Proxy presence model production method, proxy presence model production device, proxy action device, proxy action method, proxy action program, and recording medium

Abstract

Provided is a proxy presence model production method capable of producing a model reflecting the idea of an information producer.　A proxy presence model producing method according to the present disclosure comprises a knowledge information acquisition step, a construction information extraction step, and a model construction step. The knowledge information acquisition step acquires knowledge information of a subject, the construction information extraction step extracts personality information of the subject from the knowledge information, and the model construction step constructs a proxy presence model imitating the subject on the basis of the personality information.

Description

Method for manufacturing an agent presence model, apparatus for manufacturing an agent presence model, agent behavior device, agent behavior method, agent behavior program, and recording medium

[0001] The present disclosure relates to a method for manufacturing an agent presence model, an apparatus for manufacturing an agent presence model, an agent behavior device, an agent behavior method, an agent behavior program, and a recording medium.

[0002] In recent years, research has been conducted on constructing dialogue systems such as chatbots using machine-learned models. For example, in Patent Document 1, an acquisition unit that acquires an inquiry sentence, and a learned answer sentence for a positive example for a learning inquiry sentence and a learned answer sentence for a negative example sampled according to the frequency distribution of the learned answer sentence are used, so that the similarity between the learning inquiry sentence and the learned answer sentence for the positive example is large, and the similarity between the learning inquiry sentence and the learned answer sentence for the negative example is small. A question-and-answer device is described that includes a calculation unit that calculates the similarity between the inquiry sentence and a plurality of candidate answer sentences, and an extraction unit that extracts one or a plurality of candidate answer sentences from the plurality of candidate answer sentences based on the similarity.

[0003] Japanese Patent Application Laid-Open No. 2021-124824

[0004] In the invention as described in Patent Document 1, learning is performed by giving information such as a specialized book so that answers to questions can be generated in advance. However, since the information input for learning contains the thoughts of the information producer (author) and objective knowledge (formal knowledge), the information learned by the model may be in an uncertain state. Here, if only formal knowledge is extracted in advance from information such as a specialized book and learning is performed, there is a problem that the author's thoughts are not reflected and the handling of formal knowledge in the learned model becomes opaque.

[0005] Therefore, an object of the present disclosure is to provide a method for manufacturing an agent presence model capable of manufacturing a model that reflects the thoughts of the information producer.

[0006] To achieve the aforementioned objective, the method for manufacturing a surrogate existence model according to this disclosure includes a knowledge information acquisition step, a construction information extraction step, and a model construction step, wherein each step is performed by a computer, the knowledge information acquisition step involves acquiring knowledge information of a subject, the construction information extraction step involves extracting personality information of the subject from the knowledge information, and the model construction step involves constructing a surrogate existence model that mimics the subject based on the personality information.

[0007] The proxy existence model manufacturing apparatus of this disclosure includes a knowledge information acquisition unit, a construction information extraction unit, and a model construction unit, wherein the knowledge information acquisition unit acquires knowledge information of a subject, the construction information extraction unit extracts personality information of the subject from the knowledge information, and the model construction unit constructs a proxy existence model that mimics the subject based on the personality information.

[0008] The proxy action device of this disclosure includes an input receiving unit and a proxy action output unit, wherein the input receiving unit receives input information to a proxy existence model that mimics a target person, and the proxy action output unit inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

[0009] The proxy action method of this disclosure includes an input reception step and a proxy action output step, wherein each step is performed by a computer, the input reception step receives input information for a proxy existence model that mimics a target person, and the proxy action output step inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

[0010] The proxy action program of this disclosure includes an input reception procedure and a proxy action output procedure, wherein the input reception procedure receives input information to a proxy existence model that mimics a target person, and the proxy action output procedure inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information; the program is designed to cause a computer to execute each of these procedures.

[0011] The recording medium of this disclosure is a computer-readable recording medium that records a proxy action program for causing a computer to execute each of the following procedures: the input receiving procedure receives input information to a proxy existence model that mimics a target person; and the proxy action output procedure inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

[0012] According to this disclosure, it is possible to manufacture models that reflect the ideas of the information's creator.

[0013] Figure 1 is a block diagram showing the configuration of an example of the proxy existence model manufacturing apparatus of this disclosure. Figure 2 is a block diagram showing an example of the hardware configuration of the proxy existence model manufacturing apparatus of this disclosure. Figure 3 is a flowchart showing an example of processing in the proxy existence model manufacturing apparatus of this disclosure. Figure 4 is a block diagram showing the configuration of an example of the proxy behavior apparatus of this disclosure. Figure 5 is a block diagram showing an example of the hardware configuration of the proxy behavior apparatus of this disclosure. Figure 6 is a flowchart showing an example of processing in the proxy behavior apparatus of this disclosure. Figure 7 is a schematic diagram illustrating an example of a method for manufacturing a proxy existence model in the proxy existence model manufacturing apparatus of this disclosure.

[0014] Next, embodiments of the present disclosure will be described with reference to the drawings. The present disclosure is not limited to the following embodiments. In the following drawings, the same parts are denoted by the same reference numerals. Furthermore, unless otherwise specified, the descriptions of each embodiment can be used interchangeably with those of the others, and unless otherwise specified, the configurations of each embodiment can be combined.

