Information processing apparatus

Abstract

This disclosure provides an information processing apparatus. The information processing apparatus is capable of executing an AI agent, which is a virtual AI agent capable of engaging in dialogue with a user using natural language. The information processing apparatus has a control unit that performs the following processing: when executing a first AI agent corresponding to a first user among a plurality of users, if the first user moves more than a predetermined distance from the device providing the interface between the first AI agent and the first user, the executing first AI agent is switched to a second AI agent corresponding to a second user among the plurality of users, different from the first user; dialogue information is transferred to the second AI agent, the dialogue information being information accumulated in the first AI agent concerning the dialogue with the first user from the time the first AI agent is executed until the switch to the second AI agent.

Description

Technical Field

[0001] This disclosure relates to an information processing apparatus, etc., that provides information through dialogue with a user. Background Technology

[0002] There are technologies that use natural language processing to conduct dialogues with users and provide information to users based on those dialogues. For example, Japanese Patent Application Publication No. 2021-533400 discloses a method that temporally adapts the responsiveness of an automatic aid to a given verbal expression that has received confirmation responses from two or more assisted devices, in order to accommodate dialogue preferences that vary from user to user. Summary of the Invention

[0003] The purpose of this disclosure is to properly manage information in conversations with users using machine learning.

[0004] One aspect of this disclosure is an information processing apparatus capable of executing an AI agent, which is a virtual AI agent capable of engaging in dialogue with a user using natural language. The information processing apparatus includes a control unit that performs the following processing: when executing a first AI agent corresponding to a first user among a plurality of users, if the first user moves more than a predetermined distance from the device providing the interface between the first AI agent and the first user, the executing first AI agent is switched to a second AI agent corresponding to a second user among the plurality of users, different from the first user; dialogue information is transferred to the second AI agent, the dialogue information being information accumulated in the first AI agent concerning the dialogue with the first user from the time the first AI agent is executed until the switch to the second AI agent.

[0005] In addition, other methods may include the information processing method executed by the aforementioned information processing apparatus, a program for causing a computer to execute the aforementioned information processing method, or a computer-readable storage medium that non-temporarily stores the program.

[0006] The information processing apparatus disclosed herein is capable of appropriately managing information in dialogues with users using machine learning. Attached Figure Description

[0007] The features, advantages, and technical and industrial significance of exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, wherein like symbols denote like elements, and wherein: Figure 1 A diagram illustrating the network structure including the server apparatus according to the first embodiment.

[0008] Figure 2 This diagram illustrates the structural elements of the server device according to the first embodiment.

[0009] Figure 3 A flowchart of the processes performed by the server device according to the first embodiment.

[0010] Figure 4 This is a flowchart of the word substitution process performed by the server device according to a variation of the first embodiment.

[0011] Figure 5 A flowchart of the processes performed by the server device according to the second embodiment. Detailed Implementation

[0012] summary There is a need for services that provide appropriate information to users based on dialogues with them using natural language.

[0013] Relatedly, there is a technology for assistive devices to respond in accordance with each user's preferences by reflecting information accumulated in conversations with the user.

[0014] For example, existing assistive devices respond to user requests based on context, including the user's location, time, the user providing verbal communication, the assistive device's configuration, and / or the status of each device, reflecting each user's preferences.

[0015] Specifically, existing assistive devices, along with the user's request, acquire contextual data indicating the user's condition at the time of the request or the user's intent within the request. The assistive device then uses this contextual data to instruct the client device to perform the action corresponding to the request. For example, consider a situation where a user has previously verbally expressed "Assistant, stop" when an assistive device sounds an alarm, a television is playing video, and a vehicle is charging. In cases where responding to the verbal expression "Assistant, stop" incorrectly stops actions other than issuing an alarm, the user may sometimes provide corrective verbal expressions such as "No, stop the alarm." In such cases, automated assistance can reflect these user corrections and / or preferences in subsequent responses.

[0016] However, there is room for improvement in existing assistive devices. For example, when multiple user accounts are registered in an assistive device, each user uses their associated account to converse with the device. In such conversations, while the assistive device learns from the user's dialogue and refines its responses as described above, the dialogues that are being learned are essentially not shared between different accounts. However, when multiple users express common topics to the assistive device, the conversations could proceed more smoothly if the information learned from conversations with other users was reflected in the dialogue with the currently engaged user regarding matters related to those common topics. Therefore, it is desirable for the dialogue device to facilitate the exchange of information obtained from conversations with users across multiple accounts.

