Chat control system, chat control method and program

Abstract

To provide a chat control system, a chat control method, and a program that enable a user to obtain information with the feeling of naturally chatting with a person, and increase the possibility of obtaining an appropriate response to a prompt from an LLM. [Solution] A determination generation process executing unit 26 executes a determination generation process for each piece of necessary information that has not been generated. A determination generation control unit 28, when at least one of the ungenerated necessary information is not generated based on a chat message, causes the determination generation process executing unit 26 to execute a determination generation process for the next chat message in response to receiving the next chat message from a user terminal 14. A prompt generating unit 30, when all of the ungenerated necessary information is generated based on the chat message, generates a prompt based on the multiple pieces of necessary information.

Description

[Technical Field]

[0001] The present invention relates to a chat control system, a chat control method, and a program. [Background technology]

[0002] In recent years, techniques utilizing large-scale language models (LLMs) have been attracting attention. As an example of such techniques, Patent Literature 1 describes a technique that uses LLMs when generating automated assistant responses. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2024-521053 Summary of the Invention [Problem to be solved by the invention]

[0004] By repeatedly outputting prompts to the LLM and receiving responses from it, users can obtain information as if they were chatting naturally with a person.

[0005] However, LLMs may respond to prompts inappropriately, such as by providing vague or irrelevant answers, because they are not provided with the information necessary to respond appropriately to the prompt.

[0006] The present invention has been made in consideration of the above-mentioned problems, and one of its objectives is to provide a chat control system, chat control method, and program that allows users to obtain information as if they were chatting naturally with a person, while increasing the likelihood that they will receive an appropriate response to a prompt from the LLM. [Means for solving the problem]

[0007] (1) A chat control system according to the present invention includes a determination and generation process execution means that, in response to receiving a chat message from a terminal that is a communication partner of the chat message, determines whether or not each of a given plurality of pieces of necessary information that has not been generated can be generated based on the chat message, and executes a determination and generation process to generate the necessary information that is determined to be generatable based on the chat message; a determination and generation control means that, when at least one of the pieces of necessary information that has not been generated based on the chat message is not generated, causes the determination and generation process execution means to execute the determination and generation process for the next chat message in response to receiving the next chat message from the terminal; a prompt generation means that, when all of the pieces of necessary information that have not been generated based on the chat message are generated based on the chat message, generates a prompt based on the plurality of pieces of necessary information; a prompt output means that outputs the prompt to a large-scale language model; an answer receiving means that receives an answer from the large-scale language model in response to the prompt; and a chat message output means that outputs a chat message based on the received answer to the terminal.

[0008] (2) The chat control system described in (1) above may further include an inquiry message output means for outputting a chat message to the terminal inquiring about the necessary information that has not been generated when at least one of the necessary information that has not been generated based on the chat message is not generated.

[0009] (3) The determination generation process execution means may be a chat control system as described in (2) above, which generates the other necessary information based on a chat message received from the terminal in response to a chat message inquiring about the necessary information that is not to be generated, when it is determined that other necessary information can be generated based on the chat message.

[0010] (4) The chat control system may further include a log determination and generation means for determining whether each of the plurality of pieces of necessary information can be generated based on a log of the chat message in communication with the terminal in response to receiving a chat message from the terminal requesting an answer from the large-scale language model, and generating the necessary information determined to be generable based on the log, wherein if at least one of the plurality of pieces of necessary information cannot be generated based on the log, the determination and generation control means causes the determination and generation process execution means to execute the determination and generation process for the next chat message in response to receiving the next chat message from the terminal, and if all of the necessary information is generated based on the log, the prompt generation means generates a prompt based on the plurality of pieces of necessary information.

[0011] (5) The chat control system may be any of the above (1) to (4), in which the determination generation process execution means generates the plurality of pieces of necessary information based on a single chat message when it is determined that the plurality of pieces of necessary information can be generated based on the chat message.

[0012] (6) A chat control method according to the present invention includes the steps of: in response to receiving a chat message from a terminal that is a communication partner of the chat message, determining whether or not each of a given plurality of pieces of necessary information that has not been generated can be generated based on the chat message, and executing a determination and generation process to generate the necessary information that is determined to be generable based on the chat message; in the case where at least one of the pieces of necessary information that has not been generated based on the chat message is not generated based on the chat message, executing the determination and generation process for the next chat message in response to receiving the next chat message from the terminal; in the case where all of the pieces of necessary information that have not been generated based on the chat message are generated based on the chat message, generating a prompt based on the plurality of pieces of necessary information; outputting the prompt to a large-scale language model; accepting a response from the large-scale language model to the prompt; and outputting a chat message based on the accepted response to the terminal.