[0015] [Embodiment 1] The proxy existence model manufacturing apparatus of this embodiment will be described with reference to Figure 1. Figure 1 is a block diagram showing the configuration of an example of the proxy existence model manufacturing apparatus 10 of this embodiment. As shown in Figure 1, the proxy existence model manufacturing apparatus 10 (hereinafter also referred to as "this apparatus 10") includes a knowledge information acquisition unit 11, a construction information extraction unit 12, and a model construction unit 13. In addition, although not shown, this apparatus 10 may also include, for example, an input unit, an output unit, a display unit and / or a storage unit.

[0016] The device 10 may be, for example, a single device including the aforementioned parts, or it may be a device in which the aforementioned parts can be connected via a communication network. Furthermore, the device 10 can be connected to an external device described later via a communication network. The communication network is not particularly limited and a known network can be used, for example, it may be wired or wireless. Examples of communication networks include the Internet, WWW (World Wide Web), telephone lines, LAN (Local Area Network), SAN (Storage Area Network), DTN (Delay Tolerant Networking), LPWA (Low Power Wide Area), L5G (Local 5G), etc. Examples of wireless communication include Wi-Fi®, Bluetooth®, Local 5G, LPWA, etc. The aforementioned wireless communication may be in the form of direct communication between devices (Ad Hoc communication), infrastructure communication, or indirect communication via an access point. The device 10 may, for example, be incorporated into a server as a system. The device 10 may also be, for example, a personal computer (PC, e.g., desktop or notebook), smartphone, tablet terminal, etc., on which the program disclosed herein is installed. Furthermore, the device 10 may be in the form of cloud computing or edge computing, for example, in which at least one of the aforementioned parts is on a server and the other parts are on a terminal.

[0017] Figure 2 illustrates a block diagram of the hardware configuration of the device 10. The device 10 includes, for example, a central processing unit 101, memory 102, bus 103, storage device 104, input device 105, output device 106, communication device (communication unit) 107, etc. Each part of the device 10 is interconnected via the bus 103 through its respective interface (I / F).

[0018] The central processing unit 101 operates in coordination with other components via controllers (system controller, I / O controller, etc.) and is responsible for the overall control of the device 10. In the device 10, the central processing unit 101 executes, for example, the program disclosed herein and other programs, and also reads and writes various types of information. Specifically, for example, the central processing unit 101 functions as a knowledge information acquisition unit 11, a construction information extraction unit 12, and a model construction unit 13. The device 10 may also include other computing devices such as a CPU, GPU (Graphics Processing Unit), APU (Accelerated Processing Unit), or a combination thereof as computing devices.

[0019] Bus 103 can also be connected to external devices, for example. Examples of such external devices include external storage devices (external databases, etc.), electrocardiographs, printers, external input devices, external display devices, audio output devices such as speakers, external imaging devices such as cameras, and various sensors such as acceleration sensors, geomagnetic sensors, and direction sensors. The device 10 can be connected to an external network (the aforementioned communication network) by a communication device 107 connected to bus 103, for example, and can also be connected to other devices via the external network.

[0020] Memory 102 may be, for example, main memory. When the central processing unit 101 performs processing, memory 102 reads various operational programs, such as the program of this disclosure, stored in the storage device 104 (described later), and the central processing unit 101 receives data from memory 102 and executes the program. The main memory may be, for example, RAM (random access memory). Alternatively, memory 102 may be, for example, ROM (read-only memory).

[0021] The storage device 104 is also called an auxiliary storage device, for example, in relation to the main memory (primary memory). As described above, the storage device 104 stores an operating program including the program of this disclosure. The storage device 104 may be, for example, a combination of a recording medium and a drive for reading and writing to the recording medium. The recording medium is not particularly limited and may be internal or external, for example, an HD (hard disk), CD-ROM, CD-R, CD-RW, MO, DVD, flash memory, memory card, etc. The storage device 104 may be, for example, a hard disk drive (HDD) in which the recording medium and the drive are integrated, or a solid state drive (SSD). If the device 10 includes, for example, the storage device 104 functions as the storage unit.

[0022] In this device 10, the memory 102 and storage device 104 can also store various types of information, such as log information, information obtained from an external database (not shown) or external devices, information generated by this device 10, and information used by this device 10 when executing processing. At least some of the information may be stored on an external server other than the memory 102 and storage device 104, or it may be stored in a distributed manner across multiple terminals using blockchain technology or the like.

[0023] The device 10 further includes, for example, an input device 105 and an output device 106. The input device 105 may include, for example, a pointing device such as a touch panel, trackpad, or mouse; a keyboard; imaging means such as a camera or scanner; a card reader such as an IC card reader or magnetic card reader; an audio input means such as a microphone; and so on. The output device 106 may include, for example, a display device such as an LED display or liquid crystal display; an audio output device such as a speaker; a printer; and so on. In this embodiment 1, the input device 105 and the output device 106 are configured separately, but the input device 105 and the output device 106 may be configured as an integrated unit, such as a touch panel display.

[0024] Next, an example of the proxy existence model manufacturing method of this embodiment will be described based on the flowchart in Figure 3. The proxy existence model manufacturing method of this embodiment can be carried out as follows, for example, using the proxy existence model manufacturing apparatus 10 shown in Figures 1 and 2. Note that the proxy existence model manufacturing method of this embodiment is not limited to the use of the proxy existence model manufacturing apparatus 10 shown in Figures 1 and 2.