[0017] One aspect of this disclosure relates to an information processing apparatus capable of executing an AI agent, which is a virtual AI agent capable of engaging in dialogue with a user using natural language. The information processing apparatus includes a control unit that performs the following processing: when executing a first AI agent corresponding to a first user among a plurality of users, if the first user moves more than a predetermined distance from the device providing the interface between the first AI agent and the first user, the executing first AI agent is switched to a second AI agent corresponding to a second user among the plurality of users, different from the first user; dialogue information is transferred to the second AI agent, the dialogue information being information accumulated in the first AI agent concerning the dialogue with the first user from the time the first AI agent is executed until the switch to the second AI agent.

[0018] The first user refers to a user who has an account that is used to utilize AI agents.

[0019] The first AI agent refers to the AI ​​agent corresponding to the first user's account. Here, the AI ​​agent is a virtual agent that receives the user's voice or input and engages in dialogue with the user through natural language processing. This virtual agent is implemented through software programs that provide natural language processing-based dialogue functionality with the user. The first AI agent is capable of learning from its dialogue with the first user.

[0020] The second user refers to a user who is different from the first user and has an account for utilizing the AI ​​agent.

[0021] The second AI agent refers to the AI ​​agent corresponding to the second user's account. The second AI agent is able to learn from the conversations with the second user.

[0022] Dialogue information refers to the information accumulated by the AI ​​agent during its dialogue with the user. Dialogue information can also refer to the information obtained by parsing and accumulating the user's vocalizations or input to the AI ​​agent. Furthermore, dialogue information can be data calculated as feature quantities by analyzing the aforementioned vocalizations or inputs.

[0023] When the first user moves more than a predetermined distance from the device that provides the interface between the first AI agent and the first user, the control unit switches the executing AI agent from the first AI agent corresponding to the first user to the second AI agent corresponding to the second user, and transfers the dialogue information accumulated in the first AI agent to the second AI agent.

[0024] In this way, the information processing apparatus disclosed herein can share the conversation history of each account among multiple accounts in a dialogue with a user implemented by natural language processing.

[0025] Based on the structure involved, the information processing apparatus disclosed herein is capable of appropriately managing information in dialogues with users using machine learning.

[0026] Furthermore, the information processing apparatus disclosed herein may also include a storage unit that stores a database containing information representing the relationships between the plurality of users. The control unit switches the first AI agent to a second AI agent corresponding to the second user selected from the plurality of users based on the relationship information representing the relationships between the plurality of users contained in the database.

[0027] For example, consider a scenario where multiple users are riding in a vehicle capable of conversing with an AI agent. For instance, suppose the driver, their relatives, and guests are in the vehicle. While in such cases the AI ​​agent corresponding to the driver is generally executed, it may temporarily switch to another user's AI agent when the driver leaves the vehicle. If the switched AI agent is that of a guest or other user with a less close relationship to the driver, it is possible that information that the user (guest) does not wish to be known by others, i.e., other passengers, could be disclosed through conversation with the AI ​​agent. This is inappropriate from a privacy perspective. Therefore, the information processing apparatus of this disclosure switches to an AI agent corresponding to a user (generally the driver's) relative or other user with an appropriate relationship to the user of the executing AI agent. Thus, the information processing apparatus of this disclosure can consider the relationships between users when switching to an appropriate AI agent.

[0028] The control unit may also transfer information selected from the dialogue information based on the relationship between the first user and the second user as represented by the relational information to the second AI agent.

[0029] For example, if all the dialogue information accumulated in the executing AI agent is transferred to the AI ​​agent at the changed destination, information that is not intended to be known by the user of the AI ​​agent at the changed destination may be known by the user of the AI ​​agent at the changed destination through the dialogue. Therefore, the information processing apparatus of this disclosure selects which information to transfer based on the relationship between the user of the executing AI agent and the user of the AI ​​agent at the changed destination. Thus, the information processing apparatus of this disclosure can transfer appropriate information based on the relationship between the user of the executing AI agent and the user corresponding to the AI ​​agent at the changed destination.

[0030] Furthermore, the control unit can also, during the execution of the first AI agent, parse the conversations conducted within a specified range from the device providing the interface between the first AI agent and the first user, thereby determining the relationship between the first user and the second user, and if it is determined that the second user and the first user are in a specific relationship, transfer the dialogue information to the second AI agent.