[0013] (7) A program according to the present invention causes a computer to execute the following steps: in response to receiving a chat message from a terminal that is the communication partner of the chat message, determining whether or not each of a given plurality of pieces of necessary information that has not been generated can be generated based on the chat message, and executing a determination and generation process to generate the necessary information that is determined to be generable based on the chat message; in the case where at least one of the pieces of necessary information that has not been generated based on the chat message is not generated, executing the determination and generation process for the next chat message in response to receiving the next chat message from the terminal; in the case where all of the pieces of necessary information that have not been generated based on the chat message are generated based on the chat message, generating a prompt based on the plurality of pieces of necessary information; outputting the prompt to a large-scale language model; accepting a response from the large-scale language model to the prompt; and outputting a chat message based on the accepted response to the terminal. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a diagram illustrating an example of a configuration of a chat system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a functional block diagram showing an example of functions of the intermediary agent system according to an embodiment of the present invention. [Figure 3] FIG. 10 is a diagram illustrating an example of necessary information. [Figure 4] FIG. 10 is a diagram illustrating an example of necessary information. [Figure 5] FIG. 10 is a diagram illustrating an example of necessary information management data. [Figure 6A] FIG. 2 is a flowchart showing an example of the flow of processing performed in the intermediary agent system according to one embodiment of the present invention. [Figure 6B] FIG. 2 is a flowchart showing an example of the flow of processing performed in the intermediary agent system according to one embodiment of the present invention. [Figure 6C] FIG. 2 is a flowchart showing an example of the flow of processing performed in the intermediary agent system according to one embodiment of the present invention. [Figure 6D] FIG. 2 is a flowchart showing an example of the flow of processing performed in the intermediary agent system according to one embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0015] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

[0016] Fig. 1 is a diagram showing an example of the configuration of a chat system 1 according to this embodiment. As shown in Fig. 1, the chat system 1 according to this embodiment includes a mediation agent system 10, a processing agent system 12, and a user terminal 14. The mediation agent system 10, the processing agent system 12, and the user terminal 14 are connected to a computer network 16 such as the Internet.

[0017] The facilitating agent system 10 may be configured, for example, by a computer such as a server computer. The facilitating agent system 10 may include a processor 10a such as a CPU, a storage unit 10b such as a memory or a hard disk drive, a communication unit 10c such as a network interface card, etc. The facilitating agent system 10 may include multiple computers or may include a single computer.

[0018] The processing agent system 12 may be configured, for example, by a server computer including a CPU, a GPU, a memory, a communication interface, etc. The processing agent system 12 may include multiple computers or may include a single computer. As shown in Fig. 1, the processing agent system 12 according to this embodiment is equipped with a trained large-scale language model (LLM) 12a.

[0019] The user terminal 14 may be configured, for example, by a computer (for example, a personal computer, a smartphone, a tablet terminal, etc.) including a CPU, a memory, a communication interface, an input device, and a display.

[0020] In this embodiment, chat messages are communicated between the intermediary agent system 10 and the user terminal 14. That is, in this embodiment, for the intermediary agent system 10, the user terminal 14 corresponds to a terminal that is the communication partner of the chat message. Also, in this embodiment, the intermediary agent system 10 outputs a prompt to the LLM 12a included in the processing agent system 12 as necessary. Then, the intermediary agent system 10 generates a chat message based on a response from the LLM 12a to the prompt, and outputs the generated chat message to the user terminal 14.

[0021] Fig. 2 is a functional block diagram showing an example of functions implemented in the facilitating agent system 10 according to this embodiment. Note that the facilitating agent system 10 according to this embodiment does not need to implement all of the functions shown in Fig. 2, and functions other than those shown in Fig. 2 may also be implemented.