[0025] First, the knowledge information acquisition unit 11 acquires the subject's knowledge information (S1, knowledge information acquisition step). The format of the knowledge information is not particularly limited; for example, it may be text information, image information, audio information, or a combination thereof. The knowledge information is, for example, information linked to predetermined information and subject identification information that identifies the creator of the information (the subject). The predetermined information is, for example, information that includes the subject's personality information and explicit knowledge information. The personality information is, for example, information that represents the subject's thoughts from the knowledge information. The personality information is also called, for example, a partial stance. The explicit knowledge information is, for example, the part of the knowledge information excluding the personality information, and includes objective knowledge. If the knowledge information is a book, paper, etc., the explicit knowledge information may include, but is not limited to, information such as technical terms and experimental results. The subject identification information may include, for example, name, address, telephone number, email address, identification number (for example, My Number (individual number), etc.). Specific examples of the aforementioned knowledge information include, but are not limited to, books and papers written by the subject, video data of lectures given by the subject, audio data of lectures given by the subject, and image data created by the subject. The knowledge information acquisition unit 11 may, for example, acquire knowledge information recorded in the storage unit of the device 10, or it may acquire the aforementioned knowledge information from outside the device 10 via the input device 105. The knowledge information acquisition unit 11 may, for example, acquire one type of knowledge information of the subject, or it may acquire two or more types. The knowledge information acquisition unit 11 may, for example, record the acquired knowledge information in the storage unit of the device 10.

[0026] The information extraction unit 12 for construction extracts personality information of the subject from the knowledge information (S2, information extraction step for construction). The information extraction unit 12 for construction may also, for example, further extract explicit knowledge information from the knowledge information of the subject. The information extraction unit 12 for construction may, for example, use known natural language processing techniques to extract at least one of the personality information and explicit knowledge information of the subject, or use a large-scale language model to extract at least one of the personality information and explicit knowledge information of the subject. If the knowledge information is textual information such as a book or a paper, the information extraction unit 12 for construction can extract the personality information or explicit knowledge information based, for example, on the end of sentences in a document or the chapter structure of a book. The information extraction unit 12 for construction may, for example, record at least one of the extracted personality information and explicit knowledge information in the storage unit of the device 10. In this case, the information extraction unit 12 for construction can, for example, record the personality information and explicit knowledge information linked to the knowledge information from which they were extracted and the creator identification information of the knowledge information.

[0027] The information extraction unit 12 for construction can, for example, analyze the knowledge information and extract sentences whose sentence ends with a word that expresses the author's thoughts as the personality information. Examples of words that express the author's thoughts include, but are not limited to, words such as "I want to," "I think," "I believe," and "I want." The information extraction unit 12 for construction can also analyze the knowledge information and extract sentences whose sentence ends with a word that indicates explicit knowledge as the explicit knowledge information. Examples of words that indicate explicit knowledge include, but are not limited to, words such as "It is," "It was," and "As a result."

[0028] The information extraction unit 12 for construction may, for example, analyze the knowledge information and extract sentences contained in chapters that describe the author's ideas as personality information. Examples of chapters that describe the author's ideas include the "preface," "introduction," and "afterword."

[0029] The information acquisition unit 12 for construction may, for example, use AI to extract personality information and explicit knowledge information from knowledge information. In this disclosure, AI may refer to, for example, a large-scale learning model called a "foundation model." The foundation model is a machine learning model pre-trained on predetermined big data and is not limited to a large-scale language model (LLM) that has learned natural language, but may also include a large-scale model for speech, a large-scale model for images, and a multimodal model (such as a visual language model) that handles language, images, speech, and video across the board. Furthermore, a configuration may be adopted in which a small-scale language model (SLM) is placed on the terminal side and cooperates with the large-scale model on the cloud side, a configuration that includes search extension generation (RAG) using an external knowledge source, tool execution / function call, agent-oriented control logic, etc. The providers of large-scale language models are not particularly limited. Examples include, but are not limited to, various LLM / multimodal models provided by companies such as OpenAI, Anthropique, Alphabet (Google), META, Microsoft, Cohere, Mistral, xAI, NEC Corporation, and NTT.

[0030] When the information acquisition unit 12 for construction uses AI to extract personality information and explicit knowledge information from knowledge information, the information acquisition unit 12 can cause the AI ​​to extract the user's personality information and explicit knowledge information by inputting extraction instruction information, which instructs the AI ​​to extract personality information and explicit knowledge information, along with the knowledge information. The extraction instruction information may be recorded in the storage unit of the device 10, stored externally, or input by the user each time. Specific examples of the extraction instruction information include, for example, "classify the information into the following two categories: sentences and paragraphs that show the author's thoughts and opinions, and sentences and paragraphs that show knowledge, such as scientific verification results and historical facts." This disclosure is not limited to the above examples.

[0031] Furthermore, the AI ​​may be, for example, a model finely tuned to extract personality information and explicit knowledge information from knowledge information. In this case, for example, fine tuning for extracting personality information and explicit knowledge information from knowledge information can be performed by training the AI ​​with knowledge information and a set of personality information and explicit knowledge information previously extracted from the knowledge information. In this case, the AI ​​may be, for example, a multilayer neural network model having a large number of parameters (e.g., 100,000 or more, and in some cases, several million or more). Such fine tuning processing requires the rapid and repeated execution of a huge amount of numerical calculations, making it practically impossible for a human to perform it in their head or with paper and pencil, and thus relies on automated processing by electronic computers such as processors and GPUs. During the fine tuning, in order to suppress overfitting of the training data, known regularization methods such as L2 regularization, dropout, and early stopping may be combined and applied. This makes it possible to obtain a model that can extract personality information and explicit knowledge information with high generalization performance even from unknown knowledge information, and improves the accuracy and robustness of information extraction compared to simple threshold judgment or rule-based processing.