[0031] For example, depending on the relationship between the user of the AI ​​agent currently executing the transaction and the user of the AI ​​agent changing destinations, there may be situations where it is inappropriate to transfer accumulated information. Specifically, while it is possible to transfer information to the AI ​​agent of the driver's wife, from a privacy perspective, it is inappropriate to transfer information to the AI ​​agent of a temporarily traveling passenger. Therefore, the information processing apparatus of this disclosure transfers the accumulated information only to the AI ​​agent corresponding to the user of the currently executing AI agent who has a specific relationship with the user of the currently executing AI agent. Thus, the information processing apparatus of this disclosure is able to transfer information to the AI ​​agent corresponding to the user who has an appropriate relationship with the user of the currently executing AI agent.

[0032] Furthermore, the control unit also transfers the dialogue information, in which one or more first words with prescribed properties are replaced by one or more second words of the same word class as the first words, to the second AI agent.

[0033] Therefore, the information processing device disclosed herein can conceal information of a specified nature when handing over information to an AI agent switching destinations.

[0034] The specific embodiments of this disclosure will now be described based on the accompanying drawings. Unless otherwise specified, the hardware structures, module structures, and functional structures described in the various embodiments are not intended to limit the scope of the disclosed technology to these terms.

[0035] First Implementation Method Summary of the processing performed by the server device Reference Figure 1 Here is a brief description of the network structure including the server device according to the first embodiment. Figure 1 This diagram illustrates the network structure of the server device 100 included in the embodiments. An information processing apparatus according to one aspect of this disclosure is implemented as the server device 100.

[0036] Server device 100 is an information processing device that provides appropriate information to user 20 based on dialogue with user 20 using natural language processing such as generative AI (Artificial Intelligence). Server device 100 communicates with vehicle terminal 200 via communication network 40 through wide area network. Furthermore, users 20a and 20b are collectively referred to as user 20.

[0037] Typically, user 20 rides in vehicle 10 and speaks to an in-vehicle terminal 200 that communicates with server device 100. Server device 100 analyzes user 20's spoken voice obtained via in-vehicle terminal 200 and generates a response corresponding to the meaning of the spoken voice. This processing is performed by an AI agent corresponding to user 20, executed within server device 100. For example, server device 100 reads the user's intent from the spoken content and generates an appropriate response. Alternatively, natural language processing to generate a dialogue with user 20 can also be performed by in-vehicle terminal 200. In this case, the in-vehicle terminal 200 is equipped with artificial intelligence such as generative AI.

[0038] The server device 100 obtains information for identifying the user 20 from sensors 30 installed on the vehicle 10. Specifically, the sensors 30 send the obtained information to the communication unit (described later) of the vehicle 10, which is connected via the vehicle network 50, and the communication unit of the vehicle 10 sends the information to the server device 100 via the communication network 40. Thus, the information obtained by the sensors 30 for identifying the user 20 who is riding in the vehicle is sent to the server device 100.

[0039] For example, consider the scenario where users 20a and 20b are riding in a vehicle. If user 20a is the driver, server device 100 executes the AI ​​agent corresponding to user 20a's account and provides it to vehicle 10. Furthermore, if it is determined based on information obtained from sensor 30 that user 20a has disembarked from vehicle 10, server device 100 can also switch the executed AI agent to the one corresponding to user 20b's account.

[0040] Structure of a server device Next, the hardware and software structures of the system, including the server device 100, will be described. Figure 2 This diagram illustrates the structural elements of the server device 100 according to the first embodiment.

[0041] Server device 100 can be configured as a computer having a processor (CPU, GPU, etc.), main storage (RAM, ROM, etc.), and auxiliary storage (EPROM, hard disk drive, removable media, etc.). The auxiliary storage contains an operating system (OS), various programs, various tables, etc. By executing the programs stored therein, various functions (software modules) that conform to a predetermined purpose can be achieved, as described later. However, some or all of the functions can also be implemented as hardware modules using hardware circuits such as ASICs or FPGAs.

[0042] The server device 100 is configured to include a control unit 110, a storage unit 120, and a communication unit 130.

[0043] The control unit 110 is a computing unit that implements various functions of the server device 100 by executing a prescribed program. The control unit 110 can be implemented, for example, by a hardware processor such as a CPU. In addition, the control unit 110 may also be configured to include RAM, ROM (Read-Only Memory), cache memory, etc.

