted by the avatar transmission unit 114 to the terminal device 2 of the user. In addition, in the above, when the user who has temporarily used the avatar before uses the avatar again, it is sufficient to perform an operation or the like for newly using the avatar in the same manner as described above.

(Specific example 2)
In the above specific example, when the evaluation of the response information by the evaluation acquisition unit 123 is low, the dialogue model is added by the learning function, and when the evaluation of the automatic processing is low, the automatic processing model is added by the additional function. Specifically, when the evaluation of the response information by the evaluation acquisition unit 123 is low, teacher data for the interactive function is added by the learning function, and when the evaluation of the automatic processing is low, teacher data for the automatic processing function is added by the additional function. Examples are described below.

FIG. 8 shows a second avatar management table for managing avatar information stored in the avatar storage unit 101. This second avatar management table is referred to as "dialogue teacher data" in the avatar management table shown in FIG. , to which an attribute of "automatic processing training data" is added. "Dialogue teacher data" is teacher data for interactive functions prepared for avatars by default, and indicates the file name of the teacher data here. The teacher data for the dialogue function is, for example, information having a plurality of sets of utterances for dialogue and responses to the utterances. However, teacher data is not limited to such information. "Automatic processing teaching data" and teaching data for an automatic processing function prepared for avatars by default, and here indicate the file name of the teaching data. The training data for the automatic processing function is, for example, information comprising multiple sets of information specifying a request from the user to perform processing and the processing to be performed in response to this request. However, teacher data is not limited to such information. The dialogue model indicated by the "dialogue model" value of each record is a dialogue model obtained by performing machine learning using the teacher model indicated by the "dialogue teacher model" value of the same record, such as a learner. and In addition, the dialogue model indicated by the "automatic processing model" value of each record is a dialogue model obtained by performing machine learning using the training model indicated by the "automatic processing training model" value of the same record. Suppose it is a learner.

FIG. 9 is a second user identifier management table for managing user identifiers stored in the user identifier storage unit 102. In the user identifier management table shown in FIGS. Instead of "dialogue model" and "additional automatic processing model", it has "dialogue model" and "automatic processing model" as attributes, and further has attributes "additional dialogue training data" and "additional automatic processing training data". have. "Interaction model" is an identifier of an interaction model and an automatic processing model used by the avatar indicated by the "avatar ID" when the user indicated by the "user ID" of the same record uses the avatar indicated by the "avatar ID". The “additional dialogue training data” is training data for the dialogue function acquired by the learning processing unit 124 for learning the dialogue function according to the evaluation of the evaluation acquisition unit 123, and the “additional automatic processing training data” is: This is teacher data for the automatic processing function that the addition processing unit 125 has acquired for adding the automatic processing function.

FIG. 10(a) is a dialog teaching data management table for managing teaching data for interactive functions stored in the avatar storage unit 101. FIG. The dialogue training data management table has attributes of “dialogue training data” and “field”. “Dialogue training data” is training data for the dialogue function. “Field” is the field name of the field corresponding to the teacher data for the interactive function, for example, the field name indicating the field of dialogue included in the teacher data.

FIG. 10(b) is an automatic processing teaching data management table for managing teaching data stored in the avatar storage unit 101 for the automatic processing function. The automatic processing teaching data management table has attributes of "automatic processing teaching data" and "field". "Automatic processing teaching data" is teaching data for the automatic processing function. "Field" is the field name of the field corresponding to the training data for the automatic processing function, for example, the field name indicating the field of automatic processing executed by the training data.

For example, when a user newly uses an avatar, the second user identifier management table as shown in FIG. The record associated with the avatar identifier, and the default dialogue model identifier and automatic processing model identifier associated with this acquired avatar identifier in the second avatar management table shown in FIG. ” and “automatic processing model” are newly added by the storage unit 112 . The values of the "additional dialogue teaching data" and the "additional automatic processing teaching data" of the record to be added are both blank or NULL by default.

Then, when the instruction receiving unit 111 receives the dialogue input information associated with the user identifier and the avatar identifier, the response information acquiring unit 121 stores the received user identifier in the second user identifier management table shown in FIG. and the response information is acquired using the dialogue model indicated by the dialogue model identifier, which is the value of the “dialogue model” of the record associated with the avatar identifier.

Further, when the instruction receiving unit 111 receives a request associated with the user identifier and the avatar identifier, the response information acquiring unit 121 stores the received user identifier in the second user identifier management table shown in FIG. Using the automatic processing model indicated by the automatic processing model identifier, which is the value of the "automatic processing model" of the record associated with the avatar identifier, the automatic processing according to the request is executed.