[0032] Furthermore, the information extraction unit 12 for construction may extract the personality information of the subject by, for example, providing a large-scale language model with the subject's knowledge information and instruction information that instructs the model to extract the subject's personality information from the knowledge information based on the subject's knowledge information, thereby causing the model to extract the personality information from the knowledge information. The large-scale language model is not particularly limited and includes, but is not limited to, OpenAI®'s GPT-3®, GPT-4®, Alphabet Inc. (Google®)'s BERT, LaMDA, PaLM2, META®'s LlaMA, NEC Corporation's LLM, NTT®'s LLM, etc. The instruction information that instructs the model to extract the subject's personality information from the knowledge information based on the subject's knowledge information is not particularly limited as long as it is a document that instructs the model to divide the knowledge information into parts that describe thoughts and parts that describe knowledge. Specific examples of instruction information that instructs the extraction of personality information of the subject from the knowledge information of the subject include, but are not limited to, documents such as, "Classify the information into the following two categories: - Sentences and paragraphs that show the author's thoughts and opinions, such as the author's ideas; - Sentences and paragraphs that show knowledge, such as scientific verification results and historical facts."

[0033] The information extraction unit 12 for construction may, for example, preprocess the knowledge information and extract personality information and explicit knowledge information. Specifically, for example, the information extraction unit 12 for construction may convert the knowledge information into a vectorized embedding vector sequence for each unit text, and then classify or group the knowledge information into personality information and explicit knowledge information based on the embedding vector sequence, thereby extracting personality information and explicit knowledge information based on the knowledge information. The embedding vectors may, for example, contain hundreds to thousands of dimensions of real-valued elements for each unit text. For this reason, for example, a large number of embedding vectors are generated for the entire knowledge information, and the classification or grouping process consists of a large number of numerical operations, mainly matrix operations, which is a process that is practically impossible for a human to perform in their head or with paper and pencil.

[0034] The model building unit 13 constructs a surrogate existence model that mimics the target person based on the personality information (S3, model building process). The model building unit 13 can construct the surrogate existence model by, for example, providing a large-scale learning model with the personality information and instruction information that instructs it to construct a surrogate existence model that mimics the target person based on the personality information. The large-scale learning model is, for example, a machine learning model that has been trained using predetermined big data. The large-scale learning model may be, for example, a model that has been trained on big data of natural language (large-scale language model), a model that has been trained on big data of speech (large-scale speech model), or a model that has been trained on big data of images (large-scale image model). The model building unit 13 can construct the surrogate existence model by, for example, providing a large-scale language model, as the large-scale learning model, with the personality information and instruction information that instructs it to construct a surrogate existence model that mimics the target person based on the personality information. The instruction information is, for example, instruction information (prompt) for generating the behavior of a person handling knowledge. The aforementioned instruction information could include, but is not limited to, documents such as, "When generating text, follow the rules below: - Describe examples that reflect personal information. - Do not reflect personal information in terms of knowledge. - For example, structure the text by stating a fact that is correct as knowledge, and then expressing an opinion on that fact as personal information."

[0035] Furthermore, the model building unit 13 may generate multiple sets of training data showing the correspondence between the subject's input information and responses based on the extracted personality information. The model building unit 13 may input the training data in mini-batch units into a large-scale learning model, calculate a loss function based on the error between the model's output for the training data and the response, and train a personality model capable of outputting behavior that mimics the subject by iteratively updating a large number of model parameters using gradient descent or a modified algorithm thereof.

[0036] Furthermore, the model building unit 13 may, for example, construct an explicit knowledge model by further training the explicit knowledge information onto a large-scale language model. The model building unit 13 can construct the explicit knowledge model by, for example, providing the explicit knowledge information to the large-scale language model and fine-tuning it. The explicit knowledge model is also called, for example, a large-scale language model with domain knowledge.

[0037] The aforementioned surrogate existence model is, for example, a model that has learned the personality information of the subject from the information that constitutes the knowledge information. Therefore, the surrogate existence model has learned, for example, the subject's way of thinking and understanding when handling knowledge, values ​​such as thoughts and beliefs, and how they interact with others (personality). For this reason, the surrogate existence model manufacturing method of this disclosure makes it possible to easily manufacture a model that reflects the thoughts of the information creator. Furthermore, the surrogate existence model manufactured by the surrogate existence model manufacturing method of this disclosure is capable of outputting products that are characteristic of the subject. In other words, according to this disclosure, for example, it becomes possible to construct a model that suppresses personal hallucination. Personal hallucination refers to, for example, a hallucination (illusion) of how to handle knowledge and output policies that the creator of the knowledge information would not say. Therefore, according to the surrogate existence model of this disclosure, for example, it becomes possible to more accurately extract and output the knowledge that knowledge information (e.g., a book) explicitly or implicitly contains. The output that is characteristic of the target person is not particularly limited and may be, for example, text output, voice output, or instructions to a designated machine.

[0038] The model building unit 13 may, for example, use extracted personality information to train a personality model capable of outputting behavior that mimics the subject by iteratively updating the parameters of a large-scale learning model having many parameters based on gradient descent to reproduce the correspondence between the subject's past input information and responses. Similarly, the model building unit 13 may, for example, use extracted explicit knowledge information to train an explicit knowledge model that outputs explicit knowledge information contained in the knowledge information. In this case, the model building unit 13 can, for example, generate proxy behavior information consistent with the subject's personality and output the proxy behavior information by coordinating the operation of the trained personality model and the explicit knowledge model. By configuring the personality model and the explicit knowledge model separately in this way, and coordinating the output of explicit knowledge based on the knowledge information with the output based on the personality information, it is possible to reduce the inclusion of personal expressions not included in the knowledge information and improve the consistency and reliability of the generated responses, compared to, for example, simply providing prompts to a general-purpose large-scale language model.

