Search data processing system, method for search data processing, and program

Abstract

To search for information in a short time even when information distributed across a large number of information sources is duplicated.SOLUTION: A search data processing system for registering search data according to one aspect of the present disclosure includes: a storage unit for storing a plurality of types of source information having different data formats; an extraction unit for extracting, from the source information stored in the storage unit, subject information indicating a subject of the source information; a classification unit for classifying the source information stored in the storage unit on the basis of the subject information extracted by the extraction unit; and a registration unit for registering, as the search data, the source information classified by the classification unit for each item.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] There are various types of data, such as books and internet information, which are distributed across a large number of information sources, and there are devices for easily searching for information. For example, an information search method described in Patent Document 1 stores a plurality of data items sorted according to a predetermined rule, specifies at least one of the sorted plurality of data items, extracts data densely from the sorted plurality of data items near the specified data item, and extracts data coarsely from positions far from the specified data item, and displays the extracted data. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 10-187748 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the information search method described in Patent Document 1 has the problem that when searching for information about a person or organization, if the information distributed across multiple information sources is duplicated, it takes a long time to search for information from a large amount of data.

[0005] The present disclosure has been made in consideration of the above circumstances, and aims to provide a search data processing system, a search data processing method, and a program that can search for information in a short time even when information distributed across multiple information sources is duplicated. [Means for solving the problem]

[0006] The present disclosure has been made to solve the above-mentioned problems, and one aspect of the present disclosure is a search data processing system for registering search data, comprising: a memory unit that stores multiple types of original information with different data formats; an extraction unit that extracts subject information indicating the subject of the original information from the original information stored in the memory unit; a classification unit that classifies the original information stored in the memory unit based on the subject information extracted by the extraction unit; and a registration unit that registers the original information classified by the classification unit as search data for each item.

[0007] Another aspect of the present disclosure is a search data processing method including the steps of: a search data processing system that registers search data storing multiple types of original information having different data formats in a memory unit; the search data processing system extracting subject information indicating the subject of the original information from the original information stored in the memory unit; the search data processing system classifying the original information stored in the memory unit based on the extracted subject information; and the search data processing system registering the classified original information as the search data for each item.

[0008] Another aspect of the present disclosure is a program that causes a computer of a search data processing system that registers search data to execute the steps of storing multiple types of original information with different data formats in a memory unit, extracting subject information indicating the subject of the original information from the original information stored in the memory unit, classifying the original information stored in the memory unit based on the extracted subject information, and registering the classified original information item by item as the search data. [Effects of the Invention]

[0009] According to one aspect of the present invention, even when information distributed across multiple information sources is duplicated, it is possible to search for information in a short time. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a block diagram showing an example of the configuration of a dialogue support system 1 according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating an outline of processing performed by the dialogue support system 1 according to the embodiment. [Figure 3] FIG. 10 is a diagram showing an example of a processing procedure of a search data processing device 400 in an embodiment. [Figure 4] FIG. 10 is a diagram showing another example of the processing procedure of the search data processing device 400 in the embodiment. [Figure 5] FIG. 10 is a diagram showing an example of processing of the LLM for subject extraction and classification in the embodiment. [Figure 6] 4 is a flowchart showing an example of a processing procedure of the dialogue support system 1 according to the embodiment. [Figure 7] 10 is a diagram illustrating an example of customer-defined information including a variable name and a character string according to an embodiment. FIG. [Figure 8] FIG. 10 is a diagram illustrating an example of a persona generation prompt according to an embodiment. [Figure 9] FIG. 10 is a diagram illustrating an example of variables created from a persona generation prompt in the embodiment. [Figure 10] FIG. 10 is a diagram illustrating an example of a persona designation prompt according to an embodiment. [Figure 11] FIG. 10 is a diagram showing an example of persona designation information in the embodiment. [Figure 12] FIG. 10 is a diagram illustrating an example of an evaluation prompt according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] A search data processing system, a search data processing method, and a program to which the present invention is applied will be described below with reference to the drawings.

