Information processing method, program, and information processing system

Abstract

By utilizing chat history, responses optimized for each user can be generated. [Solution] According to one embodiment of an information processing method, the method includes an answer generation step of generating an answer to text input by a user using a first language model and presenting the answer to the text to the user; an accumulation step of accumulating the text and the answer generated in the generation step as a log; an acquisition step of acquiring the log at a predetermined timing; and a summary generation step of generating a summary from the acquired log using a second language model, wherein in the summary generation step, if a past summary is stored in a memory unit, a summary based on the past summary in addition to the log is generated and stored in the memory unit, and the past summary is deleted from the memory unit.

Description

[Technical Field]

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

[0002] Patent Document 1 discloses a chat system that can obtain information organized in text format in the form of questions and answers from a chat history between a potential customer and the chatter. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2024-150099 Summary of the Invention [Problem to be solved by the invention]

[0004] The above prior art technology can obtain standardized answers by accumulating sets of questions and answers exchanged between potential customers. However, it is difficult to utilize this technology in situations where individualized responses are desired, such as customer support and counseling. In other words, the prior art technology makes it difficult to provide individualized responses that fully utilize the individual chat history accumulated for each user.

[0005] In consideration of the above, an object of the present invention is to provide an information processing method, a program, and an information processing system that can utilize chat history to generate responses optimized for each user. [Means for solving the problem]

[0006] According to one embodiment of an information processing method, the method includes an answer generation step of generating an answer to text input by a user using a first language model and presenting the answer to the text to the user; an accumulation step of accumulating the text and the answer generated in the generation step as a log; an acquisition step of acquiring the log at a predetermined timing; and a summary generation step of generating a summary from the acquired log using a second language model, wherein in the summary generation step, if a past summary is stored in a memory unit, a summary based on the past summary in addition to the log is generated and stored in the memory unit, and the past summary is deleted from the memory unit.

[0007] According to one embodiment of the program, an information processing device is caused to execute an information processing method that includes an answer generation step of generating an answer to text input by a user using a first language model and presenting the answer to the text to the user, an accumulation step of accumulating the text and the answer generated in the generation step as a log, an acquisition step of acquiring the log at a predetermined timing, and a summary generation step of generating a summary from the acquired log using a second language model, wherein in the summary generation step, if a past summary is stored in a memory unit, an information processing method is executed in which a summary based on the past summary is generated in addition to the log and stored in the memory unit, and the past summary is deleted from the memory unit.

[0008] According to one embodiment, the information processing system is equipped with an information processing device, and performs the following steps: an answer generation step of generating an answer to text input by a user using a first language model and presenting the answer to the text to the user; an accumulation step of accumulating the text and the answer generated in the generation step as a log; an acquisition step of acquiring the log at a predetermined timing; and a summary generation step of generating a summary from the acquired log using a second language model, wherein in the summary generation step, if a past summary is stored in a memory unit, the information processing system generates a summary based on the past summary in addition to the log and stores it in the memory unit, and deletes the past summary from the memory unit. [Effects of the Invention]

[0009] According to one embodiment, chat history can be utilized to generate personalized responses. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram illustrating an example of a configuration of an information processing system according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of a hardware configuration of a server according to an embodiment. [Figure 3] FIG. 2 is a diagram illustrating an example of a hardware configuration of a user terminal according to the embodiment. [Figure 4] FIG. 2 is a diagram illustrating an example of a functional configuration of a server according to an embodiment. [Figure 5] FIG. 2 is a diagram illustrating an example of a functional configuration of a user terminal according to the embodiment. [Figure 6] 10 is a flowchart illustrating an example of a process for generating a summary of a chat log, which is executed by the information processing system according to the embodiment. [Figure 7] 10 is a flowchart illustrating an example of a process for generating an answer based on a summary, which is executed by the information processing system according to the embodiment. [Figure 8] 1A and 1B are diagrams showing an overview of a display on the server device side and a display on the user terminal side, respectively, for a chat display with a dialogue system according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] An embodiment of an information processing system according to the present invention will be described below with reference to Figures 1 to 5. In each drawing, the same or equivalent components and parts are denoted by the same reference numerals. Also, the dimensional proportions in the drawings are exaggerated for the sake of explanation and may differ from the actual proportions.