[0022] As shown in FIG. 2, the intermediary agent system 10 according to this embodiment functionally includes, for example, a necessary information storage unit 20, a log storage unit 22, a chat message receiving unit 24, a judgment generation process execution unit 26, a judgment generation control unit 28, a prompt generation unit 30, a prompt output unit 32, a response receiving unit 34, a chat message generation unit 36, and a chat message output unit 38. The intermediary agent system 10 according to this embodiment functions as a chat control system that controls chats held in the chat system 1. The necessary information storage unit 20 and the log storage unit 22 are implemented primarily in the storage unit 10b. The chat message receiving unit 24, the prompt output unit 32, the response receiving unit 34, and the chat message output unit 38 are implemented primarily in the communication unit 10c. The judgment generation process execution unit 26, the judgment generation control unit 28, the prompt generation unit 30, and the chat message generation unit 36 ​​are implemented primarily in the processor 10a.

[0023] The above functions may be implemented by executing a program including instructions corresponding to the above functions, which is installed in the intermediary agent system 10, which is a computer, on the intermediary agent system 10. This program may also be supplied to the intermediary agent system 10 via a computer-readable information storage medium such as an optical disk, magnetic disk, magnetic tape, or magneto-optical disk, or via the Internet, for example.

[0024] In this embodiment, the necessary information storage unit 20 stores, for example, necessary information generated by a determination generation process execution unit 26, which will be described later. FIGS. 3 and 4 are diagrams showing examples of necessary information. As shown in FIGS. 3 and 4, the necessary information includes a combination of an index and a value. FIG. 3 shows three pieces of necessary information, and FIG. 4 shows five pieces of necessary information.

[0025] Furthermore, the necessary information storage unit 20 may store necessary information management data shown in Fig. 5 in advance. The necessary information management data may include, for example, a list of indexes of necessary information to be generated. The necessary information management data may then be referenced in the determination generation process described below. The necessary information management data shown in Fig. 5 includes five indexes.

[0026] In this embodiment, the log storage unit 22 stores, for example, a log of chat messages in communication between the intermediary agent system 10 and the user terminal 14. Here, for example, chat messages associated with a timestamp indicating the date and time when the chat was accepted or output, and a flag indicating whether the chat message was accepted by the intermediary agent system 10 or output by the intermediary agent system 10 may be accumulated in the log storage unit 22.

[0027] In this embodiment, the chat message receiving unit 24 receives chat messages from the user terminal 14, for example. In addition, in this embodiment, the chat message receiving unit 24 stores the chat messages in the log storage unit 22, for example.

[0028] In this embodiment, for example, in response to receiving a chat message from the user terminal 14, the determination generation process execution unit 26 determines whether or not each piece of necessary information that has not yet been generated among a given plurality of pieces of necessary information can be generated based on the chat message, and executes a determination generation process to generate the necessary information that is determined to be generable based on the chat message. Then, in this embodiment, for example, the determination generation process execution unit 26 stores the generated necessary information in the necessary information storage unit 20. Note that the determination generation process execution unit 26 may start execution of the determination generation process when triggered by the chat message reception unit 24 receiving a chat message requesting an answer from the LLM 12a, such as "Please tell me about products you recommend" (hereinafter referred to as an answer request message).

[0029] Here, the determination and generation process execution unit 26 may identify necessary information that has not yet been generated based on the necessary information management data and the generated necessary information. Then, the determination and generation process execution unit 26 may determine whether each piece of necessary information identified as not yet being generated can be generated based on the received chat message, using a known natural language processing technique such as name recognition technology, topic modeling technology, or estimation technology using a trained machine learning model. Then, the determination and generation process execution unit 26 may generate necessary information that is determined to be generable. Then, the determination and generation process execution unit 26 may output the generated necessary information to the necessary information storage unit 20.

[0030] In this embodiment, for example, in an initial state, there is no necessary information stored in the necessary information storage unit 20. Then, as necessary information is generated in the judgment generation process, the generated necessary information is accumulated in the necessary information storage unit 20.

[0031] In the present embodiment, for example, when at least one piece of necessary information that has not been generated based on a chat message is not generated, the determination generation control unit 28 causes the determination generation process execution unit 26 to execute the determination generation process for the next chat message in response to receipt of the next chat message from the user terminal 14. Here, for example, when at least one piece of necessary information that has not been generated based on a chat message is not generated, the determination generation control unit 28 may output an instruction to execute the determination generation process for the next chat message to the determination generation process execution unit 26. Furthermore, when at least one piece of necessary information that has not been generated based on a chat message is not generated, the determination generation control unit 28 may output an instruction to generate a chat message inquiring about the necessary information that has not been generated to the chat message generation unit 36.