[0012] FIG. 1 is a block diagram showing an example of the configuration of a dialogue support system 1 according to an embodiment. A dialogue support system 1 according to an embodiment supports a dialogue between a user and an avatar characterized by a specific persona. For example, the dialogue support system 1 generates a persona based on information specified by the user, and controls an avatar corresponding to the generated persona, thereby allowing the user and the avatar to engage in role-playing. Role-playing is, for example, training new employees, sales training, communication training, etc., with the user and the pseudo-avatar as conversation partners.

[0013] The dialogue support system 1 includes, for example, a processing server device 100, a generation server device 200, a user terminal device 300, and a search data processing device 400. The processing server device 100, the generation server device 200, the user terminal device 300, and the search data processing device 400 are communicatively connected via a network NW such as the Internet. The processing server device 100, the generation server device 200, the user terminal device 300, and the search data processing device 400 may be connected via either wired or wireless communication, including a general-purpose network such as the Internet, or a private network such as local 5G or Wi-Fi (registered trademark). The processing server device 100, the generation server device 200, the user terminal device 300, and the search data processing device 400 may have a communication interface, such as a network interface card (NIC) or a wireless communication module, for connecting to a network, allowing them to exchange information with each other.

[0014] The user terminal device 300 is, for example, an information processing device operated by a user who interacts with an avatar. The user terminal device 300 includes, for example, a speaker, a microphone, a display device, an operation unit, and a processing unit such as a CPU.

[0015] The processing server device 100 is, for example, a server device that includes a processor that performs processing in response to requests received from the generation server device 200 and the user terminal device 300 and transmits processing results to the generation server device 200 and the user terminal device 300. The processing server device 100 includes, for example, a customer generation unit 110, a dialogue control unit 120, an operation control unit 130, and a storage unit 140. The customer generation unit 110, the dialogue control unit 120, and the operation control unit 130 are functional units realized by, for example, an information processing circuit that performs various processes by causing a central processing unit (CPU) to execute a program. Furthermore, some or all of these functional units may be realized by hardware such as a large-scale integration (LSI), an application-specific integrated circuit (ASIC), or a field-programmable gate array (FPGA), or may be realized by a combination of software and hardware. The storage unit 140 is realized by, for example, a hard disk drive (HDD), a solid state drive (SSD), a flash memory, an electrically erasable programmable read-only memory (EEPROM), a read-only memory (ROM), or a random access memory (RAM), or a hybrid storage device using a combination of these. Part or all of the storage unit 140 may be realized by an external storage device accessible via various networks. An example of an external storage device is a network attached storage (NAS) device.

[0016] The customer generation unit 110 generates customer information. The customer information is information indicating a customer envisioned by the user. A customer corresponds to an avatar who will be the user's conversation partner in, for example, role-playing. The customer generation unit 110 includes, for example, a reception unit 111 and a customer definition unit 112. The reception unit 111 receives persona information based on information received from the user terminal device 300. The persona information is customer definition information indicating a persona that characterizes a customer (conversation partner) for the user. A persona is a virtual character image. The customer definition unit 112 generates customer information based on the persona information received by the reception unit 111.

[0017] The dialogue control unit 120 controls an avatar corresponding to a persona based on the persona generation prompt generated by the persona generation unit 211, and performs processing to control a dialogue between the avatar and a user. The dialogue control unit 120 includes, for example, an utterance acquisition unit 121, an emotion parameter processing unit 122, a response prompt generation unit 123, a response text conversion unit 124, and a conversation history generation unit 125. The utterance acquisition unit 121 acquires utterance information indicating the user's utterance input from the user terminal device 300, and converts the acquired utterance information into text data. Emotion parameter processing unit 122 performs processing to set and update emotion parameters. Emotion parameters are numerical values ​​that indicate the emotion of an avatar (customer). Emotion parameters are information that expresses emotions such as joy, anger, sadness, enjoyment, confidence, confusion, and fear on a scale of 1 to 5. Note that, although the present embodiment describes a configuration related to the emotion of the customer, such as emotion parameter processing unit 122, the present invention is not limited to this, and the configuration related to the emotion of the customer may not be provided. The response prompt generation unit 123 generates a response prompt including text data of the user's voice and emotion parameters, and transmits the generated response prompt to the generation server device 200. The response text conversion unit 124 converts the response text acquired from the generation server device 200 into voice data. The conversation history generating unit 125 generates history information indicating the history of conversations between the user and the avatar.