[0039] According to this disclosure, it is possible to create a model that reflects the thoughts of the information creator. Therefore, according to this disclosure, for example, it is possible to construct a surrogate model for individuals with limited human resources (e.g., busy researchers, supervisors, managers, teachers, etc.), thereby reducing the burden on those individuals.

[0040] [Embodiment 2] Embodiment 2 is an example of the proxy action device of the present disclosure.

[0041] The proxy action device of this embodiment will be described with reference to Figure 4. Figure 4 is a block diagram showing the configuration of an example of the proxy action device 20 of this embodiment. As shown in Figure 4, the proxy action device 20 (hereinafter also referred to as "this device 20") includes an input receiving unit 21 and a proxy action output unit 22. In addition, although not shown, this device 20 may also include, for example, an input unit, an output unit, a display unit and / or a storage unit.

[0042] The proxy action device 20 may be, for example, a single device including the aforementioned parts, or each of the aforementioned parts may be a device that can be connected via a communication network. Furthermore, the proxy action device 20 can be connected to an external device described later via a communication network. The communication network is not particularly limited and a known network can be used, for example, it may be wired or wireless. Examples of communication networks include the Internet, WWW (World Wide Web), telephone lines, LAN (Local Area Network), SAN (Storage Area Network), DTN (Delay Tolerant Networking), LPWA (Low Power Wide Area), L5G (Local 5G), etc. Examples of wireless communication include Wi-Fi®, Bluetooth®, local 5G, and LPWA. The wireless communication may be in the form of direct communication between devices (Ad Hoc communication), infrastructure communication, or indirect communication via an access point. The proxy action device 20 may be incorporated into a server as a system, for example. Alternatively, the proxy action device 20 may be a personal computer (PC, e.g., desktop or notebook), smartphone, or tablet terminal on which the program disclosed herein is installed. Furthermore, the proxy action device 20 may be in the form of cloud computing or edge computing, for example, in which at least one of the aforementioned parts is on a server and the other parts are on a terminal.

[0043] Figure 5 illustrates a block diagram of the hardware configuration of the proxy action device 20. As shown in Figure 5, the proxy action device 20 includes, for example, a central processing unit 201, a memory 202, a bus 203, a storage device 204, an input device 205, an output device 206, a communication device 207, etc. The explanation of each component of the proxy action device 20 can be made by referring to the explanation of each component of the proxy existence model manufacturing device 10. Each part of the proxy action device 20 is connected via the bus 203 by its respective interface (I / F). In the proxy action device 20, the central processing unit 201 functions as an input receiving unit 21 and a proxy action output unit 22.

[0044] Next, an example of the proxy behavior method of this embodiment will be described based on the flowchart of FIG. 6. The proxy behavior method of this embodiment is implemented as follows, for example, using the proxy behavior device 20 shown in FIGS. 4 and 5. Note that the proxy behavior method of this embodiment is not limited to the use of the proxy behavior device 20 shown in FIGS. 4 and 5.

[0045] The input reception unit 21 receives input information for the proxy presence model that mimics the target person (S21, input reception step). The input information may be, for example, character information (text information), image information, voice information, or a combination thereof. The input reception unit 21 receives, for example, text information for the target person as the input information. The input information may be, for example, a question for the target person mimicked by the proxy presence model.

[0046] The proxy behavior output unit 22 inputs the input information into the proxy presence model that mimics the target person to generate the proxy behavior information of the target person in the proxy presence model, and outputs the proxy behavior information (S22, proxy behavior output step). The proxy behavior output unit 22 may, for example, generate the proxy speech text of the target person based on the text information in the proxy presence model as the proxy behavior information and output the proxy speech text. When the input information is a question, the proxy behavior output unit 22 may, for example, generate an answer to the question in the proxy presence model as the proxy behavior information and output the answer. The proxy behavior output unit 22 may, for example, generate instruction information for instructing the proxy presence model to generate an answer to the question from another large language model as the proxy behavior information, input the instruction information into the large language model to generate an answer from the large language model, and output the answer. In this case, the proxy behavior output unit 22 may input, for example, the answer generated by the large language model into the proxy presence model as additional input information, correct the answer generated by the large language model to an answer output that seems to be the target person, and output the corrected answer as the proxy behavior information.

[0047] The proxy action output unit 22 may, for example, output the proxy action information with an authenticity display image attached thereto. The authenticity display image may be, for example, an image indicating that the proxy action information is an output by a proxy existence model. The shape, color, size, etc. of the authenticity display image are not particularly limited, and any image can be used. The authenticity display image is also referred to as an authentication badge, for example. The proxy existence model of the present disclosure reveals the history of the knowledge information underlying the personality information or formal knowledge information used in learning. For this reason, the proxy existence model of the present disclosure reveals, for example, the history of the information sources used in learning. Therefore, regarding the proxy action information generated by the proxy existence model of the present disclosure, since the reliability of the information sources used in learning is ensured, by attaching an authenticity display image, it is possible to give the user a sense of security regarding the generated data.

[0048] A specific example of the proxy existence model manufacturing method and the proxy action method of the present disclosure will be described with reference to FIG. 7. In the following description, as the knowledge information, when constructing a proxy existence model of the author of the specialized book A and a large-scale language model with domain knowledge (formal knowledge model) that has learned the knowledge of the specialized book A using the specialized book A, this will be described as an example, but the present disclosure is not limited to the following description in any way.