[0044] In this embodiment, the control unit 110 of the server device 100 is configured to have four software modules: a dialogue unit 111, an acquisition unit 112, a switching unit 113, and a transfer unit 114. Each software module can also be implemented by the control unit 110 (CPU, etc.) executing a program stored in the storage unit 120. Furthermore, the information processing performed by the software modules has the same meaning as the information processing performed by the control unit 110 (CPU, etc.).

[0045] The dialogue unit 111 is a processing unit for engaging in natural language-based dialogue with user 20. The dialogue unit 111 executes an AI agent corresponding to user 20, and parses user 20's speech or input text recognized by the microphone, thereby generating a response corresponding to the meaning expressed by the speech or text. The dialogue unit 111 can communicate with an external network via the communication unit 130 to obtain information. The dialogue unit 111 outputs responses via speech or text.

[0046] The acquisition unit 112 acquires data obtained by the sensor 30. The data acquired by the acquisition unit 112 can be either data used to identify the user 20 currently riding in the vehicle, or data used to infer the relationship between the users 20 currently riding in the vehicle. The acquisition unit 112 also acquires voice data acquired by the sensor 30. Furthermore, the sensor 30 can also be a camera, and the acquisition unit 112 can also acquire images captured by the camera. Additionally, the sensor 30 can also be a range sensor or other sensor capable of detecting the actions of the user 20, and the acquisition unit 112 can also acquire the result of detecting the user 20's actions of riding in the vehicle. The sensor 30 can also detect other actions of the user 20. Furthermore, the sensor 30 can also be a device for detecting the location information of the user 20, and the acquisition unit 112 can also acquire the user 20's location information. Additionally, the user 20's location information can also be obtained via GNSS (Global Navigation Satellite System) from the location information of a terminal associated with the user 20.

[0047] In addition, the acquisition unit 112 acquires information from external devices indicating the relationships between users.

[0048] The switching unit 113 performs a switch of the AI ​​agent to be executed. When the switching unit 113 is executing the AI ​​agent corresponding to user 20a (here referred to as the first AI agent), and user 20a moves more than a predetermined distance from the device providing the interface between the first AI agent and user 20a (i.e., the vehicle terminal 200), the following process is performed: The switching unit 113 switches the executing first AI agent to the AI ​​agent corresponding to user 20b (here referred to as the second AI agent).

[0049] The transfer unit 114 transfers the dialogue information, which is the information about the dialogue with user 20a accumulated in the first AI agent from the time the first AI agent is executed until the second AI agent is switched to, to the second AI agent.

[0050] The storage unit 120 is a unit for storing information and is constructed using storage media such as RAM, disk, or flash memory. The storage unit 120 stores programs executed by the control unit 110, data used by those programs, and the like.

[0051] The communication unit 130 is a wireless communication interface for connecting the server device 100 to an external network. The communication unit 130 is configured, for example, to communicate with external devices via a wireless LAN or a cellular communication network such as 3G, 4G, or 5G.

[0052] Vehicle 10 is a vehicle capable of communicating with communication network 40. Vehicle 10 can be a passenger car, a bus, a truck, or other large vehicle. Vehicle 10 includes a control unit 11, a storage unit 12, a communication unit 13, a drive unit 14, sensors 30, and an on-board terminal 200.

[0053] The control unit 11 is a computing unit that executes prescribed programs to implement various functions of the vehicle 10. The control unit 11 can be implemented using a hardware processor such as a CPU. Alternatively, the control unit 11 may be configured to include RAM, ROM, cache memory, etc. The control unit 11 executes various software programs to control the driving and other functions of the vehicle 10.

[0054] The storage unit 12 is a unit for storing information and is composed of storage media such as RAM, disk, or flash memory. The storage unit 12 stores programs executed by the control unit 11, data used by those programs, and the like.

[0055] The communication unit 13 is a wireless communication interface for connecting the vehicle 10 to an external network. The communication unit 13 is configured, for example, to communicate with external devices via a wireless LAN or a cellular communication network such as 3G, 4G, or 5G.

[0056] The drive unit 14 is a mechanism that functions to move the vehicle 10. The drive unit 14 can be configured, for example, to include a motor and converter for driving the wheels, brakes, and a steering mechanism. The drive unit 14 can also be operated by power supplied from a battery.

[0057] Sensor 30 is a sensor component installed on vehicle 10. Sensor 30 may be, for example, a temperature sensor, a radio wave ranging sensor, an infrared sensor, a touch sensor, a microphone, etc.

[0058] The vehicle terminal 200 is a terminal that accepts operations from the user 20 of the vehicle 10 regarding various devices equipped in the vehicle, or provides information to the user 20. The vehicle terminal 200 is configured to include a control unit 210, a storage unit 220, a communication unit 230, a display 240, and a speaker 250.

