Program, computer, system, and information processing method

Abstract

To provide a program, a computer, a system, and an information processing method that allows chatbots to vary their responses.SOLUTION: In an information processing system 1, a computer (server 10) executes a program to function as acquisition means 18, selection means 20, and transmission means 26. The acquisition means 18 acquires dialogue information in a chatbot that is input and output at a user terminal 34. The selection means 20 selects one mode from one or more modes on the basis of at least one of the acquired dialogue information and user information, which is information about the user operating the user terminal 34. The transmission means 26 transmits output instruction information to the user terminal 34 to cause a chat interface to output a response corresponding to the mode selected by the selection means 20.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a program, a computer, a system, and an information processing method. [Background technology]

[0002] In recent years, there has been an increase in AI chatbot services in which AI (artificial intelligence) answers questions from users. For example, Patent Document 1 discloses a dialogue system that uses AI technology to answer questions from users in customer support situations. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6218057 Summary of the Invention [Problem to be solved by the invention]

[0004] Previously, chatbots had the problem of not being able to vary the content of the robot's answers.

[0005] The present disclosure has been made in consideration of these points, and aims to provide a program, computer, system, and information processing method that can vary the answers given by a chatbot. [Means for solving the problem]

[0006] The program of the present disclosure is A program that causes a computer to function as an acquisition means, a selection means, and a transmission means, The acquisition means acquires dialogue information in the chatbot inputted and outputted to the user terminal, the selection means selects one mode from one or a plurality of modes based on at least one of the acquired dialogue information and user information that is information about a user operating the user terminal; The transmitting means transmits to the user terminal output instruction information for causing a chat interface to output a response corresponding to the mode selected by the selecting means.

[0007] The program of the present disclosure is causing the computer to further function as a setting means, an instruction sentence generating means, and an answer obtaining means; The setting means sets a mode including information of a setting statement, the selection means selects one mode from one or more modes preset by the setting means, the instruction statement generating means generates an instruction statement based on the setting statement corresponding to at least the mode selected by the selecting means; the answer acquisition means acquires information about an answer to the instruction statement generated by the instruction statement generation means, The transmitting means may transmit output instruction information to the user terminal to cause the chat interface to output the answer acquired by the answer acquiring means as an answer corresponding to the mode selected by the selecting means.

[0008] In the program of the present disclosure, The selection means may further select one mode from one or more modes based on the immediately preceding mode.

[0009] In the program of the present disclosure, The instruction sentence generation means may generate the instruction sentence based on the setting sentence corresponding to the mode selected by the selection means and the user's input content in the chatbot sent from the user terminal.

[0010] In the program of the present disclosure, The mode information set by the setting means further includes trigger information, The selection means may select a mode by comparing at least one of the dialogue information and the user information with the trigger information included in each of one or more modes preset by the setting means.

[0011] In the program of the present disclosure, The selection means may select the normal mode when the comparison result indicates that the trigger for any mode is not satisfied.

[0012] In the program of the present disclosure, The selection means may input a mode selection instruction sentence to a large-scale language model, which causes a mode to be selected based on at least one of the dialogue information and the user information, and select a mode based on response information of the mode selected by the large-scale language model.

[0013] In the program of the present disclosure, The answer acquisition means may acquire information about the answer by inputting the instruction sentence generated by the instruction sentence generation means into a large-scale language model and receiving from the large-scale language model the answer to the instruction sentence calculated by the large-scale language model.

[0014] The program of the present disclosure is causing the computer to further function as a registration means; The registration means registers a character to be a chat partner of the user, The setting means may set a mode for each character registered by the registration means.

[0015] In the program of the present disclosure, The character registered by the registration means is associated with retained knowledge information, The instruction generating means may generate the instruction by also referring to the retained knowledge information associated with a character with which the user will chat.

[0016] In the program of the present disclosure, The mode information set by the setting means further includes retained knowledge information, The instruction sentence generating means may generate the instruction sentence by also referring to the retained knowledge information corresponding to the mode selected by the selecting means.

[0017] In the program of the present disclosure, The user information is associated with retained knowledge information, The instruction sentence generating means may generate the instruction sentence by also referring to the retained knowledge information associated with the user information.

[0018] In the program of the present disclosure, the retained knowledge information includes text, image, video, or audio information; the answer acquired by the answer acquisition means includes displaying text, an image, or a video as the retained knowledge information on the user terminal, or emitting sound as the retained knowledge information from the user terminal; The transmitting means may transmit to the user terminal, based on the answer acquired by the answer acquiring means, display instruction information for displaying text, an image or a video on the user terminal, or voice generation instruction information for emitting voice from the user terminal.

[0019] The computer of the present disclosure includes: A computer that functions as an acquisition means, a selection means, and a transmission means by executing a program, The acquisition means acquires dialogue information in the chatbot inputted and outputted to the user terminal, the selection means selects one mode from one or a plurality of modes based on at least one of the acquired dialogue information and user information that is information about a user operating the user terminal; The transmitting means transmits to the user terminal output instruction information for causing a chat interface to output a response corresponding to the mode selected by the selecting means.