[0049] As shown in FIG. 7, first, the specialized book A is scanned and converted into electronic data (text data) that is knowledge information. Then, the knowledge information and a prompt for dividing the document into knowledge and the author's thoughts are given to a large-scale language model, and from the knowledge information, partial stances (author's personality information) and specialized formal knowledge (formal knowledge information) are extracted. Next, the extracted personality information and a prompt for generating the behavior of a person who handles knowledge are given to a large-scale language model, and a proxy existence model that has learned the author's portrait is constructed. Also, by giving the extracted formal knowledge information to a large-scale language model and fine-tuning the large-scale language model, a formal knowledge model (large-scale language model with domain knowledge) capable of generating a document based on the knowledge of the specialized book A can be constructed.

[0050] Users of the proxy existence model input questions to the model as input information. The proxy existence model, acting as a proxy for the subject on which the proxy existence model is based, directly generates answers that mimic the subject, thereby simulating a conversation (text chat, etc.) with the subject. Alternatively, the proxy existence model may generate prompts to a large-scale language model with domain knowledge to provide appropriate answers to the questions, and the large-scale language model with domain knowledge may generate answers based on these prompts.

[0051] The above example illustrates a case where text is input as input information and text generated from the input information is output as proxy action information, but this disclosure is not limited to this example.

[0052] [Embodiment 3] The proxy action program of this embodiment is a program that causes a computer to execute each step of the proxy action method described above. Specifically, the proxy action program of this embodiment is a program that causes a computer to execute an input acceptance procedure and a proxy action output procedure.

[0053] The input reception procedure receives input information for a proxy existence model that mimics the target person, and the proxy action output procedure inputs the input information into the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

[0054] Furthermore, the proxy action program of this embodiment can also be described as a program that causes a computer to function as an input receiving procedure and a proxy action output procedure.

[0055] The proxy action program of this embodiment can be based on the descriptions in the proxy action device and proxy action method of the present disclosure. Each of the aforementioned steps can be read as, for example, "step" or "process". The program of this embodiment may also be recorded on, for example, a computer-readable recording medium. The recording medium is not particularly limited and includes, for example, random access memory (RAM), read-only memory (ROM), hard disk (HD), flash memory (e.g., SSD (Solid State Drive), USB flash memory, SD / SDHC card, etc.), optical disc (e.g., CD-R / CD-RW, DVD-R / DVD-RW, BD-R / BD-RE, etc.), magneto-optical disk (MO), floppy disk (FD), etc. The proxy action program of this embodiment (also referred to as, for example, a programming product or a proxy action program product) may also be distributed, for example, from an external computer. The aforementioned "distribution" may be, for example, distribution via a communication network, or distribution via a wired connected device. The proxy action program of this embodiment may be installed and executed on the distributed device, or it may be executed without being installed.

[0056] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure are possible, which can be understood by those skilled in the art within the scope of the present disclosure.

[0057] This application claims priority based on Japanese Patent Application No. 2024-229564, filed on 26 December 2024, and incorporates all of its disclosures herein.