[0059] The vehicle terminal 200 is an example of the "device providing an interface between a first AI agent and a first user" in this disclosure.

[0060] The control unit 210 receives input from the user 20 and sends the input to the server device 100 via the communication unit 230. In addition, text or voice sent from the server device 100 representing a response to the input is output via the display 240 or the speaker 250.

[0061] The storage unit 220 is a unit for storing information and is composed of storage media such as RAM, disk, or flash memory. The storage unit 220 stores programs executed by the control unit 210, data used by those programs, and the like.

[0062] The communication unit 230 is a wireless communication interface for connecting the vehicle terminal 200 to the communication network 40. The communication unit 230 is configured, for example, to communicate with external devices via a wireless LAN or a cellular communication network such as 3G, 4G, or 5G.

[0063] Display 240 is a display for showing images and the like to provide information to a user. Display 240 may also be a liquid crystal display (LCD) or an organic EL (Electro-Luminescence) display. Furthermore, display 240 may also be implemented as a touch panel display.

[0064] Speaker 250 is a device for outputting voice information to a user. Speaker 250 has at least a built-in voice coil, magnet, and cone. Speaker 250 can also be a full-range speaker. Alternatively, speaker 250 can be installed in vehicle 10 as a vehicle speaker instead of on vehicle terminal 200.

[0065] in addition, Figure 2 The structure shown is an example; all or part of the functions illustrated can also be executed using specially designed circuitry. Furthermore, programs can be stored or executed using a combination of main and auxiliary storage devices, other than those shown in the diagram.

[0066] Server device processing Next, the specific details of the processing performed by the server device 100 according to one aspect of this disclosure will be described.

[0067] Here, the switching process of the AI ​​agent implemented by the server device 100 will be explained. Figure 3This is a flowchart of the processing performed by the server device 100 according to the first embodiment. In this embodiment, an AI agent corresponding to the driver (user 20a) is executed while the vehicle 10 is in motion. Figure 3 The processing described herein is performed when vehicle 10 begins to move. Alternatively, Figure 3 The processing described herein can also be executed when passengers are getting on or off in vehicle 10, or when a request is made to switch accounts of the AI ​​agent being executed.

[0068] First, in step S10, the acquisition unit 112 identifies each of the plurality of users 20 and determines the relationship between the plurality of users 20. For example, the acquisition unit 112 acquires an image from the vehicle-mounted camera included in the sensor 30 and identifies each user 20 by analyzing the image. Alternatively, the acquisition unit 112 may communicate with portable terminals that have established associations with the plurality of users 20 respectively via the communication unit 230 of the vehicle-mounted terminal 200 and acquire information for identifying each user 20 respectively. Then, the acquisition unit 112 may also identify the plurality of users 20 respectively based on the information acquired from the portable terminals.

[0069] Furthermore, in this step, the acquisition unit 112 acquires information indicating the relationships between the users 20 (hereinafter referred to as relationship information). The acquisition unit 112 communicates with an external device via the communication network 40 through the communication unit 130 and refers to a database containing relationship information stored in that device. The external device can be a server device installed in a designated location or a portable terminal that has been associated with any of the users 20. Then, the acquisition unit 112 determines the relationships between the identified users 20 based on the relationship information.

[0070] Next, in step S11, the dialogue unit 111 executes the AI ​​agent (first AI agent) corresponding to user 20a. Here, user 20a is an example of a "first user". The dialogue unit 111 conducts dialogue with user 20a, etc., through the first AI agent. Specifically, the dialogue unit 111 parses the voices of user 20a or passengers in the vehicle 10 or the input content input to the vehicle terminal 200 through natural language processing, and outputs the responses generated by the generating AI to these contents via the speaker 250 or the display 240.

[0071] Next, in step S12, the switching unit 113 determines whether the user 20a has moved more than a predetermined distance from the vehicle terminal 200. The switching unit 113 determines whether the user 20a has moved more than a predetermined distance from the vehicle terminal 200 based on the data acquired by the sensor 30 obtained by the acquisition unit 112.