[0020] The system of the present disclosure comprises: A user terminal; a computer communicatively connected to the user terminal for executing a dialogue with a chatbot on the user terminal; Equipped with The computer functions as an acquisition means, a selection means, and a transmission means by executing a program; The acquisition means acquires dialogue information in the chatbot inputted and outputted to the user terminal, the selection means selects one mode from one or a plurality of modes based on at least one of the acquired dialogue information and user information that is information about a user operating the user terminal; The transmitting means transmits to the user terminal output instruction information for causing a chat interface to output a response corresponding to the mode selected by the selecting means.

[0021] The information processing method of the present disclosure includes: An information processing method performed by a computer having a control unit, The control unit acquires dialogue information in the chatbot inputted and outputted by the user terminal; a step of selecting one mode from one or a plurality of modes by the control unit based on at least one of the acquired dialogue information and user information which is information about a user operating the user terminal; a step of the control unit transmitting output instruction information to the user terminal for causing a chat interface to output a response corresponding to the selected mode; The present invention is characterized by the following features. [Effects of the Invention]

[0022] According to the program, computer, system, and information processing method disclosed herein, it is possible to vary the responses from a chatbot. [Brief explanation of the drawings]

[0023] [Figure 1] FIG. 1 is a diagram schematically illustrating a configuration of an exemplary information processing system according to an embodiment of the present disclosure. [Figure 2] FIG. 10 is a diagram illustrating an example of an administrator screen on which mode information is input. [Figure 3] 1 is a diagram illustrating an outline of the relationship between pieces of information in an information processing system and an information processing method according to an embodiment of the present disclosure. [Figure 4] FIG. 10 is a diagram illustrating an example of the flow of information processing when registering character settings in an information processing system and an information processing method according to an embodiment of the present disclosure. [Figure 5] 1 is a diagram illustrating an example of a flow of information processing during a chat by a user in an information processing system and an information processing method according to an embodiment of the present disclosure. [Figure 6] 10A and 10B are diagrams illustrating an example of the flow of information processing when a mode is set in an information processing system and an information processing method according to an embodiment of the present disclosure. [Figure 7] FIG. 1 is a diagram illustrating an example of the flow of information processing in an information processing method according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0024] 1 to 7 are diagrams illustrating an information processing system 1 and an information processing method according to the present embodiment.

[0025] [Information Processing System 1] The information processing system 1 according to the present disclosure shown in FIG. 1 includes a server (computer) 10, a user terminal 34, an administrator terminal 36, and a large-scale language model server (computing unit) 32. The server 10 is a management server that can be used via a communication network (not shown) such as the Internet. The user terminal 34, the administrator terminal 36, and the like are communicatively connected to the server 10 via the communication network. With this information processing system 1 according to the present disclosure, it is possible to vary the answers from a chatbot even when the same question is asked. A "chatbot" refers to a program or application that interacts with people using text, voice, or the like, and preferably uses a large-scale language model to interact. Each component of the information processing system 1 will be described below.

[0026] <Server (Computer) 10> The server (computer) 10 is used to execute a dialogue with a chatbot on a user terminal 34. The configuration of the server (computer) 10 of the present disclosure will be described with reference to FIG. 1. FIG. 1 is a schematic diagram illustrating the configuration of the server 10 of the present disclosure. The server 10 of this embodiment is composed of an industrial computer, a personal computer, or the like, and as shown in FIG. 1, includes a control unit 12, a storage unit 28, and a communication unit 30. Note that the server 10 may be composed of multiple computers.

[0027] (Control unit 12) The control unit 12 is configured with a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and the like, and controls the operation of the server 10. Specifically, the control unit 12 executes programs stored in a storage unit 28 (described later) to function as a registration means 14, a setting means 16, an acquisition means 18, a selection means 20, an instruction statement generation means 22, a response acquisition means 24, a transmission means 26, and the like (see FIG. 1). Note that these functions may be realized by executing one or more independent programs or applications. Furthermore, these programs and applications may be provided in a single terminal (the server 10) or may be distributed across multiple terminals (including the server 10), and in the latter case, may be connected to each other via a wired cable or a communication network. Each of these means will be described later.

[0028] (Storage unit 28) The storage unit 28 is configured, for example, with an HDD (Hard Disk Drive), RAM (Random Access Memory), ROM (Read Only Memory), SSD (Solid State Drive), etc. Furthermore, the storage unit 28 is not limited to being built into the server 10, but may be a storage medium (for example, a USB memory) that can be detachably attached to the server 10. Furthermore, instead of providing the storage unit 28, various pieces of information may be stored in other storage means (such as a cloud server).