[0058] <Notes> Some or all of the above embodiments may be described as follows, but are not limited to the following. (Note 1) A method for manufacturing a surrogate existence model, comprising a knowledge information acquisition step, a construction information extraction step, and a model construction step, wherein the knowledge information acquisition step acquires knowledge information of a subject, the construction information extraction step extracts personality information of the subject from the knowledge information, and the model construction step constructs a surrogate existence model that mimics the subject based on the personality information, with each step being executed by a computer. (Note 2) The method for manufacturing a surrogate existence model according to Note 1, wherein the model construction step causes a large-scale learning model to construct a surrogate existence model by providing it with the personality information and instruction information that instructs it to construct a surrogate existence model that mimics the subject based on the personality information. (Note 3) The method for manufacturing a surrogate existence model according to Note 2, wherein the model construction step involves providing a large-scale language model with the personality information and instruction information that instructs the large-scale language model to construct a surrogate existence model that mimics the target person based on the personality information, thereby causing the surrogate existence model to be constructed. (Note 4) The method for manufacturing a surrogate existence model according to Note 3, wherein the information extraction step for construction involves providing a large-scale language model with the knowledge information of the target person and instruction information that instructs the large-scale language model to extract the personality information of the target person from the knowledge information based on the knowledge information of the target person, thereby causing the surrogate existence model to be extracted from the knowledge information. (Note 5) The method for manufacturing a surrogate existence model according to Note 3 or 4, wherein the information extraction step for construction involves extracting explicit knowledge information from the knowledge information of the target person, and the model construction step involves providing a large-scale language model with an explicit knowledge model that has been further trained with the explicit knowledge information. (Appendix 6) A proxy existence model manufacturing apparatus comprising a knowledge information acquisition unit, a construction information extraction unit, and a model construction unit, wherein the knowledge information acquisition unit acquires knowledge information of a subject, the construction information extraction unit extracts personality information of the subject from the knowledge information, and the model construction unit constructs a proxy existence model that mimics the subject based on the personality information.(Note 7) The proxy existence model manufacturing apparatus according to Note 6, wherein the model construction unit causes the proxy existence model to be constructed by providing the large-scale learning model with the personality information and instruction information that instructs the large-scale learning model to construct a proxy existence model that mimics the target person based on the personality information. (Note 8) The proxy existence model manufacturing apparatus according to Note 7, wherein the model construction unit causes the proxy existence model to be constructed by providing the large-scale language model, as the large-scale learning model, with the personality information and instruction information that instructs the large-scale language model to construct a proxy existence model that mimics the target person based on the personality information. (Note 9) The proxy existence model manufacturing apparatus according to Note 8, wherein the construction information extraction unit extracts the personality information of the target person by providing the large-scale language model with the knowledge information of the target person and instruction information that instructs the large-scale language model to extract the personality information of the target person from the knowledge information based on the knowledge information of the target person. (Note 10) The proxy existence model manufacturing apparatus according to Note 8 or 9, wherein the information extraction unit for construction extracts explicit knowledge information from the subject's knowledge information, and the model construction unit constructs an explicit knowledge model by further training a large-scale language model with the explicit knowledge information. (Note 11) A proxy action device including an input receiving unit and a proxy action output unit, wherein the input receiving unit receives input information for a proxy existence model that mimics a subject, and the proxy action output unit inputs the input information to the proxy existence model that mimics a subject, causing the proxy existence model to generate proxy action information for the subject, and outputs the proxy action information. (Note 12) The proxy action device according to Note 11, wherein the input receiving unit receives text information for a subject as the input information, and the proxy action output unit causes the proxy existence model to generate proxy statement text for the subject based on the text information, and outputs the proxy statement text as the proxy action information. (Note 13) The proxy action device according to Note 12, wherein the input receiving unit receives a question to the subject as input information, and the proxy action output unit causes the proxy existence model to generate an answer to the question and outputs the answer as proxy action information.(Note 14) The proxy action device according to Note 12, wherein the input receiving unit receives a question to the target person as input information, the proxy action output unit generates instruction information as proxy action information, instructing the proxy existence model to generate an answer to the question from another large-scale language model, inputting the instruction information to the large-scale language model to generate an answer from the large-scale language model, and outputting the answer. (Note 15) A proxy action method comprising an input receiving step and a proxy action output step, wherein the input receiving step receives input information to a proxy existence model that mimics the target person, and the proxy action output step inputs the input information to the proxy existence model that mimics the target person to generate proxy action information for the target person, and outputs the proxy action information, each step being performed by a computer. (Note 16) The proxy action method according to Note 15, wherein the input receiving step receives text information for the target person as input information, and the proxy action output step causes the proxy existence model to generate a proxy statement text for the target person based on the text information as proxy action information, and outputs the proxy statement text. (Note 17) The proxy action method according to Note 16, wherein the input receiving step receives a question for the target person as input information, and the proxy action output step causes the proxy existence model to generate an answer to the question as proxy action information, and outputs the answer. (Note 18) The proxy action method according to Note 16, wherein the input receiving step receives a question for the target person as input information, and the proxy action output step generates instruction information as proxy action information that instructs the proxy existence model to generate an answer to the question from another large-scale language model, inputs the instruction information to the large-scale language model to generate an answer from the large-scale language model, and outputs the answer.(Note 19) A proxy action program that causes a computer to execute each of the following procedures, including an input reception procedure and a proxy action output procedure, wherein the input reception procedure receives input information for a proxy existence model that mimics a target person, and the proxy action output procedure inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information. (Note 20) The proxy action program according to Note 19, wherein the input reception procedure receives text information for a target person as the input information, and the proxy action output procedure causes the proxy existence model to generate proxy statement text for the target person based on the text information, and outputs the proxy statement text, as the proxy action information. (Note 21) The proxy action program according to Note 20, wherein the input reception procedure receives a question for a target person as the input information, and the proxy action output procedure causes the proxy existence model to generate an answer to the question, and outputs the answer, as the proxy action information. (Note 22) The proxy action program described in Note 20, wherein the input receiving procedure receives a question to the subject as input information, the proxy action output procedure generates instruction information to instruct the proxy existence model to generate an answer to the question from another large-scale language model as proxy action information, the instruction information is input to the large-scale language model to generate an answer from the large-scale language model, and the answer is output. (Note 23) A computer-readable recording medium that includes an input receiving procedure and a proxy action output procedure, wherein the input receiving procedure receives input information to a proxy existence model that mimics the subject, the proxy action output procedure inputs the input information to the proxy existence model that mimics the subject to generate proxy action information for the subject, and outputs the proxy action information. (Note 24) The recording medium described in Note 23, wherein the input reception procedure receives text information for the subject as input information, and the proxy action output procedure causes the proxy existence model to generate proxy statement text for the subject based on the text information as proxy action information, and outputs the proxy statement text.(Note 25) The recording medium described in Note 24, wherein the input reception procedure receives a question to the subject as input information, and the proxy action output procedure causes the proxy existence model to generate an answer to the question as proxy action information, and outputs the answer. (Note 26) The recording medium described in Note 24, wherein the input reception procedure receives a question to the subject as input information, and the proxy action output procedure causes the proxy existence model to generate instruction information instructing another large-scale language model to generate an answer to the question as proxy action information, inputs the instruction information to the large-scale language model to generate an answer from the large-scale language model, and outputs the answer.

[0059] This disclosure enables the creation of models that reflect the intentions of the information creators. Therefore, this disclosure can reduce the workload for individuals with limited human resources. For this reason, this disclosure is useful in a wide range of industries, for example.