[0018] The movement control unit 130 performs processing to control the movement of the avatar. The movement control unit 130 includes, for example, an avatar generation unit 131, a voice generation unit 132, a voice information processing unit 133, a motion processing unit 134, an emote processing unit 135, and a lip sync processing unit 136. The avatar generation unit 131 generates an avatar. The avatar generation unit 131 creates component information representing content for displaying an avatar based on, for example, an image showing the appearance of a customer. The voice generating unit 132 generates voice data to be output to the user. For example, the voice generating unit 132 generates voice data that reproduces the actual voice of the customer reading out loud. The voice timbre information processing unit 133 processes the voice data based on the voice timbre information corresponding to the emotion parameters. The motion processing unit 134 controls the motion of the avatar based on the emotion parameters and the content of the response text. The motion of the avatar represents, for example, the movement of the entire avatar or the movement of the avatar's hands. The emotion processor 135 controls the facial expression of the avatar based on the emotion parameters and the content of the response text, for example, by controlling the movements of the avatar's eyes, eyebrows, mouth, and the like. Lip sync processor 136 controls the movement of the avatar's lips based on the emotion parameters and the content of the response text.

[0019] The storage unit 140 stores, for example, customer information 141, response information 142, voice information 143, and action information 144. The customer information 141 includes, for example, persona information, utterance information, a persona generation prompt, and a persona designation prompt. The persona generation prompt is detailed information for generating a persona. The persona designation prompt is information indicating a persona that is designated when a user and an avatar actually have a dialogue, such as a role-play. The response information 142 includes, for example, user voice text and response text, but may also include initial emotional parameter values ​​and current emotional parameter values. The voice information 143 includes, for example, voice data such as user voice and response voice, and tone of voice information, but may also include emotional parameters. The action information 144 includes, for example, emotional parameters, component information, motion information, emote information, and lip-sync information. The motion information is a default value representing the avatar's motion, the emote information is a default value representing the avatar's emote, and the lip-sync information is a default value representing the avatar's lip-sync.

[0020] The generation server device 200 is, for example, a server device that performs processing in response to a request received from the processing server device 100 and transmits the processing results. The generation server device 200 includes, for example, a generation unit 210, a storage unit 220, and an LLM learning unit 230. The generation unit 210 and the LLM learning unit 230 are functional units realized by, for example, an information processing circuit that performs various processes by causing a CPU to execute a program. The storage unit 220 is realized, for example, by a recording device such as an HDD or SSD, or a hybrid storage device using a plurality of these, and may also be realized by an external storage device accessible via various networks, such as a NAS device.

[0021] The generation unit 210 includes, for example, a persona generation unit 211, a response text generation unit 212, and a unique information acquisition unit 214. The persona generation unit 211 inputs persona information acquired from the processing server device 100 into a first large-scale language model and generates a persona generation prompt based on the output of the first large-scale language model. The persona generation unit 211 may input persona information and information about a specific field into the first large-scale language model and create a persona generation prompt that indicates the characteristics of a persona corresponding to the specific field based on the output of the first large-scale language model. The information about the specific field is various information about a field that becomes a topic of conversation. The information about the specific field may be, for example, customer characteristic information such as customer issues related to product purchases that are empirically assumed for a specific industry or a specific generation. The information about the specific field is acquired as unique information by the unique information acquisition unit 214. The response text generation unit 212 generates a response text from the response prompt generated by the response prompt generation unit 123, the conversation history generated by the conversation history generation unit 125, and the unique information acquired by the unique information acquisition unit 214. The response text generation unit 212, for example, inputs the response prompt, the conversation history between the user and the avatar, and the unique information into a second large-scale language model, and generates a response text based on the second large-scale language model. Note that the response text generation unit 212 may extract context information of the conversation and generate a response text based on the context information in addition to the response prompt, the conversation history, and the unique information. The first large-scale language model is, for example, a large-scale language model (LLM) using a neural network. The second large-scale language model may be the same LLM as the first large-scale language model, or they may be different LLMs. Emotion parameter generation section 213 generates or updates emotion parameters according to the content of the generated response text. The unique information acquisition unit 214 acquires unique information that is information unique to a dialogue such as a role-play. Unique information is, for example, information about a specific person, information about a specific organization, or information about a specific product. For example, information about a specific person is information that is registered by classifying multiple types of original information with different data formats.