[0029] The memory unit 28 can store programs executed by the control unit 12, as well as registration information such as accounts corresponding to the user terminal 34 operated by the user and the administrator terminal 36 operated by the character administrator, mode information, retained knowledge information, user information, dialogue information, prompts, and so on. The administrator (character) account, question information, and answer information may be stored in association with each other. This association allows for efficient data output, which is preferable. In this specification, a "character" refers to an avatar that interacts with the user in a chatbot. It refers to a real person with a unique personality, such as a celebrity, famous person, or ordinary person, or their personality, character, or personality that appears on the server 10 platform (chatbot). While a deceased person may be present, this concept differs from stereotypical characters or personas such as cram school teachers, flight attendants, idols, or announcers. A "user" refers to a person who interacts with a character (avatar) using the server 10 platform. An "administrator" refers to the character himself or the person who manages the character (including an organization or corporation).

[0030] "Mode information" refers to information about how a character interacts in each mode, and includes information such as mode basic settings, mode message settings, mode voice settings, and dialogue output settings, as shown in FIG. 2. In this specification, "mode" refers to a special output format that changes the character's responses compared to normal mode (the basic setting, when no mode is selected), while maintaining the identity of the character with whom the user chats. This change, i.e., the "type of mode," is not particularly limited, but examples include emotions, speech patterns (e.g., accents, baby talk, casual speech, romantically-spoken speech, and impersonations), and topic specificity (e.g., quizzes, local topics, private information, schedules, recommended information, and advice) (including the ranking and strength of each output format). Multiple modes may be selected simultaneously, or only one mode may be selected at a time.

[0031] As part of the "mode basic settings," the administrator can set the "mode name," "mode description," and "trigger settings" (Figure 2). The "mode description" is a description of the mode. In the example shown in Figure 2, the game mode is described as "playing rock-paper-scissors, a word chain, or a word association game with the user." As another example, in the anger mode, for example, the user may be prompted to limit the content of certain statements to specific words. In this way, a mode description can be added to the prompt to generate answer information corresponding to the selected mode. To accommodate the selected mode, a large-scale language model may be trained in advance, or a separate program corresponding to each mode may be executed. The "trigger settings (trigger information)" describe the conditions for switching from normal mode to each mode (Figures 2 and 3(A)). In the illustrated game mode, it is preferable to set the mode to switch to game mode not only when the user says "game" but also when the user says "game + positive words." This is to prevent the game mode from starting in response to a statement such as "I hate games."

[0032] As a "mode message setting," the administrator can set a "setting sentence" so that the character (chatbot) will respond according to the set mode. For example, in the case of angry mode, an instruction sentence such as "Please speak in an angry tone" can be generated. As will be described later, an instruction sentence can be generated for the large-scale language model server 32 based on this setting sentence and the input content from the user.

[0033] When having a chatbot converse (output) by voice, the administrator may be able to select a voice model appropriate for the mode as a "mode voice setting." The voice output (tone of voice, intonation) in each mode may include recording and editing to match the character, voice quality conversion, etc. The administrator may also be able to select whether the chatbot converses in one or multiple languages. In this way, the present disclosure differs from chatbots that only provide standardized answers.

[0034] The platform according to the present disclosure can be incorporated into various chat apps, social networking sites, etc., and the display method of the character (avatar) may be changed to suit the form (e.g., type, size) of the user terminal 34 (FIG. 2). That is, the administrator can set the facial expressions and movements of the character (avatar) to be output in the chat interface according to the selected mode in the "Dialogue Output Settings." For example, in angry mode, the avatar may have an angry expression or pose. In this way, the administrator may set the desired sounds (sound effects, background music), images, and videos to be output in a given mode in addition to dialogue information in the form of text or voice.

[0035] "Retained knowledge information" is information about a character that can be referenced when the character (chatbot) converses with an individual user. Examples include, but are not limited to, the character's hometown, publicly available information, quiz topics, topics, and additional information related to each mode (taboo phrases, funny jokes, etc.). Types of retained knowledge information may include text, images, videos, audio, etc. Each piece of retained knowledge information may be linked not only to the character, but also to user information and the set mode (Figure 3(B)). In this way, the character's answers can be further varied each time the mode is switched to, allowing the user to enjoy conversations with the character.

[0036] "User information" is information about a user that can be referenced by a character (chatbot) when interacting with an individual user. Examples include, but are not limited to, information that can be used as conversation material, such as age, gender, family structure, occupation, hobbies and interests, preferred mode names, schedule, available chat times, and purchasing information. Modes can be selected based on such information, allowing the character to engage in conversation tailored to each user (e.g., it can be used in a toddler mode, a lover mode, etc.). User information can be extracted from a conversation, but may also be registered by the user at any time without a conversation. In some embodiments, both "user information" and "retained knowledge information" may be used only in modes other than normal mode.

[0037] (Communication unit 30) The communication unit 30 connects the control unit 12 to an external device (e.g., a user terminal 34) wirelessly or via a wired connection so that the control unit 12 can communicate with the external device. The control unit 12 can send and receive signals to and from the external device via the communication unit 30.

[0038] (others) In addition, the server 10 may be provided with an operation unit (input unit) such as a keyboard that enables an administrator of the server 10 to issue various commands to the control unit 12, and a display unit such as a monitor that displays various screens in response to display command signals from the control unit 12. In one embodiment, a display operation unit such as a touch panel that integrates the operation unit and display unit may be used. Also, a calculation unit corresponding to the large-scale language model server 32 may be provided within the server 10.