When the evaluation of the evaluation acquisition unit 123 for the response information acquired by the response information acquisition unit 121, that is, the evaluation of the avatar's interactive function is low, the learning processing unit 124 acquires , the user identifier associated with the dialogue input information, and the combination of the user identifier and the avatar identifier associated with the dialogue input information, the teaching data for the dialogue function, The identifier of the teacher data in the field corresponding to the dialogue input information received by the receiving unit 111 (that is, the dialogue teacher data identifier) is acquired. Then, the storage unit 112 stores the acquired identifier in the second user identifier management table shown in FIG. Append as a value. Furthermore, the learning processing unit 124 stores teacher data for the interactive function indicated by the added identifier and teacher data indicated by the value of "interactive teacher data" associated with this avatar identifier by default in the second avatar management table. is used to perform machine learning to acquire a learner, and the acquired learner is newly given a dialogue model identifier and stored in the avatar storage unit 101 . Then, the accumulation unit 112 uses this dialogue model identifier to store the value of the "dialogue model" of the record having the user identifier and the avatar identifier associated with the dialogue input information in the second user identifier management table shown in FIG. update (eg, overwrite) the . After updating, the response information acquisition unit 121 acquires response information corresponding to the interaction input information using the interaction model indicated by the updated interaction model identifier. As a result, it is possible to improve the interactive function of the avatar used by the user whose evaluation of the interactive function was low.

Further, when the evaluation of the evaluation acquisition unit 123 for the automatic processing acquired by the response information acquisition unit 121, that is, the evaluation of the automatic processing function of the avatar is low, the learning processing unit 124 acquires the automatic processing teacher data shown in FIG. For automatic processing functions that are not associated with any of the user identifier associated with the request received by the instruction receiving unit 111 and the combination of the user identifier and the avatar identifier associated with the request, from the management table The identifier of the training data in the field corresponding to the request received by the instruction receiving unit 111 (that is, the automatic processing training data identifier) is acquired. The storage unit 112 adds the acquired identifier to the second user identifier management table shown in FIG. 9 as the value of "additional automatic processing teacher data" of the record having the user identifier and avatar identifier associated with the request. do. Furthermore, the learning processing unit 124 stores teacher data for the automatic processing function indicated by the added identifier and teacher data indicated by the value of "automatic processing teacher data" associated with this avatar identifier by default in the second avatar management table. Machine learning is performed using the data, a learning device is acquired, an automatic processing model identifier is newly assigned to the acquired learning device, and the acquired learning device is stored in the avatar storage unit 101 . Using this automatic processing model identifier, the storage unit 112 stores the "automatic Update (eg, overwrite) the value of "processing model". After updating, the response information acquisition unit 121 acquires response information corresponding to the dialogue input information using the automatic processing model indicated by the updated automatic processing model identifier. As a result, it is possible to improve the automatic processing function of the avatar used by the user whose evaluation of the interactive function was low.

In the case of charging, in the processing for determining the amount to be charged in the specific example above, instead of the processing for determining the amount to be charged according to the number of added dialogue models or the number of added automatic processing models (or processing), the amount to be charged can be determined according to the number of values of "additional dialogue training data" and the number of values of "additional automatic processing training data" in the second user identifier management table. good.

As described above, according to the present embodiment, it is possible to change at least one of the information that the avatar used by the user uses for interaction and the information that is used for automatic processing, so that at least one of appropriate interaction and automatic processing can be realized. avatars can be provided. Thereby, for example, the convenience of the avatar used by the user can be improved.

In the above embodiments, each process (each function) may be implemented by centralized processing by a single device (system), or may be implemented by distributed processing by a plurality of devices. may

Further, in the above embodiments, two or more means of communication (instruction receiving unit, input information receiving unit, avatar transmitting unit, response information transmitting unit, etc.) existing in one device are physically realized by one medium. It goes without saying that this is a good thing.

Also, in the above embodiment, the avatar provision device is a stand-alone device, but the avatar provision device may be a stand-alone device or a server device in a server-client system. In the latter case, the output unit and reception unit receive inputs and output screens via communication lines.

Further, in the above embodiments, each component may be configured by dedicated hardware, or components that can be realized by software may be realized by executing a program. For example, each component can be realized by reading and executing a software program recorded in a recording medium such as a hard disk or a semiconductor memory by a program execution unit such as a CPU. During execution, the program execution unit may execute the program while accessing a storage unit (for example, a recording medium such as a hard disk or memory).