[0012] (System Overview) First, an overview of an information processing system 10 according to this embodiment will be described. The information processing system 10 according to this embodiment is a system that dynamically summarizes chat logs in a dialogue system that utilizes artificial intelligence (AI) to generate answers. In this embodiment, the information processing system 10 is a system that summarizes individual chats with users according to predetermined timing, such as an upper limit on the number of tokens or a change in topic, and generates answers by including the latest summary in a prompt. Specifically, it is possible to create a new summary based on the chat log and past summaries, and consistently maintain the dialogue context with the user.

[0013] (System Configuration) Fig. 1 is a diagram showing an example of the configuration of an information processing system 10 according to this embodiment. As shown in Fig. 1, the information processing system 10 according to this embodiment includes a server device 1 and a user terminal 2, which are connected to each other so as to be able to communicate with each other via a network N. The network N is, for example, a wired local area network (LAN), a wireless LAN, the Internet, a public line network, a mobile data communication network, or a combination thereof. In the example of Fig. 1, the information processing system 10 includes one server device 1 and one user terminal 2, but may include multiple of each.

[0014] The server device 1 is an example of an information processing device that monitors the screen of the user terminal 2 and issues operational instructions to the avatar 3 in response to changes in the screen. The server device 1 may be a PC (Personal Computer), a smartphone, a tablet terminal, a server device, a microcomputer, or a combination of these. The specific configuration and operation of the server device 1 will be described later.

[0015] The user terminal 2 is an example of an information processing device that performs operations for inputting and displaying various types of information. The user terminal 2 may be a PC (Personal Computer), a smartphone, a tablet terminal, a server device, a microcomputer, a wearable device, or a combination of these. In this embodiment, an information processing device that allows a user to input text into a chat and that can display a chat history in an interactive system is used as an example. Note that the user can input text in natural language.

[0016] (Hardware configuration - Server) 2 is a block diagram showing the hardware configuration of the server device 1. The server device 1 includes a processor 101, a memory 102, a storage 103, and a communication I / F 104, which are communicably connected to each other via a bus B.

[0017] The processor 101 controls each component of the server device 1 and realizes the functions of the server device 1 by loading various programs stored in the storage 103 into the memory 102 and executing them. The programs executed by the processor 101 include, but are not limited to, an OS (Operating System) and various programs described below. Execution of these programs by the processor 101 realizes part of the state visualization method according to this embodiment. The processor 101 is, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), a DSP (Digital Signal Processor), or a combination thereof.

[0018] The memory 102 is, for example, a read-only memory (ROM), a random access memory (RAM), or a combination thereof. The ROM is, for example, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a combination thereof. The RAM is, for example, a dynamic random access memory (DRAM), a static random access memory (SRAM), a magnetoresistive random access memory (MRAM), or a combination thereof.

[0019] The storage 103 stores an OS, various programs (described later), and various data. The storage 103 is, for example, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), a storage class memory (SCM), or a combination of these.

[0020] The communication I / F 104 is an interface for connecting the server device 1 to external devices including the user terminal 2 via the network N and controlling communication. The communication I / F 104 is, for example, an adapter compliant with Bluetooth (registered trademark), Wi-Fi (registered trademark), ZigBee (registered trademark), Ethernet (registered trademark), or optical communication (e.g., Fibre Channel), but is not limited to these.

[0021] The input device 105 is a device for inputting information to the server device 1. The input device 105 is, for example, a mouse, a keyboard, a touch panel, a microphone, a scanner, a photographing device (camera), various sensors, or an operation button, but is not limited to these.