[0072] Furthermore, the switching unit 113 can also identify the ID of the portable terminal associated with the user 20a present in the vehicle 10. Then, the switching unit 113 can determine that the user 20a has moved more than a predetermined distance from the vehicle terminal 200 when the signal strength of the portable terminal with that ID acquired by the acquisition unit 112 within the vehicle 10 is below a predetermined value. Furthermore, the switching unit 113 can also analyze the image of the vehicle-mounted camera acquired by the acquisition unit 112 to identify the image of the user 20a. Then, the switching unit 113 can determine whether the user 20a is present in the vehicle, and if it determines that the user 20a is not present in the vehicle, it determines that the user 20a has moved more than a predetermined distance from the vehicle terminal 200.

[0073] This step becomes a positive determination if the switching unit 113 determines that the user 20a has left the vehicle terminal 200 by more than a specified distance.

[0074] If the determination is positive in step S12, the process proceeds to step S13.

[0075] If the decision is negative in step S12, the process is transferred back to step S12.

[0076] When the process moves to step S13, the switching unit 113 selects user 20b based on the relationship information obtained in step S10 and a pre-set selection criterion. Here, user 20b is an example of a "second user". The switching unit 113 selects the person (user 20b in this embodiment) to become the second user based on the relationship with the first user and the pre-set selection criterion. Here, the selection criterion can also be a criterion that ranks the relationships between the person who may become the second user and the user 20a who is the first user. For example, the selection criterion can also be a criterion that specifies "selecting the second user in the order of wife, child, friend".

[0077] For example, the switching unit 113 may select a user who has a relationship with user 20a such as "wife" as the second user. The switching unit 113 may also select a user who has a relationship with user 20a such as "relative" as the second user. The switching unit 113 may also select a user that user 20a has pre-specified as the switching destination as the second user.

[0078] Next, in step S14, the transfer unit 114 selects information from the information accumulated in the dialogue conducted by the dialogue unit 111 using the first AI agent, based on relational information, to be transferred to the AI ​​agent corresponding to the second user (the second AI agent). For example, the transfer unit 114 selects information from the accumulated information that is unrelated to user 20a's privacy or that has a low necessity for anonymization, based on the relationship between user 20b and user 20a. Specifically, if user 20b is user 20a's wife, the transfer unit 114 does not select business-related information from the accumulated information, but selects information related to kinship or leisure.

[0079] Next, in step S15, the transfer unit 114 transfers the information selected in step S14 to the second AI agent. For the transfer unit 114, the information provided is the learning data of the generative AI used by the second AI agent in the dialogue, which is accumulated in the generative AI used by the first AI agent. Furthermore, the accumulated information can be data representing the voice or input of the user 20a, etc., or data output from the intermediate layer of the generative AI being used in the first AI agent.

[0080] Next, in step S16, the dialogue unit 111 executes the second AI agent. The dialogue unit 111 conducts dialogue with the user 20b, etc., through the second AI agent.

[0081] As described above, in this embodiment, the server device 100 appropriately transfers information when switching from an AI agent corresponding to user 20a to an AI agent corresponding to user 20b. Therefore, the server device 100 is able to appropriately manage information during conversations with users using machine learning.

[0082] Variations of the first embodiment In the first embodiment, the server device 100 does not process the selected information but transfers it directly as is. However, even information that is determined to be non-transferable can sometimes be transferred by processing a portion of the information. Therefore, in this variation, in order to expand the range of transferable information, a portion of information that would otherwise be non-transferable is processed. Figure 4 This is a flowchart of the word substitution process performed by the server device 100, which is a variation of the first embodiment. Figure 4 The processing described herein is to Figure 3 Step S15 specifies the processing procedure.

[0083] First, in step S20, the transfer unit 114 determines whether the dialogue information contains one or more words with a predetermined nature (here referred to as the first word). For example, a word with a predetermined nature is a word that expresses an expression related to privacy. A word with a predetermined nature may be personal information, such as name, date of birth, personal identification number, travel history, or places visited in the past.

[0084] The transfer unit 114 can also use methods such as natural language processing to parse the content of the dialogue between the dialogue unit 111 and the user 20a, thereby determining whether the first word is contained. This step becomes an affirmative determination if the transfer unit 114 determines that the dialogue information contains the first word.

[0085] If the determination in this step is positive, the process proceeds to step S21.

[0086] If the decision is negative in this step, the process proceeds to step S23.

[0087] When the process moves to step S21, the transfer unit 114 replaces the first word in the dialogue information with a second word, which is a word of the same part of speech as the first word but different from it. For example, the transfer unit 114 replaces a word representing specific personal information with a word of the same part of speech as the word representing specific personal information, but whose meaning of the specific personal information is uncertain. Specifically, the transfer unit 114 replaces the specific store name "〇〇 store" that user 20a visited along the way with other words of the same part of speech (in this case, a noun) such as "food store" that do not represent a specific store name.