[0039] <Large-scale language model server 32> The large-scale language model server 32 is a server such as an industrial computer. The large-scale language model server 32 preferably includes a learning model, particularly a large-scale language model (hereinafter, the language model in the large-scale language model server 32 is also referred to as LLM32), which inputs a question sentence and outputs a response sentence, an acoustic model, a WaveNet model, or the like. The learning method is not particularly limited, and methods described in Patent Application No. 2023-197471 by the present applicant may be used. The same applies to methods for generating and learning dialogue information within a chatbot mode. Examples of LLM32 include BERT, XLNeT, T5, and GPT, with multimodal GPT-4 being particularly preferred. The LLM32 calculates answer information in response to question information from a user terminal 34 on a website (platform) on which a program by the server 10 according to the present disclosure is executed. Additionally, the LLM32 may accept a "mode selection instruction" that selects a mode based on at least one of dialogue information and user information, and select a mode instead of the server 10 (selection means 20). Furthermore, the LLM 32 may calculate answer information based on an instruction (i.e., a mode) from the server 10 (instruction generation means 22). These dialogue functions, mode selection functions, and in-mode dialogue functions may also be realized by multiple learning models (such as the LLM 32). The LLM 32 may also have a function to convert the format of the answer information (text, audio, video, etc.). In this case, it can perform voice recognition, natural language processing, etc., analyze the content of the question, and then calculate and output audio information or text information of an answer that matches the content of the question. Other methods may also be adopted.

[0040] <User terminal 34 and administrator terminal 36> The user terminal 34 and the administrator terminal 36 are terminals operated by a user and an administrator, respectively. The user terminal 34 and the administrator terminal 36 may include a personal computer, a tablet terminal, a mobile terminal, etc., but are not particularly limited as long as they are capable of transmitting and receiving information to and from the server 10 and inputting and outputting information. The user terminal 34 and the administrator terminal 36 do not need to be the same terminal; the same user (user, administrator) may transmit and receive information using different terminals at different times. On a website (platform) on which the program of the present disclosure is executed, each user and each administrator may use different terminals, as long as each terminal is identified by the same registration information (account information). In the present invention, interaction with a character is not limited to being performed on a website, but may be performed by a so-called app (application software) installed on the user terminal 34. Furthermore, the program of the present disclosure may be executed by both a website and an app.

[0041] (Details of control unit 12) (Method of registration 14) The registration means 14 registers characters that users can chat with. Furthermore, the registration means 14 can also register administrators and users of characters. In this specification, "registration" refers to, but is not limited to, assigning an identification number or the like to registrants (administrators and users) in the server 10 and setting up an account for the registrant, thereby enabling character mode management and interaction on the platform. For example, a registration method on this platform may involve a registrant expressing a predetermined intention to be registered (including, for example, consenting to the management terms and conditions, providing a name, etc., and verifying identity), after which the registrant is assigned an account and an identification number that enables interaction with the registrant. For each character registered by the registration means 14, basic mode settings, etc., as described above, may be performed and retained knowledge information may be associated with the characters.

[0042] (Setting means 16) The setting means 16 sets a mode including setting sentence information. The setting means 16 can set a mode for each character registered by the registration means 14. As described above, such mode information may further include the mode name, mode content, trigger information, retained knowledge information, etc., and the administrator can register or additionally input mode information as needed. In addition, the server 10 platform can also input settings for modes that can be commonly applied to each character.

[0043] (Acquisition method 18) The acquisition means 18 acquires dialogue information in the chatbot inputted and outputted by the user terminal 34. "Dialogue information" includes a dialogue history such as question information and answer information, and may also include voice information and text information thereof. The latest question information from the user terminal 34 can also be stored in the storage unit 28 at any time. In this specification, unless otherwise specified, "question information" and "answer information" refer to dialogue information from the user terminal 34 and the character (large-scale language model server 32 in the example of FIG. 6), respectively. For example, even if the content of dialogue information from a character is a question to the user, it is referred to as answer information for convenience.

[0044] (Selection means 20) The selection means 20 selects one mode from one or more modes based on at least one of the acquired dialogue information and user information, which is information about the user operating the user terminal 34. When a user instructs in the dialogue to switch to a specific mode, or when the user wants to interact with a character in a specific mode set in advance, the selection means 20 may select that mode.

[0045] The selection means 20 can select one mode from one or more modes preset by the setting means 16. In this way, the administrator may freely set the mode.

[0046] The selection means 20 may further select one mode from one or more modes based on the immediately preceding mode during the dialogue. For example, the quiz level may be switched in stages, such as a beginner quiz mode and an advanced quiz mode. Alternatively, multiple anger modes may be set, and the anger tone may be gradually switched to become more scathing. Alternatively, a reconciliation mode, which is the exact opposite of an anger mode, may be selected after an anger mode. Furthermore, when multiple modes can be selected, the selection means 20 may select a mode so as to continue the immediately preceding mode (i.e., select the immediately preceding mode again). In particular, when a mode is switched from a normal mode, the selection means 20 may preferentially select the immediately preceding mode so that the switched mode continues. When a language model is used to select such a mode, a prompt may be given, or the mode may be set not to change until a predetermined time or a predetermined number of questions and answers have been answered.