[0060] 10 Proxy existence model manufacturing device 11 Knowledge information acquisition unit 12 Construction information extraction unit 13 Model construction unit 14 Recurrence prediction unit 101 Central processing unit 102 Memory 103 Bus 104 Storage device 105 Input device 106 Output device 107 Communication device 20 Proxy action device 21 Input reception unit 22 Proxy action output unit 201 Central processing unit 202 Memory 203 Bus 204 Storage device 205 Input device 206 Output device 207 Communication device

Claims

1. A method for manufacturing a surrogate existence model, comprising a knowledge information acquisition step, a construction information extraction step, and a model construction step, wherein the knowledge information acquisition step acquires knowledge information of a subject, the construction information extraction step extracts personality information of the subject from the knowledge information, and the model construction step constructs a surrogate existence model that mimics the subject based on the personality information, with each step being executed by a computer.

2. The method for manufacturing a surrogate existence model according to claim 1, wherein the model construction step involves providing a large-scale learning model with the personality information and instruction information that instructs the model to construct a surrogate existence model that mimics the target person based on the personality information, thereby causing the surrogate existence model to be constructed.

3. The method for manufacturing a surrogate existence model according to claim 2, wherein the model construction step involves providing a large-scale language model, as the large-scale learning model, with the personality information and instruction information that instructs the model to construct a surrogate existence model that mimics the target person based on the personality information, thereby causing the surrogate existence model to be constructed.

4. The method for manufacturing a surrogate existence model according to claim 3, wherein the information extraction step for construction is performed by providing a large-scale language model with knowledge information of the subject and instruction information that instructs the model to extract personality information of the subject from the knowledge information based on the knowledge information of the subject, thereby causing the model to extract personality information from the knowledge information.

5. The method for manufacturing a surrogate existence model according to claim 3 or 4, wherein the information extraction step for construction extracts explicit knowledge information from the subject's knowledge information, and the model construction step constructs an explicit knowledge model by further training a large-scale language model with the explicit knowledge information.

6. A proxy existence model manufacturing apparatus comprising a knowledge information acquisition unit, a construction information extraction unit, and a model construction unit, wherein the knowledge information acquisition unit acquires knowledge information of a subject, the construction information extraction unit extracts personality information of the subject from the knowledge information, and the model construction unit constructs a proxy existence model that mimics the subject based on the personality information.

7. The surrogate existence model manufacturing apparatus according to claim 6, wherein the model construction unit causes the surrogate existence model to be constructed by providing the large-scale learning model with the personality information and instruction information that instructs the model to construct a surrogate existence model that mimics the target person based on the personality information.

8. The surrogate existence model manufacturing apparatus according to claim 7, wherein the model construction unit causes the surrogate existence model to be constructed by providing the large-scale language model with the personality information and instruction information that instructs the large-scale language model to construct a surrogate existence model that mimics the target person based on the personality information.

9. The surrogate existence model manufacturing apparatus according to claim 8, wherein the construction information extraction unit extracts the personality information of the target person by providing the large-scale language model with knowledge information of the target person and instruction information that instructs the model to extract the personality information of the target person from the knowledge information based on the knowledge information of the target person, thereby causing the model to extract the personality information from the knowledge information.

10. The surrogate existence model manufacturing apparatus according to claim 8 or 9, wherein the information extraction unit for construction extracts explicit knowledge information from the subject's knowledge information, and the model construction unit constructs an explicit knowledge model by further training a large-scale language model with the explicit knowledge information.

11. A proxy action device comprising an input receiving unit and a proxy action output unit, wherein the input receiving unit receives input information for a proxy existence model that mimics a target person, and the proxy action output unit inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

12. The proxy action device according to claim 11, wherein the input receiving unit receives text information for the target person as input information, and the proxy action output unit causes the proxy existence model to generate proxy statement text for the target person based on the text information, and outputs the proxy statement text as proxy action information.

13. The proxy action device according to claim 12, wherein the input receiving unit receives a question to the subject as input information, and the proxy action output unit causes the proxy existence model to generate an answer to the question and outputs the answer as proxy action information.

14. The proxy action device according to claim 12, wherein the input receiving unit receives a question to the subject as input information, the proxy action output unit generates instruction information as proxy action information, instructing the proxy existence model to generate an answer to the question from another large-scale language model, inputs the instruction information to the large-scale language model to generate an answer from the large-scale language model, and outputs the answer.

15. A proxy action method comprising an input receiving step and a proxy action output step, wherein the input receiving step receives input information for a proxy existence model that mimics a target person, and the proxy action output step inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information, each step being performed by a computer.

16. The proxy action method according to claim 15, wherein the input receiving step receives text information for the target person as input information, and the proxy action output step causes the proxy existence model to generate proxy statement text for the target person based on the text information, and outputs the proxy statement text as proxy action information.

17. The proxy action method according to claim 16, wherein the input receiving step receives a question to the subject as input information, and the proxy action output step causes the proxy existence model to generate an answer to the question and outputs the answer as proxy action information.

18. The proxy action method according to claim 16, wherein the input receiving step receives a question to the subject as input information, the proxy action output step generates instruction information as proxy action information, instructing the proxy existence model to generate an answer to the question from another large-scale language model, inputs the instruction information to the large-scale language model to generate an answer from the large-scale language model, and outputs the answer.

19. A proxy action program that causes a computer to execute an input receiving procedure and a proxy action output procedure, wherein the input receiving procedure receives input information to a proxy existence model that mimics a target person, and the proxy action output procedure inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

20. A computer-readable recording medium that records a proxy action program for causing a computer to execute each of the following procedures: the input receiving procedure receives input information for a proxy existence model that mimics a target person, and the proxy action output procedure inputs the input information to the proxy existence model that mimics the target person, causing the proxy existence model to generate proxy action information for the target person, and outputs the proxy action information.

Images