[0022] The output device 106 is a device for outputting information from the server device 1. The output device 106 is, for example, a display device (display), a projector, a printer, a speaker, or a vibrator, but is not limited to these.

[0023] (Hardware configuration - user terminal) 3 is a block diagram showing the hardware configuration of the user terminal 2. The user terminal 2 includes a processor 201, a memory 202, a storage 203, a communication I / F 204, an input / output I / F 205, an input device 206, and an output device 207, which are communicably connected to each other via a bus B. The communication I / F 204 is an interface for connecting the user terminal 2 to external devices including the server device 1 via a network N and for controlling communication. The communication I / F 204 is, for example, an adapter compliant with Bluetooth (registered trademark), Wi-Fi (registered trademark), ZigBee (registered trademark), Ethernet (registered trademark), or optical communication (e.g., Fibre Channel), but is not limited to these.

[0024] The input / output I / F 205 is an interface for connecting an input device 206 and an output device 207 to the user terminal 2. The input device 206 is, for example, a mouse, a keyboard, a touch panel, a microphone, a scanner, a camera, various sensors, operation buttons, or a combination of these. The output device 207 as a user interface is, for example, a display, a projector, a printer, a speaker, a vibrator, or a combination of these. In this embodiment, as an example, the output device 207 and the input device 206 are an integrally configured touch panel display.

[0025] In this embodiment, the program may be written to the memory 202 or the storage 203 during the manufacturing stage of the server device 1, or may be provided to the server device 1 via the network N. Alternatively, the program may be provided to the server device 1 via a non-transitory computer-readable recording medium such as a disk medium (not shown).

[0026] (Functional configuration - Server) Next, the functional configuration of the server device 1 will be described. Fig. 4 is a diagram showing an example of the functional configuration of the server device 1. When executing various programs, the server device 1 realizes various functions using the above-mentioned hardware resources. The server device 1 has a communication unit 11, a storage unit 12, and a control unit 13 as functional configurations realized by the server device 1. Each functional configuration is realized by the processor 101 reading and executing a program stored in the memory 102 or storage 103.

[0027] The storage unit 12 also stores a summary DB 121 that stores the latest summaries, and a chat history DB 122 that stores the history of chat exchanges with users. The summary DB 121 stores summaries generated at predetermined times during chat exchanges between the user and the dialogue system. The latest summary is always stored, and older summaries are deleted each time a summary is generated. The chat history DB 122 stores chat exchanges with users as a log. By storing past chat histories, it is possible to generate replies to users from past exchanges. Note that by providing a summary DB 121 and a chat history DB 122 for each user, it is possible to generate replies optimized for each user.

[0028] The control unit 13 includes an answer generation unit 131 that generates answers from users, and an accumulation unit 132 that accumulates chat exchanges with users in a chat history DB 122 via the screen of the user terminal 2. The control unit 13 also includes an acquisition unit 133 that acquires chat logs from the accumulated chat history DB 122 at a predetermined timing, and a summary generation unit 134 that generates a summary based on the chat logs acquired by the acquisition unit.

[0029] The answer generation unit 131 generates an answer using, for example, LLM (Large Language Models) in response to text input by the user using the user terminal 2. At this time, for example, if a summary is stored in the summary DB 121, the answer is generated by including the summary in the prompt. Furthermore, the answer generation unit 131 searches the chat history DB 122 for keywords in the text input by the user to determine whether or not a similar past chat log is stored based on the keywords, and if a corresponding chat log is found, the answer is generated taking the chat log into consideration.

[0030] The accumulation unit 132 accumulates the text input by the user and the response generated by the response generation unit 131 in response to the text as a chat log in the chat history DB 122. The acquisition unit 133 acquires, at a predetermined timing, the chat log from the timing at which a previous summary was generated from the chat history DB 122. The summary generation unit 134 generates a summary of the chat log acquired by the acquisition unit 133 at the predetermined timing, for example, using LLM, and stores the generated summary in the summary DB 121. At this time, if a previously generated summary is stored in the summary DB 121, the previous summary is deleted.