[0022] The storage unit 220 includes, for example, unique information 221 acquired by the unique information acquisition unit 214 and LLM information 222. The LLM information 222 is parameter information of an LLM (first large-scale language model) for generating a persona generation prompt. The LLM for generating a persona generation prompt is a machine learning model trained to output a persona generation prompt indicating a detailed prediction result of a persona when information for defining a persona input based on a user operation is input, for example. The LLM information 222 may include parameter information for an LLM that generates a persona-specific prompt based on the persona-generating prompt. The LLM for generating the persona-specific prompt is, for example, a machine learning model trained to output a persona-specific prompt that indicates a customer in a role-play when the persona-generating prompt is input. The LLM information 222 may include parameter information of a second large-scale language model (LLM) for generating a response text based on a persona-specified prompt. The LLM for generating a response text is a machine learning model trained to output a response text for an avatar to converse with when, for example, a persona-specified prompt and user utterance information are input. The LLM information 222 may include parameter information for an LLM that generates an evaluation prompt based on a conversation history. The LLM that generates the evaluation prompt is a machine learning model that is trained to output an evaluation prompt including information that predicts the user's evaluation when the conversation history is input. In addition, the LLM for generating persona generation prompts, the LLM for generating persona specification prompts, the LLM for generating response text, and the LLM for generating evaluation prompts may be a single LLM or may be different LLMs.

[0023] The LLM training unit 230 performs a process of training an LLM (first large-scale language model) for generating persona-generating prompts and an LLM (second language model) for generating response texts. The LLM training unit 230 may also train an LLM for generating persona-specifying prompts and an LLM for generating evaluation prompts.

[0024] The search data processing device 400 is an information processing device that communicates with other devices via a communication network NW and performs various processes. The search data processing device 400 functions as a search data processing system that registers search data. The search data is data used to search for specific information, such as information about a specific person, information about a specific organization, or information about a specific product. The process of searching for specific information is, for example, a hybrid search of keyword search and vector search, but is not limited to this.

[0025] The search data processing device 400 includes, for example, an information acquisition unit 410, a storage unit 420, an extraction unit 430, a classification unit 440, and a registration unit 450. The information acquisition unit 410, the storage unit 420, the extraction unit 430, the classification unit 440, and the registration unit 450 correspond to a search data processing system that registers search data. The information acquisition unit 410, the extraction unit 430, the classification unit 440, and the registration unit 450 are functional units realized, for example, by an information processing circuit that performs various processes by causing a CPU to execute a program. The storage unit 420 is realized, for example, by a recording device such as an HDD or SSD, or a hybrid storage device using multiple devices, and may also be realized by an external storage device accessible via various networks, such as a NAS device.

[0026] The information acquisition unit 410 acquires various types of information via the communication network NW. For example, when acquiring information about a specific person, the information acquisition unit 410 searches for information about the specific person via the communication network NW and acquires multiple types of original information with different data formats. The original information is information that serves as the source of information registered by processing by the extraction unit 430, the classification unit 440, and the registration unit 450. The multiple types of original information with different data formats include information that is saved in different data formats, such as book data, web data, audio data, photo data, and video data.

[0027] The original information may be data sent by a specific person or organization and distributed across multiple information sources, such as a server device that provides web content, a server device that provides SNS services, and a server device that provides video distribution services, all of which are managed by different information processing devices. The storage unit 420 stores a plurality of types of raw information with different data formats. The storage unit 420 also stores information registered by the registration unit 450.

[0028] The extraction unit 430 extracts subject information indicating the subject of the original information from the original information stored in the storage unit 420. The extraction unit 430 may standardize the original information stored in the storage unit 420 into a predetermined format, input the format to a large-scale language model, and extract the subject based on the output of the large-scale language model.