[0088] Next, in step S22, the transfer unit 114 transfers the dialogue information in step S21, in which the first word was replaced with the second word, to the second AI agent. Alternatively, the transferred information may be information in the data representing the user 20a's speech or input itself, in which the first word was replaced with the second word. Alternatively, the transferred information may be data that is input into the generating AI being used in the first AI agent and output, in which the information representing the user 20a's speech or input itself was replaced with the second word. Furthermore, the transferred information may be data output from the intermediate layer of the generating AI.

[0089] Second Implementation Method In the first embodiment, the server device 100 selects a second AI agent as the switching destination based on relational information obtained from an external device. However, the server device 100 may also independently determine the relational relationship between the user corresponding to the AI ​​agent to be switched and the user corresponding to the AI ​​agent currently in operation, and select the AI ​​agent to be switched. Therefore, in the second embodiment, the server device 100 determines the relational relationship between the user corresponding to the AI ​​agent to be switched and the user corresponding to the AI ​​agent currently in operation.

[0090] Figure 5 A flowchart illustrating the processes performed by the server device 100 according to the second embodiment. Figure 5 In, for and in Figure 3 The processing annotation described in the text is the same as the processing annotation. Figure 3 Same symbol. In Figure 5 In China, only for... Figure 3 The different processing methods will be explained.

[0091] In step S30, the acquisition unit 112 identifies each of the plurality of users 20. The acquisition unit 112 acquires an image from the vehicle-mounted camera included in the sensor 30 and identifies each user 20 by analyzing the image. Alternatively, the acquisition unit 112 may communicate with portable terminals that are associated with the plurality of users 20 respectively via the communication unit 230 of the vehicle-mounted terminal 200, and acquire information for identifying each user 20 respectively. Then, the acquisition unit 112 may also identify the plurality of users 20 respectively based on the information acquired from the portable terminals.

[0092] Following step S12, in step S31, the switching unit 113 parses the conversations between multiple users within a specified range of the vehicle terminal 200 and determines the relationship between user 20a and other users. For example, if the switching unit 113 determines that user 20a and other users have a specific relationship if the conversation contains specific words, such as spouses or relatives.

[0093] For example, the switching unit 113 can determine that the users are a couple or lovers if the words used to address each other in conversations between user 20a and other users contain the personal names of user 20a and other users. Furthermore, the switching unit 113 can determine that the users are relatives if the words used to address at least one of the users in conversations between user 20a and other users contain words representing family roles such as "father" or "mother".

[0094] Next, in step S32, the switching unit 113 selects user 20b (the second user) based on the relationship between user 20a and other users determined in step S31. For example, the switching unit 113 selects a user who is a relative of user 20a from among the other users as the second user. The switching unit 113 may also select a user whose relationship with user 20a is considered the closest among multiple other users as the second user. For example, if there are no relatives of user 20a among the other users, a friend of user 20a may be selected as the second user.

[0095] Next, in step S33, the switching unit 113 determines whether the user 20b selected in step S32 has a specific relationship with user 20a. For example, although the switching unit 113 selected user 20b as a relative of user 20a in step S32, it further determines whether the selected user 20b is user 20a's wife. A specific relationship includes, for example, husband, wife, mother, father, etc. Alternatively, the switching unit 113 may also determine whether the user 20b selected in step S32 has a pre-defined relationship.

[0096] This step becomes an affirmative determination if the switching unit 113 determines that user 20b and user 20a have a specific relationship.

[0097] If the determination in this step is positive, the process proceeds to step S34.

[0098] If the decision is negative in this step, the process proceeds to step S16.

[0099] If the process is transferred to step S34, the transfer unit 114 will transfer the information accumulated in the dialogue conducted by the dialogue unit 111 using the first AI agent to the AI ​​agent (second agent) corresponding to the user 20b.

[0100] As described above, the server device 100 in this embodiment determines the AI ​​agent for switching destinations by parsing the conversations between multiple users and determining the relationships between them. At this time, information is appropriately transferred based on the determined relationships. Therefore, even when there are users who are not known in advance, the server device 100 can appropriately manage information in conversations with users conducted using machine learning.

[0101] Other variations The above-described embodiments are merely examples, and this disclosure can be implemented with appropriate modifications without departing from its spirit. For example, the processes or units described in this disclosure can be freely combined and implemented as long as they do not create technical contradictions.