[0031] In this embodiment, the predetermined timing refers to any of the following: when the number of tokens input by the user exceeds a preset threshold; when a predetermined time has elapsed since the answer generation unit 131 presented an answer to the user; or when the topic changes in the text input by the user. The models used by the answer generation unit 131 and the summary generation unit 134 are not limited and may be selected as appropriate. In this embodiment, the answer generation unit 131 searches chat logs in the chat history DB 122 when generating an answer to text input by the user. However, a separate language model dedicated to search may be used for this search. That is, the language model for generating an answer and the language model for searching chat logs from the chat history DB may be different models and processed in parallel. In this embodiment, the language model used by the answer generation unit 131 when generating an answer is referred to as a first language model, the language model used by the summary generation unit 134 when generating a summary is referred to as a second language model, and the language model used when searching chat logs from the chat history DB is referred to as a third language model. These will be described in detail later.

[0032] (Functional configuration - user terminal) Next, the functional configuration of the user terminal 2 will be described. Fig. 5 is a diagram showing an example of the functional configuration of the user terminal 2. When executing various programs, the user terminal 2 realizes various functions using the above-mentioned hardware resources. The user terminal 2 has, as functional components realized by the user terminal 2, a communication unit 21, a storage unit 22 in which a program 221 is stored, and a control unit 23. Each functional component is realized by the processor 201 reading and executing the program 221 stored in the memory 202 or the storage 203. The control unit 23 includes an information acquisition control unit 231 that acquires information sent from the server device 1, and a display unit 232 that displays the acquired information on the output device 207.

[0033] Next, the flow of processing of the information processing system according to this embodiment will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of processing for generating a summary of a chat log executed by the information processing system 10 according to this embodiment. Specifically, the processing for generating a summary of an exchange in a dialogue with a user at a predetermined timing will be described.

[0034] First, in step S101, the accumulation unit 132 accumulates, as a chat log, in the chat history DB 122 (accumulation step), the text input by the user via the user terminal 2 and the answer generated by the answer generation unit 131 for the text using the first language model (answer generation step). Next, in step S102, the acquisition unit 133 determines whether or not a predetermined timing has arrived, and if the predetermined timing has arrived (YES), the process proceeds to step S103, where the acquisition unit 133 acquires the chat log from the chat history DB 122. On the other hand, if the predetermined timing has not arrived (NO), the acquisition unit 133 waits until the predetermined timing arrives.

[0035] In this embodiment, the predetermined timing is one of the following timings: when the number of tokens input by the user exceeds a preset threshold; when a predetermined time has elapsed since the answer generation unit 131 presented an answer to the user; or when the topic changes in the text input by the user. Here, the threshold for the number of tokens in the log can be changed as appropriate; for example, an appropriate threshold may be set based on system limitations. The predetermined time can be set as appropriate, such as when five minutes have elapsed since the answer was presented to the user. Regarding the topic change, for example, when a user inputs a setting to change the topic (e.g., start a new chat) from the chat menu, the text input by the user may be analyzed by LLM, and the answer generation unit 131 may detect the topic change and notify the acquisition unit 133 before generating an answer. Note that the predetermined timing is not limited to these; it may be increased or decreased, or may be set as appropriate. Furthermore, if the acquisition unit 133 determines that any of the timings has occurred, the process proceeds to step S103. Next, in step S103, the acquisition unit 133 acquires a log from the chat history DB 122 (acquisition step). The log acquired from the chat history DB 122 in step S103 is a chat log covering the period from the previous acquisition until a predetermined timing.