[0032] Furthermore, the determination and generation control unit 28 may output an instruction to generate a prompt to the prompt generating unit 30 when all of the necessary information that has not yet been generated is generated based on the chat message.

[0033] As described above, in the present embodiment, in the initial state, there is no necessary information stored in the necessary information storage unit 20. In this situation, the determination generation process execution unit 26 identifies that five pieces of necessary information have not been generated, based on the necessary information management data shown in FIG. 5. Then, based on the received chat message, the determination generation process execution unit 26 determines whether or not the necessary information indexed as "gender," the necessary information indexed as "age group," the necessary information indexed as "address," the necessary information indexed as "hobbies," and the necessary information indexed as "number of people in household" can be generated.

[0034] For example, suppose it is determined that necessary information with an index of "gender," necessary information with an index of "age," and necessary information with an index of "address" can be generated. In this case, the three pieces of necessary information shown in FIG. 3 are generated, and these three pieces of necessary information are stored in the necessary information storage unit 20. In this way, when it is determined that multiple pieces of necessary information can be generated based on one chat message, the determination generation process execution unit 26 may generate the multiple pieces of necessary information based on the chat message.

[0035] In this case, in response to receiving the next chat message from the user terminal 14, the judgment generation control unit 28 causes the judgment generation process execution unit 26 to execute the judgment generation process for the next chat message.

[0036] 5 and the three pieces of necessary information that have already been generated, the determination and generation process executing unit 26 determines whether or not the necessary information indexed as "hobbies" and the necessary information indexed as "number of people in household" can be generated, based on the chat message received next.

[0037] For example, assume that it is determined that necessary information with the index "hobbies" and necessary information with the index "number of people in household" can be generated. In this case, necessary information with the index "hobbies" and necessary information with the index "number of people in household" are generated, and these two pieces of necessary information are stored in the necessary information storage unit 20. As a result, the five pieces of necessary information shown in FIG. 4 are stored in the necessary information storage unit 20.

[0038] In this case, the determination generation control unit 28 outputs an instruction to the prompt generation unit 30 to generate a prompt.

[0039] In this embodiment, for example, when all of the necessary information that has not yet been generated is generated based on chat messages, the prompt generation unit 30 generates a prompt based on a given plurality of pieces of necessary information. The prompt generation unit 30 may generate a prompt based on a given plurality of pieces of necessary information in response to receiving an instruction to generate a prompt from the determination and generation control unit 28. For example, in a situation where the five pieces of necessary information shown in FIG. 4 are stored in the necessary information storage unit 20, a prompt is generated based on these five pieces of necessary information.

[0040] Here, for example, the prompt generation unit 30 may generate a prompt including a string indicating a combination of an index and a value of the required information and a string indicated by the above-mentioned answer request message (e.g., "Please tell me about some recommended products.").

[0041] In this embodiment, for example, the prompt output unit 32 outputs the prompt generated by the prompt generation unit 30 to the LLM 12a.

[0042] In this embodiment, the answer receiving unit 34 receives an answer from the LLM 12a in response to a prompt output by the prompt output unit 32, for example.

[0043] In this embodiment, the chat message generating unit 36 ​​generates chat messages to be output to the user terminal 14, for example.

[0044] In this embodiment, for example, the chat message output unit 38 outputs the chat message generated by the chat message generation unit 36 ​​to the user terminal 14. In addition, in this embodiment, for example, the chat message output unit 38 stores the chat message in the log storage unit 22.

[0045] In this embodiment, for example, when the answer receiving unit 34 receives an answer from the LLM 12a, the chat message generating unit 36 ​​may generate a chat message based on the answer in response to the reception of the answer. Here, a chat message including the answer may be generated. Note that the chat message may be the same character string as the answer, or may be a different character string. For example, a chat message may be generated by processing the answer (e.g., by translating the answer). Then, the chat message output unit 38 may output a chat message based on the answer to the user terminal 14.

[0046] Furthermore, the chat message generation unit 36 ​​may generate a chat message inquiring about the necessary information that has not been generated when at least one piece of necessary information that has not been generated based on a chat message is not generated by the determination and generation process execution unit 26. For example, the chat message generation unit 36 ​​may generate a chat message inquiring about the necessary information that has not been generated in response to receiving an instruction to generate a chat message from the determination and generation control unit 28. Then, the chat message output unit 38 may output the chat message inquiring about the necessary information that has not been generated to the user terminal 14.