[0102] Furthermore, although this embodiment describes an example of a server device 100 as the information processing apparatus involved in this disclosure, the processing performed by the information processing apparatus involved in this disclosure in the first embodiment, variations of the first embodiment, and the second embodiment can also be performed by the vehicle terminal 200. In the above cases, the acquisition of relational information or the determination of the relationship between users, the execution of the first AI agent, the selection of the second AI agent, and the switching to the second AI agent are performed by the vehicle terminal 200. In addition, the vehicle terminal 200 also performs the selection of information to be transferred to the second AI agent and the transfer of the selected information.

[0103] Furthermore, when the vehicle terminal 200 performs the processing performed in the first embodiment, a variation of the first embodiment, and the second embodiment, the vehicle terminal 200 obtains relational information from an external device through communication via the communication network 40.

[0104] Then, the vehicle terminal 200 identifies the user 20a based on data obtained from sensors 30, etc., and executes a first AI agent corresponding to the user 20a. In addition, if the vehicle terminal 200 determines that the user 20a has moved more than a predetermined distance from the vehicle terminal 200, it performs the following actions: The vehicle terminal 200 selects a second AI agent to replace the first AI agent based on relational information. Then, the information selected based on the relational information is transferred to the second AI agent.

[0105] Additionally, some of the processing in the vehicle terminal 200, such as dialogue generation by the AI, can be performed by the server device 100, or by the vehicle terminal 200 itself, as well as other processing described above.

[0106] Furthermore, although in the first embodiment, the transfer unit 114 performs the processing of step S14, the transfer unit 114 may also omit the processing of step S14 and transfer all the accumulated information to the second AI agent. Additionally, the transfer unit 114 may transfer the accumulated information to the second AI agent only when user 20b and user 20a have a specific relationship. In this case, the transferred information may be all the accumulated information or the information selected in the same way as in step S14.

[0107] Furthermore, in the second embodiment, the transfer unit 114 transfers all the accumulated information to the second agent when user 20b and user 20a have a specific relationship. However, the transfer unit 114 may also select and transfer only a portion of the accumulated information based on the relationship between user 20a and user 20b.

[0108] This disclosure can also be implemented by supplying a computer program containing the functions described in the above embodiments to a computer, which then reads and executes the program using one or more processors. Such a computer program can be provided to the computer either via a non-transitory computer-readable storage medium connectable to the computer's system bus or via a network. Non-transitory computer-readable storage media include, for example, any type of disk (floppy disk, registered trademark, hard disk drive (HDD), etc.), optical disk (CD-ROM, DVD, Blu-ray disc, etc.), read-only memory (ROM), random access memory (RAM), EPROM, EEPROM, magnetic cards, flash memory, optical cards, and any type of medium suitable for storing electronic instructions.

Claims

1. An information processing device capable of executing an AI agent, said AI agent being a virtual agent capable of engaging in dialogue with a user using natural language, wherein, It has a control unit that performs the following processing: While executing a first AI agent corresponding to a first user among the plurality of users, if the first user moves more than a predetermined distance from the device providing the interface between the first AI agent and the first user, the executing first AI agent is switched to a second AI agent corresponding to a second user among the plurality of users who is different from the first user; and The dialogue information is transferred to the second AI agent, which is information about the dialogue with the first user that has been accumulated in the first AI agent from the time the first AI agent is executed until the switch to the second AI agent.

2. The information processing apparatus as described in claim 1, wherein, It also includes a storage unit that stores a database containing information representing the relationships between the multiple users. The control unit performs the following processing: based on the information representing the relationships between the multiple users contained in the database, i.e., relational information, the first AI agent is switched to the second AI agent corresponding to the second user selected from the multiple users.

3. The information processing apparatus as described in claim 2, wherein, The control unit performs the following processing: transferring information selected from the dialogue information based on the relationship between the first user and the second user as represented by the relational information to the second AI agent.

4. The information processing apparatus as described in claim 1, wherein, The control unit performs the following processing: during the execution of the first AI agent, it parses the conversations conducted within a specified range from the device providing the interface between the first AI agent and the first user, thereby determining the relationship between the first user and the second user, and if it is determined that the second user and the first user have a specific relationship, it transfers the dialogue information to the second AI agent.

5. The information processing apparatus as described in claim 1, wherein, The control unit further performs the following processing: transferring the dialogue information in which one or more first words with a prescribed nature are replaced by one or more second words of the same word class as the first words to the second AI agent.

Images