[0036] Next, in step S104, the summary generation unit 134 generates a summary using a second language model based on the summary stored in the summary DB and the chat log acquired by the acquisition unit 133 in step S103 (summary generation step). That is, the summary generation unit 134 generates a new summary by reflecting the past summary. Note that, although the second language model is used to generate the summary, the present invention is not limited to this. For example, a configuration may be adopted in which multiple language models are combined to improve the accuracy of the summary. Then, in step S105, the summary generation unit 134 stores the generated summary in the summary DB 121 and deletes the summary (past summary) used when generating the summary in step S104. Note that, in this embodiment, the past summary is deleted. However, the present invention is not limited to this. For example, the past summary may be stored in the summary DB 121 without being deleted, and only the latest summary may always be used without using the past summaries. As a result, only the latest summary is always stored in the summary DB 121. Here, when generating a summary, the summary generation unit 134 summarizes a preset percentage of the chat log acquired by the acquisition unit 133. For example, if the ratio is set to 70%, the summary generation unit 134 will extract and summarize the first 70% of the chat log acquired by the acquisition unit 133. The ratio can be set appropriately and can be changed as needed. When generating a summary, a preset ratio is randomly extracted from the chat log, but if the LLM makes an inquiry to the summary generation unit 134 and there is a corresponding response, the relevant exchange is extracted as a set.

[0037] According to this embodiment, even if the chat history becomes enormous, summaries are automatically generated at a predetermined timing, so that necessary context information can always be aggregated into the latest summary. Furthermore, summaries are updated based on the number of tokens input by the user, the elapsed time since the answer was submitted, or the condition of a change in topic, so summaries are generated at appropriate timing. In addition, new summaries are created while reflecting past summaries, and old summaries that have been reflected are deleted, so the number of summaries does not increase excessively, and chat content for each user can be consistently managed. This makes it possible to generate answers using summaries of the chat history that are always up-to-date and optimized, and to consistently maintain the conversation context with the user. This, in turn, improves usability and makes efficient use of system resources.

[0038] Next, the flow of processing in the information processing system according to this embodiment will be described with reference to Fig. 7. Fig. 7 is a flowchart showing an example of processing executed by the information processing system 10 according to this embodiment to generate an answer based on the generated summary. Specifically, the processing of generating an answer based on the summary generated by the summary generation unit 134 will be described.

[0039] First, in step S201, the answer generation unit 131 receives text input by the user via the user terminal 2. In step S202, the answer generation unit 131 searches the chat history DB 122 for a similar chat log based on keywords in the received text using the third language model. If a chat log is found (YES), in step S203, the answer generation unit 131 acquires the corresponding chat log from the chat history DB 122. On the other hand, if a chat log is not found (NO), the process proceeds to step S204. Next, in step S204, the answer generation unit 131 acquires a summary from the summary DB 121. Then, in step S205, the answer generation unit 131 generates an answer to the text acquired in step S201 using the first language model based on the acquired summary, and transmits the answer to the user terminal 2 to present it to the user. At this time, if the answer generation unit 131 has acquired a chat log in step S203, the answer generation unit 131 generates an answer by adding the chat log acquired in step S203 to the prompt in addition to the summary acquired in step S204. Next, in step S206, the accumulation unit 132 accumulates in the chat history DB 122 the text received in step S201 and the response generated in step S205.

[0040] According to this embodiment, a chat log related to the text entered by the user can be searched for and referenced, and a reply can be generated by combining the generated summary, thereby always efficiently utilizing the latest context and past exchanges. In particular, since a reply can be generated using a summary even when there is no related chat log, it is possible to accurately reflect only the necessary information while reducing the duplication of exchanges or the use of unnecessary information, and smoothly obtain a reply that is tailored to each user's exchange.

[0041] Next, an example of a screen display will be shown using Fig. 8. Fig. 8(A) shows the display screen on the server device 1 side, and Fig. 8(B) shows the display screen on the user terminal 2 side. In Fig. 8(A), screen A1 is a screen displayed on the output device 106 of the server device 1. In Fig. 8(B), screen A2 is a screen displayed on the output device 106 of the user terminal 2.