[0029] The extraction unit 430 may input raw information in multiple data formats stored in the storage unit 420 into a multimodal machine learning model and extract a main theme based on the output of the multimodal machine learning model. A multimodal machine learning model is an integrated machine learning model that can process multiple types of data, such as text, images, audio, and numerical values, at the same time. An example of a multimodal machine learning model is a neural network model.

[0030] The classification unit 440 classifies the original information stored in the storage unit 420 based on the subject information extracted by the extraction unit 430. The classification unit 440 may classify the subject of the original information using the Nippon Decimal Classification or the Colon Classification by repeatedly determining to which of a plurality of classifications the subject of the data belongs. The classification format may be, for example, a multiple-branching or hierarchical structure.

[0031] The registration unit 450 registers the raw information classified by the classification unit 440 as search data for each item. The registration unit 450 registers the search data by storing it in the storage unit 420. The registration unit 450 may store the search data in the storage unit 220 via the communication network NW.

[0032] In the embodiment, as shown in Figure 1, the dialogue support system 1 distributes the functional configuration (functional units) among the processing server device 100, the generation server device 200, and the search data processing device 400, but this is not limited to this and the functional units may be distributed in other configurations, the functional units of the processing server device 100, the generation server device 200, and the search data processing device 400 may be consolidated into one device, multiple functional units may be combined into one functional unit, or one function may be distributed among multiple functional units.

[0033] FIG. 2 is a diagram illustrating an outline of the processing performed by the dialogue support system 1 according to the embodiment. The reception unit 111 and the customer definition unit 112 generate customer definition information D10 and send it to the generation server device 200. The persona generation unit 211 inputs the customer definition information D10 to a persona generation LLM (P10) and generates a persona generation prompt D12 based on the output of the persona generation LLM (P10). The persona generation unit 211 generates a persona specification prompt D14 when the user performs a role-play. The persona specification prompt D14 is output to the response text generation LLM.

[0034] The search data processing device 400 stores search data in the storage unit 220 using the registration unit 450. The search data stored in the storage unit 220 is acquired by the generation server device 200 as, for example, unique information D20 for performing role-playing. For example, when acquiring unique information of a specific field and a specific organization in role-playing, the unique information acquisition unit 214 searches for search data whose subject is the specific field and search data whose subject is the specific organization as unique information. The generation server device 200 performs a text extraction process P20, a chunk division process P21, and a vectorization process P22 on the unique information D20 in that order, and stores vectors corresponding to the unique information D20 in a vector database (storage unit 220). The processing server device 100 performs speech recognition processing P40 on the user's speech acquired from the user terminal device 300, and then performs vectorization processing P41 on the text information processed by the speech recognition processing P40. A reference result to the vector database is extracted from the vector database using the vector corresponding to the speech as a query. The vectorized unique information D20 and the text information processed by the speech recognition processing P40 are output to the response text generation LLM (P30) together with a persona specification prompt D14.

[0035] The generation server device 200 inputs the persona specification prompt D14, unique information D20, and text information processed by speech recognition processing P40 into the response text generation LLM (P30), performs speech synthesis processing P31 on the response text output from the response text generation LLM (P30), and performs avatar control processing P32 based on the emotion parameters output from the response text generation LLM (P30), thereby transmitting avatar content D30 to the user terminal device 300. This allows the user terminal device 300 to display or output sound using avatar content D30.

[0036] FIG. 3 is a diagram showing an example of a processing procedure of the search data processing device 400 according to the embodiment. For example, the information acquisition unit 410 acquires book data A from information source A and video data B from information source B, and the storage unit 420 stores the book data A and video data B. The author of the book data A and the performer of the video data A are assumed to be a specific person A. For example, the book data A contains a statement such as "I've loved tuna mayonnaise as a rice ball filling since I was a child." The video data A contains audio or video such as "The green juice I developed is good for your health."