[0047] The selection means 20 may select a mode by comparing at least one of the dialogue information and the user information with (corresponding to) trigger information included in one or more modes preset by the setting means 16 (FIG. 3(A)). The selection means 20 may select the normal mode (including maintaining the normal mode) if the comparison result shows that the trigger of any mode is not satisfied. Note that the mode does not necessarily have to be switched via the normal mode; for example, the mode may be switched from the quiz mode to the anger mode, or may remain in the quiz mode if there is no trigger.

[0048] The selection means 20 may select a mode using a large-scale language model. For example, the selection means 20 may input a mode selection prompt, which prompts the user to select a mode based on at least one of dialogue information and user information, to a large-scale language model (e.g., LLM32), and select a mode based on response information for the mode selected by the large-scale language model. Natural language processing such as semantic analysis can also be used for the above "comparison."

[0049] (Indicative sentence generation means 22) The instruction sentence generation means 22 generates an instruction sentence based on at least a setting sentence corresponding to the mode selected by the selection means 20. Answer information corresponding to the selected mode can be acquired from a chatbot or the like by the instruction sentence generation means 22 and the answer acquisition means 24 described below. The instruction sentence here corresponds to the selected mode and is a prompt for acquiring answer information corresponding to question information from the user terminal 34, and the answer information may be acquired from the instruction sentence by various methods other than the chatbot or language model.

[0050] The instruction generating means 22 may generate an instruction by also referring to retained knowledge information associated with the character with which the user will chat. Alternatively, the instruction generating means 22 may generate an instruction by also referring to retained knowledge information corresponding to the mode selected by the selecting means 20. Alternatively, the instruction generating means 22 may generate an instruction by also referring to retained knowledge information associated with user information (FIG. 3(B)). In this way, it is possible to enrich the dialogue.

[0051] Alternatively, the instruction sentence generating means 22 may generate an instruction sentence based on a setting sentence corresponding to the mode selected by the selecting means 20 and the user's input to the chatbot (i.e., question information, which may further include a dialogue history) sent from the user terminal 34. For example, in the mimicry mode, if the user requests the character to make a cat's meowing sound, the character will be able to obtain a response in a cat-stroking voice.

[0052] (Answer acquisition means 24) The answer acquisition means 24 acquires information on an answer to the instruction statement generated by the instruction statement generation means 22 .

[0053] The answer acquisition means 24 may acquire answer information by inputting the instruction sentence generated by the instruction sentence generation means 22 into a large-scale language model (e.g., LLM 32) and receiving from the large-scale language model an answer to the instruction sentence calculated by the large-scale language model. In this way, answer information corresponding to the selected mode can be obtained by the large-scale language model (e.g., LLM 32).

[0054] (Transmission means 26) The transmitting means 26 transmits output instruction information to the user terminal 34 for causing the chat interface to output an answer corresponding to the mode selected by the selecting means 20. The transmitting means 26 can transmit output instruction information to the user terminal 34 for causing the chat interface to output the answer acquired by the answer acquiring means 24 as an answer corresponding to the mode selected by the selecting means 20. More specifically, the answer acquired by the answer acquiring means 24 may include displaying text, an image, or a video as retained knowledge information on the user terminal 34 or emitting audio as retained knowledge information from the user terminal 34. In this case, the transmitting means 26 transmits display instruction information for displaying text, an image, or a video on the user terminal 34 or audio generation instruction information for emitting audio from the user terminal 34 based on the answer acquired by the answer acquiring means 24 to the user terminal 34. The "chat interface" refers to the output unit of the user terminal 34, and may be a chat screen (display, etc.) when text, an image, or a video is output, or a speaker of the user terminal 34 when audio information is output.

[0055] [Information processing method] Next, an information processing method in the information processing system 1 described above will be described with reference to the drawings, but the information processing method according to the present disclosure is not limited to the following example. In the following description, components with the same reference numerals are the same as the components described above, and duplicated descriptions will be omitted as appropriate. Note that the processing described below is performed by the control unit 12 executing a program stored in the storage unit 28, and an example will be described in which the dialogue information is text information.

[0056] First, an example of the flow of information processing during character setting registration will be described with reference to Fig. 4. Here, first, an administrator operating administrator terminal 36 logs in with an administrator ID to a website (platform) on which the program of the present disclosure is executed. Then, display instruction information for displaying an administrator screen (e.g., Fig. 2) on the display of administrator terminal 36 is transmitted from server 10. Then, the administrator registers a character, sets a mode (registration), and inputs retained knowledge information via the administrator screen, and this information is transmitted to server 10.