[0042] 8(A), screen A1 displays summary ab1 generated in step S104 of FIG. 6, text tx1 input by the user, and answer tx2 generated in step S205 of FIG. 7. Summary ab1 displays the latest summary each time a summary is generated. Text tx1 displays text input by the user using user terminal 2, and is displayed each time it is received in step S201 of FIG. 7. Answer tx2 is an answer to text tx1, and is displayed each time an answer is generated and sent to user terminal 2.

[0043] In FIG. 8(B), screen A2 displays an icon ic1 indicating that the system is a dialogue system utilizing artificial intelligence (AI), a summary ab1 generated in step S104 of FIG. 6, text tx1 entered by the user, and an answer tx2 generated in step S205 of FIG. 7. Icon ic1 indicates that the respondent is an AI. The text tx1 and answer tx2 are displayed in the same manner as in FIG. 8(A), but are displayed with their left and right reversed. This is because in FIG. 8(A), the answer is generated on the server device 1 side and presented to the user, so the generated answer tx2 is displayed on the right. On the other hand, in FIG. 8(B), the text tx1 entered by the user on the user terminal 2 side is displayed on the right. In this manner, the server device 1 and the user terminal 2 are displayed so that the side presented by them is on the right. Note that in this embodiment, the presented text tx1 or answer tx2 is displayed on the right, but this is not limited thereto. For example, they may be displayed on the left, or other display methods may be used.

[0044] According to this embodiment, the latest summary, the text entered by the user, and the AI's response can be visually grasped simultaneously on both the server device 1 and the user terminal 2, making it easy to share the conversation content or context. Furthermore, the display of an icon indicating that it is an AI makes it easier for the user to recognize that they are interacting with the system. Furthermore, by updating the latest summary each time, a smooth dialogue can always be conducted based on the latest context.

[0045] In the present embodiment, a predetermined percentage of the acquired chat logs is randomly extracted and summarized. However, this is not a limitation. For example, importance scores may be calculated in advance and only messages with high scores may be extracted and summarized. In the present embodiment, summarization is performed when the number of tokens exceeds a threshold. However, this is not a limitation. For example, summarization may be performed based on the frequency of user input. Furthermore, in the present embodiment, old summaries are deleted so that only the most recent data is always retained. However, this is not a limitation. For example, past summaries may be stored as a history and kept in a referable state. In addition, in the present embodiment, relevant chat logs are acquired using a keyword search when generating a response. However, this is not a limitation. For example, chat logs with high similarity may be extracted using a vector search or the like.

[0046] As described above, according to this embodiment, by automatically generating summaries at predetermined times and generating replies based on the generated summaries, it is possible to always maintain the latest context even in long-term exchanges. Furthermore, by reflecting past summaries while deleting unnecessary summaries when creating or updating summaries, it is possible to consistently and concisely organize necessary information. In addition, by obtaining relevant chat logs using keyword search and combining that information with summaries to generate replies, it is possible to accurately reflect replies that are tailored to the user. As a result, it is possible to generate replies that are optimized for each user by utilizing chat history and including past exchanges.

[0047] For example, this technology can be applied to chat services involving long-term, ongoing interactions, such as fortune telling. In this case, generating responses based on past interactions can provide a service similar to chatting with a personal fortune teller. In interactive counseling for mental health and health management, long-term interactions with users are accumulated. By summarizing key points and deleting old information, it is possible to grasp necessary changes in each user's condition and provide continuous, personalized advice to the user. Furthermore, in the case of chatbots at customer support desks provided by companies, by summarizing and organizing user inquiries, responses tailored to individual situations can be generated while referencing similar past cases, thereby maintaining continuous usability. In this way, the features of the present invention, such as summarizing chat logs and deleting old summaries, utilizing chat content based on each user's chat history, and realizing long-term context maintenance and personalization, can be applied in all cases. This makes it possible to provide a dialogue (chat) system that can withstand long-term, continuous use.