[0047] For example, suppose that the three pieces of necessary information shown in FIG. 3 are generated as described above, but the necessary information with the index "hobbies" and the necessary information with the index "number of people in household" are not generated. In this case, a chat message inquiring about the user's hobbies, such as "What are your hobbies?", may be generated and output to the user terminal 14. Alternatively, a chat message inquiring about the number of people in the user's household, such as "How many people are in your household?", may be generated and output to the user terminal 14.

[0048] For example, in response to the chat message "What are your hobbies?", a chat message may be received from the user terminal 14, such as "My hobby is playing golf with my family of four." In such a case, the determination generation process execution unit 26 may generate both the necessary information indexed as "hobbies" and the necessary information indexed as "number of people in household," based on the chat message received from the user terminal 14.

[0049] In this way, when the judgment generation process execution unit 26 judges that necessary information other than the necessary information can be generated based on a chat message received from the user terminal 14 in response to a chat message inquiring about necessary information that will not be generated, the judgment generation process execution unit 26 may generate the other necessary information based on the chat message.

[0050] It is possible that a chat message may be received that does not allow for generating necessary information with the index "hobbies" in response to a chat message inquiring about a user's hobbies, but allows for generating necessary information with the index "number of people in household." In this case, the determination generation process execution unit 26 may generate necessary information with the index "number of people in household" based on the chat message received from the user terminal 14.

[0051] An example of the flow of processing performed in the facilitating agent system 10 according to this embodiment will now be described with reference to the flow charts shown in FIGS. 6A to 6D.

[0052] In this processing example, it is assumed that chat message communication has already been performed between the intermediary agent system 10 and the user terminal 14. It is also assumed that chat messages output from the intermediary agent system 10 to the user terminal 14 and chat messages received by the intermediary agent system 10 from the user terminal 14 in the communications that have been performed up to now have been accumulated as logs in the log storage unit 22.

[0053] An example of a process that is executed in response to the chat message receiving unit 24 receiving an answer request message in the above-mentioned situation will be described below.

[0054] When the chat message receiving unit 24 receives an answer request message (S101), the chat message receiving unit 24 outputs the answer request message to the log storage unit 22 (S102).

[0055] Then, the determination generation process execution unit 26 determines whether or not each piece of necessary information whose index is indicated in the necessary information management data stored in the necessary information storage unit 20 can be generated based on the chat message received in the process shown in S101 (S103).

[0056] Then, the determination and generation process executing unit 26 generates the necessary information determined to be generable in the process shown in S103 based on the chat message accepted in the process shown in S101 (S104), and outputs the generated necessary information to the necessary information storage unit 20 (S105). Note that if there is no necessary information determined to be generable in the process shown in S103, the processes shown in S104 and S105 are not executed.

[0057] Then, the determination generation control unit 28 determines whether or not all the necessary information has been generated in the process shown in S104, based on the necessary information management data and the necessary information stored in the necessary information storage unit 20 (S106).

[0058] If it is confirmed that at least one of the necessary information indexed in the necessary information management data has not been generated (S106: N), the judgment generation control unit 28 outputs an instruction to execute the judgment generation process for the next chat message to the judgment generation process execution unit 26 (S107), and outputs an instruction to generate a chat message inquiring about the necessary information that has not been generated to the chat message generation unit 36 ​​(S108).

[0059] Then, the chat message generation unit 36 ​​generates a chat message inquiring about the necessary information that has not been generated (S109). Then, the chat message output unit 38 outputs the chat message generated in the process shown in S109 to the user terminal 14 (S110), and outputs the chat message to the log storage unit 22 (S111).

[0060] Then, the chat message receiving unit 24 waits for the next chat message to be received (S112).

[0061] Then, when reception of a chat message is detected in the process shown in S112, the chat message reception unit 24 outputs the chat message to the log storage unit 22 (S113).

[0062] Then, the determination generation process execution unit 26 determines whether or not each piece of ungenerated necessary information identified based on the necessary information management data stored in the necessary information storage unit 20 and the necessary information can be generated based on the chat message whose acceptance was detected in the process shown in S113 (S114).