[0057] Next, an example of the flow of information processing when a user chats will be described with reference to FIG. 5. First, a user operating a user terminal 34 logs in with a user ID to a website on which the program of the present disclosure is executed. Then, display instruction information for displaying a chatbot screen (text information) on the display of the user terminal 34 is transmitted from the server 10. Then, when the user inputs a question on the chatbot screen, this question information is transmitted to the server 10. Next, an instruction is generated by the server 10 (instruction generation means 22), and this instruction is transmitted to the large-scale language model server 32 (LLM 32). Then, answer information is calculated by the LLM 32, and this answer information is transmitted to the user terminal 34 via the server 10 and displayed on the chatbot screen. The above is the flow of a normal dialogue that does not involve a mode.

[0058] Next, with reference to FIG. 6, an example of the flow of information processing during mode setting (selection) will be described. Here, first, the user inputs a question on the chatbot screen. This latest dialogue information (question information, which may also include a dialogue history) is transmitted from the user terminal 34 to the server 10. A mode selection instruction sentence for selecting a mode is generated by the server 10 (selection means 20) based on this dialogue information (user information may also be referenced). This mode selection instruction sentence is transmitted to the large-scale language model server 32 (LLM 32), and the LLM 32 calculates answer information indicating which mode should be selected. Based on this answer information, the server 10 (selection means 20) sets (selects) the mode. Next, the server 10 (instruction sentence generation means 22) generates an instruction sentence (prompt) based on the setting sentence corresponding to the mode selected by the selection means 20 and the latest dialogue information transmitted from the user terminal 34. This instruction sentence is transmitted to the large-scale language model server 32 (LLM 32), and the answer information calculated by the LLM 32 is transmitted to the user terminal 34 via the server 10 and displayed on the chatbot screen. The timing of generating an instruction sentence by the selection means 20 (i.e., determining whether to select or change the mode) is not limited to the predetermined time period mentioned above, but may be performed every time dialogue information is transmitted from the user terminal 34. A single dialogue may switch from the intimate mode to the furious mode.

[0059] Next, an example of an information processing method in the information processing system 1 will be described with reference to Fig. 7. First, the server 10 (control unit 12) determines whether login information has been received from the user terminal 34 (step S10). If login information has not been received ("NO"), the server 10 (control unit 12) enters a waiting state for receiving login information from the user terminal 34.

[0060] On the other hand, when the server 10 receives login information from the user terminal 34 (step S10, "YES"), the server 10 (control unit 12) transmits display instruction information for displaying the chatbot screen on the user terminal 34 (step S20).

[0061] Next, the control unit 12 (selection means 20) generates a mode selection directive based on the user information (step S30), and then transmits the generated mode selection directive to the large-scale language model server 32, and obtains a response regarding the selected mode from the large-scale language model server 32 (step S40).

[0062] Next, the control unit 12 (selection means 20) selects a mode based on the acquired response information (step S50).

[0063] Next, the control unit 12 (acquisition means 18) determines whether dialogue information (question information) in the chatbot has been received from the user terminal 34 (step S60). If question information has not been received ("NO"), the control unit 12 enters a waiting state.

[0064] On the other hand, when question information is received from the user terminal 34 (step S60, "YES"), the control unit 12 (instruction sentence generation means 22) generates an instruction sentence based on the setting sentence corresponding to the selected mode (step S70).

[0065] Next, the control unit 12 (answer acquisition means 24) transmits the generated instruction sentence to the large-scale language model server 32, and acquires the answer to the instruction sentence calculated by the large-scale language model server 32 (step S80).

[0066] Next, the control unit 12 (transmission means 26) displays, on the user terminal 34, output instruction information for outputting the acquired answer (the answer corresponding to the selected mode) to the chat interface of the user terminal 34 (step S90).

[0067] Next, the server 10 (control unit 12) determines whether logout information has been received from the user terminal 34 (step S100). If logout information has been received from the user terminal 34 (step S100, "YES"), the process ends. On the other hand, if logout information has not been received (step S100, "NO"), the process returns to step S30, the mode is selected or maintained, and question information is received from the user terminal 34 (steps S30 to S60).

[0068] The above-described information processing method is an example, and the processing flow is not limited to the above. For example, after step S20, the control unit 12 (acquisition means 18) may acquire dialogue information in the chatbot input / output to the user terminal 34 (as in step S60), and in step S30 the control unit 12 (selection means 20) may generate a mode selection instruction sentence based on at least one of the dialogue information and the user information.

[0069] Note that manual confirmation may be performed at each step. Also, instead of steps S30 to S50 for selecting a mode based on user information, the control unit 12 (selection means 20) may select a mode input by the operator of the server 10.

[0070] In the program, computer (server 10), information processing system 1, and information processing method of this embodiment configured as described above, by executing the program, the computer (server 10) functions as an acquisition unit 18, a selection unit 20, and a transmission unit 26, as described below. Specifically, the acquisition unit 18 acquires dialogue information in the chatbot input and output from a user terminal 34. The selection unit 20 selects one mode from one or more modes based on at least one of the acquired dialogue information and user information, which is information about the user operating the user terminal 34. The transmission unit 26 transmits output instruction information to the user terminal 34 to cause the chat interface to output a response corresponding to the mode selected by the selection unit 20. According to the program, computer (server 10), information processing system 1, and information processing method of this embodiment, even if the same question is asked, the chatbot can provide a variety of responses. As a result, the user can enjoy a variety of conversations with the chatbot.