[0048] Although one embodiment of the present invention has been described above, the present invention is not limited to the above-described embodiment, and modifications, improvements, etc. within the scope of achieving the object of the present invention are included in the present invention.

[0049] Furthermore, for example, the above-described series of processes can be executed by hardware or software. In other words, the functional configuration is merely an example and is not particularly limited. That is, it is sufficient that the information processing system is provided with a function that can execute the above-described series of processes as a whole, and there is no particular limit to the type of functional block used to realize this function. Furthermore, the location of the functional block is also not particularly limited and can be arbitrary. For example, a functional block of a server may be transferred to a user terminal or the like. Conversely, a functional block of a user terminal may be transferred to a server or the like. Furthermore, one functional block may be configured as a single piece of hardware, a single piece of software, or a combination thereof.

[0050] Furthermore, for example, when a series of processes is executed by software, the programs constituting the software are installed onto a computer or the like from a network or a recording medium. The computer may be a computer incorporated into dedicated hardware. Furthermore, the computer may be a computer capable of executing various functions by installing various programs thereon, such as a server, a general-purpose smartphone, or a personal computer.

[0051] Furthermore, for example, the recording medium containing such a program may be configured not only as a removable medium (not shown) that is distributed separately from the device main body in order to provide the program to the user, but also as a recording medium that is provided to the user in a state that is pre-installed in the device main body.

[0052] In this specification, the steps describing the program recorded on the recording medium include not only processes that are performed in chronological order, but also processes that are not necessarily performed in chronological order but are performed in parallel or individually. Also, in this specification, the term "system" means an overall device composed of multiple devices or multiple means, etc. [Explanation of symbols]

[0053] 10 Information Processing Systems 1 server 2 user terminals 131 Answer generation part 132 Storage Unit 133 acquisition department 134 Summary generator

Claims

1. An information processing method executed by an information processing device, an answer generation step of generating an answer to a text input by a user using a first language model and presenting the answer to the text to the user; a storage step of storing the text and the answer generated in the generation step as a log; an acquisition step of acquiring the log at a predetermined timing; a summary generation step of extracting the acquired log at a preset rate and generating a summary using a second language model; Including, In the summary generating step, if a past summary is stored in the storage unit, a summary is generated based on the past summary in addition to the log and stored in the storage unit, and the past summary is deleted from the storage unit. Information processing methods.

2. The predetermined timing is one of the timings when the number of tokens input by the user exceeds a preset threshold, when a predetermined time has elapsed since the answer was presented to the user, or when the topic in the text has changed due to the user's input. The information processing method according to claim 1 .

3. In the answer generating step, if there is a summary generated in the summary generating step, the answer is generated by including the summary as a prompt. The information processing method according to claim 1 .

4. In the answer generating step, if a keyword in the text is included in the log accumulated in the accumulation step, the log is added to the prompt to generate the answer. The information processing method according to claim 3 .

5. In the information processing device, an answer generation step of generating an answer to a text input by a user using a first language model and presenting the answer to the text to the user; a storage step of storing the text and the answer generated in the generation step as a log; an acquisition step of acquiring the log at a predetermined timing; a summary generation step of extracting the acquired log at a preset rate and generating a summary using a second language model; Including, In the summary generating step, if a past summary is stored in the storage unit, a summary is generated based on the past summary in addition to the log and stored in the storage unit, and the past summary is deleted from the storage unit. A program for executing an information processing method.

6. An information processing system including an information processing device, an answer generation step of generating an answer to a text input by a user using a first language model and presenting the answer to the text to the user; a storage step of storing the text and the answer generated in the generation step as a log; an acquisition step of acquiring the log at a predetermined timing; a summary generation step of extracting the acquired log at a preset rate and generating a summary using a second language model; and In the summary generating step, if a past summary is stored in the storage unit, a summary is generated based on the past summary in addition to the log and stored in the storage unit, and the past summary is deleted from the storage unit. Information processing system.

Images