[0063] Then, the determination and generation process executing unit 26 generates the necessary information determined to be generable in the process shown in S114 based on the chat message whose acceptance was detected in the process shown in S113 (S115), and outputs the generated necessary information to the necessary information storage unit 20 (S116). Note that if there is no necessary information determined to be generable in the process shown in S114, the processes shown in S115 and S116 are not executed.

[0064] Then, the determination and generation control unit 28 determines whether or not all of the necessary information that has not been generated in the processing shown in S115 has been generated based on the necessary information management data and necessary information stored in the necessary information storage unit 20 (S117).

[0065] If it is confirmed that at least one of the necessary information that has not been generated has not been generated (S117: N), the process returns to S107, and the processes from S107 onwards are executed again.

[0066] If it is confirmed in the process shown in S106 that all of the necessary information indexed in the necessary information management data has been generated (S106: Y), or if it is confirmed in the process shown in S117 that all of the necessary information that has not yet been generated has been generated (S117: Y), the judgment generation control unit 28 outputs an instruction to generate a prompt to the prompt generation unit 30 (S118).

[0067] Then, the prompt generation unit 30 generates a prompt based on the plurality of pieces of necessary information stored in the necessary information storage unit 20 (S119). Then, the prompt output unit 32 outputs the prompt generated by the process shown in S119 to the LLM 12a (S120).

[0068] Then, the response receiving unit 34 waits for a response to the prompt from the LLM 12a (S121).

[0069] When the reply receiving unit 34 detects the receipt of a reply, the chat message generating unit 36 ​​generates a chat message based on the reply (S122). Then, the chat message output unit 38 outputs the chat message generated in the process shown in S122 to the user terminal 14 (S123), and outputs the chat message to the log storage unit 22 (S124), and the process shown in this processing example ends.

[0070] Even after the processing shown in this processing example is completed, communication of chat messages between the intermediary agent system 10 and the user terminal 14 continues.

[0071] In this embodiment, in response to receiving an answer request message from the user terminal 14, the determination generation process execution unit 26 may determine whether or not each of a given plurality of pieces of necessary information can be generated based on a log of chat messages in communication with the user terminal 14, and may generate the necessary information determined to be generable based on the log. Then, if at least one of the plurality of pieces of necessary information cannot be generated based on the log, the determination generation control unit 28 may cause the determination generation process execution unit 26 to execute the determination generation process for the next chat message in response to receiving the next chat message from the user terminal 14. On the other hand, if all of the necessary information is generated based on the log, the prompt generation unit 30 may generate a prompt based on the plurality of pieces of necessary information.

[0072] For example, after the process shown in S102 is executed, the processes shown in S103 to S105 are not executed, and the determination and generation process executing unit 26 may determine whether or not each piece of necessary information whose index is indicated in the necessary information management data stored in the necessary information storage unit 20 can be generated based on the chat message log stored in the log storage unit 22. Then, the determination and generation process executing unit 26 may generate the necessary information determined to be generatable based on the log stored in the log storage unit 22, and output the generated necessary information to the necessary information storage unit 20. Then, the processes from S106 onwards may be executed.

[0073] Furthermore, the judgment generation process execution unit 26 may be configured to constantly execute the judgment generation process without starting execution of the judgment generation process when the response request message is received as a trigger.

[0074] Alternatively, the LLM 12a may generate a response using information that neither the user terminal 14 nor the intermediary agent system 10 has access to.

[0075] In this embodiment, as described above, a prompt generated based on a given plurality of pieces of necessary information is output to the LLM 12a only after all of the given plurality of pieces of necessary information have been generated based on the chat message.

[0076] This ensures that the LLM 12a is provided with the information necessary to provide an appropriate response to the prompt, thereby preventing the LLM 12a from providing an inappropriate response to the prompt, such as an ambiguous or irrelevant response.

[0077] In this way, according to this embodiment, the user can obtain information in a manner that feels like chatting naturally with a person, while increasing the likelihood that an appropriate response to a prompt will be obtained from the LLM 12a.

[0078] The present invention is not limited to the above-described embodiment. For example, the necessary information storage unit 20 may be provided in a system other than the facilitating agent system 10, rather than the facilitating agent system 10. Also, for example, the log storage unit 22 may be provided in a system other than the facilitating agent system 10, rather than the facilitating agent system 10.