The software that implements the avatar providing device in the above embodiment is the following program. In other words, the program includes two or more avatars having at least one function of a dialogue function and an automatic processing function, a combination of the dialogue function and a learning function related to dialogue, an automatic processing function and an automatic processing additional function. a computer capable of accessing an avatar storage unit storing two or more pieces of avatar information for respectively outputting two or more avatars including one or more avatars having at least one set of functions of a set of two or more avatars an instruction receiving unit that receives from the terminal device a selection instruction to select one of; an avatar acquisition unit that acquires avatar information of the avatar corresponding to the selection instruction from the avatar storage unit; It is a program for functioning as an avatar transmission unit that transmits to a terminal device.

In the above program, the transmission step for transmitting information and the reception step for receiving information are performed by hardware. not included).

In the program, the functions realized by the program do not include functions that can be realized only by hardware. For example, functions that can be realized only by hardware such as modems and interface cards, such as an acquisition unit that acquires information and an output unit that outputs information, are not included in the functions realized by the above program.

Also, the number of computers executing this program may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed.

FIG. 11 is a schematic diagram showing an example of the appearance of a computer that implements the avatar provision device according to the embodiment by executing the program. The above embodiments can be implemented by computer hardware and computer programs executed thereon.

In FIG. 11, computer system 900 comprises computer 901 including CD-ROM (Compact Disk Read Only Memory) drive 905 , keyboard 902 , mouse 903 and monitor 904 .

FIG. 12 is a diagram showing the internal configuration of the computer system 900. As shown in FIG. In FIG. 12, a computer 901 includes a CD-ROM drive 905, an MPU (Micro Processing Unit) 911, a ROM 912 for storing programs such as a boot-up program, and a ROM 912 for storing programs such as a boot-up program. A RAM (Random Access Memory) 913 that temporarily stores and provides a temporary storage space, a hard disk 914 that stores application programs, system programs, and data, a bus 915 that interconnects the MPU 911, ROM 912, etc. Prepare. Note that computer 901 may include a network card (not shown) that provides connection to a LAN.

A program that causes the computer system 900 to execute the functions of the avatar providing apparatus according to the above embodiments may be stored in the CD-ROM 921 , inserted into the CD-ROM drive 905 and transferred to the hard disk 914 . Alternatively, the program may be transmitted to computer 901 via a network (not shown) and stored in hard disk 914 . Programs are loaded into RAM 913 during execution. The program may be loaded directly from the CD-ROM 921 or network.

The program does not necessarily include an operating system (OS) or a third party program that causes the computer 901 to perform the functions of the avatar providing device according to the above embodiments. A program may contain only those portions of instructions that call the appropriate functions (modules) in a controlled manner to produce the desired result. How the computer system 900 operates is well known and will not be described in detail.

It goes without saying that the present invention is not limited to the above-described embodiments, and that various modifications are possible and are also included within the scope of the present invention.

As described above, the avatar providing device or the like according to the present invention is suitable as a device or the like for providing avatars, and is particularly useful as a device or the like for providing avatars capable of performing dialogue or automatic processing.

1 avatar providing device 2 terminal device 101 avatar storage unit 102 user identifier storage unit 103 learning device storage unit 111 instruction reception unit 112 storage unit 113 avatar acquisition unit 114 avatar transmission unit 115 input information reception unit 120 request processing unit 121 response information acquisition unit 122 response information transmission unit 123 evaluation acquisition unit 124 learning processing unit 125 addition processing unit 126 charging unit 201 terminal reception unit 202 terminal output unit 202a monitor 203 terminal reception unit 204 terminal transmission unit 1000 avatar use system

Claims (13)