[0071] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the computer (server 10) can further function as a setting means 16, an instruction statement generation means 22, and an answer acquisition means 24. Specifically, the setting means 16 sets a mode including setting statement information. The selection means 20 selects one mode from one or more modes pre-set by the setting means 16. The instruction statement generation means 22 generates an instruction statement based on at least a setting statement corresponding to the mode selected by the selection means 20. The answer acquisition means 24 acquires information on an answer to the instruction statement generated by the instruction statement generation means 22. The transmission means 26 transmits output instruction information to the user terminal 34 for outputting the answer acquired by the answer acquisition means 24 to the chat interface as an answer corresponding to the mode selected by the selection means 20. As described above, users can enjoy modes set by each character administrator or platform (chatbot). Furthermore, the instruction statement generation means 22 and the answer acquisition means 24 can automatically acquire answer information corresponding to the selected mode.

[0072] In the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the selection means 20 may further select one mode from one or more modes based on the immediately preceding mode. As described above, the user can continue to enjoy the immediately preceding mode, or can enjoy a mode that is related to a change in answer.

[0073] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the instruction sentence generation means 22 may generate an instruction sentence based on a setting sentence corresponding to the mode selected by the selection means 20 and the user's input content in the chatbot transmitted from the user terminal 34. In this way, it is expected that more specific answer information can be obtained based on the setting sentence and the input content (question information).

[0074] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the mode information set by the setting means 16 may further include trigger information, and the selection means 20 may select a mode by comparing at least any one of the dialogue information and the user information with the trigger information included in each of one or more modes previously set by the setting means 16. In this way, the dialogue information (including the dialogue history) and the user information can be searched using the trigger information and its similar words as keywords, allowing for efficient mode selection.

[0075] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the selection means 20 may select the normal mode if the comparison result indicates that none of the mode triggers are met. If the user knows that the mode (i.e., the manner of response by the chatbot) does not change depending on the words they use, they can select their words more carefully and enjoy interacting with the chatbot.

[0076] In the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the selection means 20 may input a mode selection directive that causes a mode to be selected based on at least one of dialogue information and user information to the large-scale language model, and select a mode based on response information for the mode selected by the large-scale language model. In this way, the large-scale language model enables efficient mode selection.

[0077] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the answer acquisition means 24 may input the instruction sentence generated by the instruction sentence generation means 22 into a large-scale language model, and acquire answer information by receiving from the large-scale language model an answer to the instruction sentence calculated by the large-scale language model. In this way, the acquisition of answer information can also be made more efficient by the large-scale language model.

[0078] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, by executing the program, it can also function as registration means 14 that registers characters that users can chat with, and the setting means 16 may set a mode for each character registered by the registration means 14. In this way, users can enjoy the mode set by each character administrator or platform for each character.

[0079] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, retained knowledge information is associated with the character registered by the registration means 14, and the instruction sentence generation means 22 may generate an instruction sentence by also referring to the retained knowledge information associated with the character with whom the user will chat. By also referring to the retained knowledge information associated with the character when generating an instruction sentence, a response specific to the character can be expected.

[0080] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the mode information set by the setting means 16 further includes retained knowledge information, and the instruction sentence generation means 22 generates an instruction sentence by also referring to the retained knowledge information corresponding to the mode selected by the selection means 20. By also referring to the retained knowledge information corresponding to the selected mode when generating an instruction sentence, it is possible to expect a response that highlights the mode.

[0081] In the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, retained knowledge information is associated with user information, and the instruction sentence generation means 22 may generate an instruction sentence by also referencing the retained knowledge information associated with the user information. By also referencing the retained knowledge information associated with the user information when generating an instruction sentence, it is possible to expect a response that is targeted to the user and that suits the user's preferences.

[0082] Furthermore, in the program, computer (server 10), information processing system 1, and information processing method according to this embodiment, the retained knowledge information includes text, image, video, or audio information, and the answer acquired by the answer acquisition means 24 includes displaying the text, image, or video as the retained knowledge information on the user terminal 34 or emitting audio as the retained knowledge information from the user terminal 34, and the transmission means 26 can also transmit to the user terminal 34 display instruction information for displaying the text, image, or video on the user terminal 34 or audio generation instruction information for emitting audio from the user terminal 34, based on the answer acquired by the answer acquisition means 24. In this way, various modes of answer information are possible, and the answers can be varied using a chatbot.