[0079] Furthermore, the specific character strings and numerical values ​​described above and the specific character strings and numerical values ​​in the drawings are examples, and the present invention is not limited to these character strings and numerical values. [Explanation of symbols]

[0080] 1 Chat system, 10 Intermediary agent system, 10a Processor, 10b Memory unit, 10c Communication unit, 12 Processing agent system, 12a Large scale language model (LLM), 14 User terminal, 16 Computer network, 20 Required information memory unit, 22 Log memory unit, 24 Chat message reception unit, 26 Judgment generation processing execution unit, 28 Judgment generation control unit, 30 Prompt generation unit, 32 Prompt output unit, 34 Answer reception unit, 36 Chat message generation unit, 38 Chat message output unit.

Claims

1. a determination and generation process execution means for executing a determination and generation process in response to reception of a chat message from a terminal that is a communication partner of the chat message, for determining whether or not each of the necessary information that has not yet been generated among a given plurality of pieces of necessary information can be generated based on the chat message, and for generating the necessary information that is determined to be generable based on the chat message; a determination and generation control means for causing the determination and generation process execution means to execute the determination and generation process for a next chat message in response to reception of the next chat message from the terminal when at least one of the necessary information that has not been generated based on the chat message has not been generated; a prompt generating means for generating a prompt based on the plurality of pieces of necessary information when all of the ungenerated pieces of necessary information are generated based on the chat message; prompt output means for outputting the prompt to a large scale language model; answer receiving means for receiving an answer from the large-scale language model in response to the prompt; a chat message output means for outputting a chat message based on the received reply to the terminal; A chat control system including:

2. further comprising an inquiry message output means for outputting, to the terminal, a chat message inquiring about the necessary information that has not been generated when at least one of the necessary information that has not been generated based on the chat message; The chat control system of claim 1 .

3. the determination and generation process execution means, when it is determined that necessary information different from the necessary information can be generated based on a chat message received from the terminal in response to a chat message inquiring about the necessary information not to be generated, generates the different necessary information based on the chat message; The chat control system of claim 2 .

4. a log determination and generation means for determining whether or not each of the plurality of pieces of necessary information can be generated based on a log of chat messages in communication with the terminal in response to receiving a chat message from the terminal requesting a response from the large-scale language model, and generating the necessary information determined to be generable based on the log; when at least one of the plurality of pieces of necessary information is not generated based on the log, the determination generation control means causes the determination generation process execution means to execute the determination generation process for the next chat message in response to reception of the next chat message from the terminal; When all the necessary information is generated based on the log, the prompt generation means generates a prompt based on the plurality of pieces of necessary information. The chat control system of claim 1 .

5. the determination and generation process execution means, when it is determined that a plurality of pieces of necessary information can be generated based on one chat message, generates the plurality of pieces of necessary information based on the chat message; The chat control system of claim 1 .

6. a step of executing a determination and generation process in which, in response to receiving a chat message from a terminal that is a communication partner of the chat message, for each of the plurality of pieces of necessary information that has not yet been generated, it is determined whether or not it can be generated based on the chat message, and the necessary information that is determined to be generable is generated based on the chat message; when at least one of the necessary information that has not been generated based on the chat message is not generated, executing the determination and generation process for the next chat message in response to receiving the next chat message from the terminal; generating a prompt based on the plurality of pieces of required information when all of the ungenerated pieces of required information are generated based on the chat message; outputting the prompt to a large-scale language model; accepting a response from the large-scale language model to the prompt; outputting a chat message based on the accepted response to the terminal; A chat control method including:

7. a step of executing a determination and generation process in which, in response to receiving a chat message from a terminal that is a communication partner of the chat message, for each of the plurality of pieces of necessary information that has not yet been generated, it is determined whether or not it can be generated based on the chat message, and the necessary information that is determined to be generable is generated based on the chat message; when at least one of the necessary information that has not been generated based on the chat message is not generated, executing the determination and generation process for the next chat message in response to receiving the next chat message from the terminal; generating a prompt based on the plurality of pieces of required information when all of the ungenerated pieces of required information are generated based on the chat message; outputting the prompt to a large-scale language model; accepting a response from the large-scale language model to the prompt; outputting a chat message based on the accepted response to the terminal; A program that causes a computer to execute the following.

Images