Two or more avatars having at least one of a dialogue function and an automatic processing function, and at least one of a set of a dialogue function and a learning function related to dialogue, and a set of an automatic processing function and an automatic processing additional function an avatar storage unit storing two or more pieces of avatar information for respectively outputting two or more avatars including one or more avatars having a set of functions;
an instruction receiving unit that receives a selection instruction to select one of the two or more avatars from a terminal device;
an avatar acquisition unit that acquires avatar information of an avatar corresponding to the selection instruction from the avatar storage unit;
and an avatar transmitting unit configured to transmit the avatar information acquired by the avatar acquiring unit to the terminal device. The avatar storage unit stores avatar information of one or more avatars having a dialogue function and a learning function related to dialogue,
2. The avatar providing apparatus according to claim 1, further comprising a learning processing unit that performs learning processing related to dialogue according to the learning function of the avatar when a predetermined condition is satisfied for the one avatar having the learning function. . The learning processing unit
3. The avatar providing device according to claim 2, which acquires all or part of teacher data or a dialog model for dialog functions possessed by other avatars, and uses the acquired teacher data or dialog model to perform learning processing related to the dialog. . The avatar storage unit stores avatar information of one or more avatars having an automatic processing function and an additional automatic processing function,
Claims 1 to 3, further comprising an additional processing unit that performs additional processing related to automatic processing according to the additional function of the avatar when a predetermined condition is satisfied for the one avatar having the additional function. The avatar providing device according to any one of claims 1 to 3. an input information receiving unit that receives a request for one avatar from the terminal device;
further comprising a request processing unit that executes automatic processing corresponding to the request using the automatic processing function;
The predetermined condition is
5. The avatar providing device according to claim 4, wherein said request processing unit cannot execute automatic processing corresponding to said request. an input information receiving unit that receives dialogue input information for dialogue with one avatar from the terminal device;
a response information obtaining unit that obtains response information corresponding to the dialogue input information using the dialogue function;
a response information transmission unit that transmits the response information to the terminal device;
An evaluation acquisition unit that acquires an evaluation result for the response information,
The predetermined condition is
4. The avatar providing device according to claim 2, wherein said evaluation result is so low as to satisfy an evaluation condition. The instruction receiving unit receives a selection instruction associated with a user identifier corresponding to the terminal device,
a user identifier storage unit in which a user identifier is stored in association with an avatar used by the user;
an accumulation unit that associates a user identifier associated with the selection instruction received by the instruction receiving unit with an avatar corresponding to the selection instruction and accumulates the user identifier in the user identifier storage unit;
7. A billing unit that bills a user identified by a user identifier corresponding to the selection instruction for use of the avatar corresponding to the selection instruction. 3. The avatar providing device according to . The avatar information stored in the avatar storage unit is associated with one or more attribute values related to the avatar output by the avatar information,
The billing unit
8. The provision of avatars according to claim 7, wherein different amounts of money are charged for the user associated with the selection instruction according to the one or more attribute values associated with the avatar corresponding to the selection instruction. Device. The user identifier storage unit further stores information indicating whether or not the avatar is temporarily used in association with the user identifier,
The billing unit
9. The avatar providing apparatus according to claim 7, wherein different amounts of money are charged to the user according to the value of information indicating whether or not said avatar is to be used temporarily. The selection instruction has one or more user attribute values,
The avatar acquisition unit
The avatar provision device according to any one of claims 1 to 9, wherein the avatar to be acquired is determined using the one or more user attribute values, and the avatar information of the avatar is acquired from the avatar storage unit. Further comprising a learning device storage unit storing learning devices obtained by performing machine learning learning processing on two or more teacher data having one or more user attribute values and avatar identifiers for identifying avatars. death,
The avatar acquisition unit
Machine learning prediction processing is performed using the one or more user attribute values included in the selection instruction and the learning device to obtain an avatar identifier, and the avatar information of the avatar identified by the avatar identifier is obtained from the 11. The avatar providing device according to claim 10, obtained from an avatar storage unit. Two or more avatars having at least one of a dialogue function and an automatic processing function, and at least one of a set of a dialogue function and a learning function related to dialogue, and a set of an automatic processing function and an automatic processing additional function an avatar storage unit storing two or more pieces of avatar information for respectively outputting two or more avatars including one or more avatars having a set of functions; an instruction reception unit; an avatar acquisition unit; An avatar providing method performed using
a step in which the instruction receiving unit receives a selection instruction for selecting one of the two or more avatars from a terminal device;
a step in which the avatar acquisition unit acquires avatar information of an avatar corresponding to the selection instruction from the avatar storage unit;
An avatar providing method, comprising: the avatar transmission unit transmitting avatar information acquired from the avatar storage unit to the terminal device. Two or more avatars having at least one of a dialogue function and an automatic processing function, and at least one of a set of a dialogue function and a learning function related to dialogue, and a set of an automatic processing function and an automatic processing additional function a computer accessible to an avatar store storing two or more pieces of avatar information for respectively outputting two or more avatars including one or more avatars having a set of functions;
an instruction receiving unit that receives a selection instruction to select one of the two or more avatars from a terminal device;
an avatar acquisition unit that acquires avatar information of an avatar corresponding to the selection instruction from the avatar storage unit;
A program for functioning as an avatar transmission unit that transmits the avatar information acquired by the avatar acquisition unit to the terminal device.

Images