[0083] The program, computer (server 10), information processing system 1, and information processing method according to the present embodiment are not limited to the above-described aspects and combinations, and various modifications can be made. [Explanation of symbols]

[0084] 1. Information Processing Systems 10 Server (computer) 12 Control Unit 14 Registration Method 16 Setting Method 18 Acquisition method 20 Selection Method 22 Instruction sentence generation means 24 Means of obtaining answers 26 Transmission Method 28 Memory section 30 Communications Department 32 Large-scale language model server (computation unit) 34 User terminal 36 Administrator terminal

Claims

1. A program that causes a computer to function as an acquisition means, a selection means, and a transmission means, The acquisition means acquires dialogue information in the chatbot inputted and outputted to the user terminal, the selection means selects one mode from one or a plurality of modes based on at least one of the acquired dialogue information and user information that is information about a user operating the user terminal; The transmitting means transmits to the user terminal output instruction information for causing a chat interface to output a response corresponding to the mode selected by the selecting means.

2. causing the computer to further function as a setting means, an instruction sentence generating means, and an answer obtaining means; The setting means sets a mode including information of a setting statement, the selection means selects one mode from one or more modes preset by the setting means, the instruction statement generating means generates an instruction statement based on the setting statement corresponding to at least the mode selected by the selecting means; the answer acquisition means acquires information about an answer to the instruction statement generated by the instruction statement generation means, 2. The program according to claim 1, wherein the transmitting means transmits output instruction information to the user terminal for causing the chat interface to output the answer acquired by the answer acquiring means as an answer corresponding to the mode selected by the selecting means.

3. 2. The program according to claim 1, wherein the selection means selects one mode from one or more modes based on the immediately preceding mode.

4. The program according to claim 2, wherein the instruction sentence generation means generates the instruction sentence based on the setting sentence corresponding to the mode selected by the selection means and the user's input content in the chatbot sent from the user terminal.

5. The mode information set by the setting means further includes trigger information, 3. The program according to claim 2, wherein the selection means selects a mode by comparing at least one of the dialogue information and the user information with the trigger information included in each of one or more modes preset by the setting means.

6. 6. The program according to claim 5, wherein said selection means selects the normal mode when the comparison result indicates that the trigger for any of the modes is not satisfied.

7. 2. The program according to claim 1, wherein the selection means inputs a mode selection directive to a large-scale language model, the directive causing a mode to be selected based on at least one of the dialogue information and the user information, and selects a mode based on response information of the mode selected by the large-scale language model.

8. 3. The program according to claim 2, wherein the answer acquisition means inputs the instruction sentence generated by the instruction sentence generation means into a large-scale language model, and acquires the answer information by receiving from the large-scale language model the answer to the instruction sentence calculated by the large-scale language model.

9. causing the computer to further function as a registration means; The registration means registers a character to be a chat partner of the user, 3. The program according to claim 2, wherein said setting means sets a mode for each character registered by said registration means.

10. The character registered by the registration means is associated with retained knowledge information, 10. The program according to claim 9, wherein the instruction generating means generates the instruction by also referring to the retained knowledge information associated with a character with which the user will chat.

11. The mode information set by the setting means further includes retained knowledge information, 3. The program according to claim 2, wherein said instruction statement generating means generates said instruction statement by also referring to said retained knowledge information corresponding to the mode selected by said selecting means.

12. The user information is associated with retained knowledge information, 3. The program according to claim 2, wherein said instruction statement generating means generates said instruction statement by also referring to said retained knowledge information associated with said user information.

13. the retained knowledge information includes text, image, video, or audio information; the answer acquired by the answer acquisition means includes displaying text, an image, or a video as the retained knowledge information on the user terminal, or emitting sound as the retained knowledge information from the user terminal; The program according to any one of claims 10 to 12, wherein the transmitting means transmits to the user terminal display instruction information for displaying text, an image, or a video on the user terminal, or voice generation instruction information for emitting voice from the user terminal, based on the answer acquired by the answer acquiring means.

14. A computer that functions as an acquisition means, a selection means, and a transmission means by executing a program, The acquisition means acquires dialogue information in the chatbot inputted and outputted to the user terminal, the selection means selects one mode from one or a plurality of modes based on at least one of the acquired dialogue information and user information that is information about a user operating the user terminal; The sending means sends to the user terminal output instruction information for causing a chat interface to output a response corresponding to the mode selected by the selecting means.

15. A user terminal; a computer communicatively connected to the user terminal for executing a dialogue with a chatbot on the user terminal; Equipped with The computer functions as an acquisition means, a selection means, and a transmission means by executing a program; The acquisition means acquires dialogue information in the chatbot inputted and outputted to the user terminal, the selection means selects one mode from one or a plurality of modes based on at least one of the acquired dialogue information and user information that is information about a user operating the user terminal; The transmission means transmits to the user terminal output instruction information for causing a chat interface to output a response corresponding to the mode selected by the selection means.

16. An information processing method performed by a computer having a control unit, The control unit acquires dialogue information in the chatbot inputted and outputted by the user terminal; a step of selecting one mode from one or a plurality of modes by the control unit based on at least one of the acquired dialogue information and user information which is information about a user operating the user terminal; a step of the control unit transmitting output instruction information to the user terminal for causing a chat interface to output a response corresponding to the selected mode; An information processing method comprising:

Images