[0037] The search data processing device 400 inputs book data A into the subject extraction and classification LLM. The search data processing device 400 extracts the subject of book data A based on the output of the subject extraction and classification LLM. The extraction unit 430 extracts the subject "I've loved tuna mayo as a rice ball filling since I was a child" from book data A based on the output from the subject extraction and classification LLM. The classification unit 440 repeatedly determines whether the subject of book data A belongs to one of multiple classifications using the Nippon Decimal Classification, such as classification items "5 Technology," "59 Home Economics and Life Sciences," and "596 Food and Cooking," based on the output of the subject extraction and classification LLM. As a result, the extraction unit 430 and classification unit 440 can classify the subject "I've loved tuna mayo as a rice ball filling since I was a child" for book data A into the classification item "596 Food and Cooking."

[0038] The search data processing device 400 inputs data converted into text information for the video data A into the subject extraction and classification LLM. The search data processing device 400 extracts the subject of the video data A based on the output of the subject extraction and classification LLM.

[0039] The extraction unit 430 extracts the subject "The green juice I developed is good for your health" from book data A based on the output from the subject extraction and classification LLM. The classification unit 440 repeatedly determines whether the subject of video data A belongs to one of multiple classifications using the Nippon Decimal Classification, such as classification items "5 Technology," "59 Home Economics and Life Science," and "596 Food and Cooking," based on the output from the subject extraction and classification LLM. The classification unit 440 also repeatedly determines whether the subject of video data A belongs to one of multiple classifications using the Nippon Decimal Classification, such as classification items "5 Technology," "59 Home Economics and Life Science," and "597 Housing and Furniture," based on the output from the subject extraction and classification LLM. As a result, the extraction unit 430 and the classification unit 440 can classify the theme "The green juice I developed is good for your health" into the classification items "596 Food and Cooking" and "597 Housing and Furniture" for the video data A. The registration unit 450 registers the theme in association with the item classified by the classification unit 440.

[0040] FIG. 4 is a diagram showing another example of the processing procedure of the search data processing device 400 according to the embodiment. The search data processing device 400 may include a summarizing unit 432 that summarizes the themes of the original information extracted by the extracting unit 430. The extracting unit 430 summarizes for each theme classified by the classifying unit 440. The registering unit 450 registers the summary sentences resulting from the theme summarization in association with the classified items. This enables the search data processing device 400 to compress the amount of information on the theme and eliminate overlapping themes.

[0041] FIG. 5 is a diagram showing an example of processing of the LLM for subject extraction and classification in the embodiment. The subject extraction and classification LLM inputs, for example, classification criteria definition information, unique information, and multiple types of source information with different data formats, and outputs data indicating classification items and subjects. The classification criteria definition information is, for example, information defining the Nippon Decimal Classification or the Colon Classification, but is not limited to these. Any classification criteria capable of classifying subjects may be used. The classification criteria definition information may, for example, be information defining a company's document management rules, a method for classifying documents based on a company's unique business operations, workflow, or relevance, rather than a general classification system, or information defining a method for classifying books, documents, and documents into unique categories based on personal preferences. The classification criteria definition information may, for example, include information indicating numerous classification items and the hierarchy of the classification items.

[0042] FIG. 6 is a flowchart showing an example of a processing procedure of the dialogue support system 1 according to the embodiment. First, the processing server device 100 inputs user information about the user who will interact with the avatar (step S100). The user information is, for example, character string data that characterizes the user, such as a new employee or a sales manager. Next, the processing server device 100 defines the customer the user is considering (step S102). At this time, the processing server device 100 receives persona information indicating a persona that characterizes the user's interaction partner from the user terminal device 300 via the reception unit 111, and stores the information as a variable, for example, as shown in FIG. 7. FIG. 7 is a diagram showing an example of customer-defined information including variable names and character strings in an embodiment. The reception unit 111 determines whether there is further input (step S104). If there is further input, the processing unit 111 repeats the processing of step S102. If there is no further input, the processing server device 100 confirms the customer-defined information. The processing server device 100 transmits the customer-defined information to the generation server device 200.

[0043] The persona generation unit 211 inputs the customer-defined information and the unique information stored in the storage unit 220 into the persona generation LLM, and generates a persona generation prompt based on the output of the persona generation LLM (step S106). FIG. 8 is a diagram showing an example of a persona generation prompt in an embodiment. The persona generation prompt includes text data indicating, for example, the customer's prerequisites, the customer's speaking style, and the customer's personality traits. The customer's prerequisites include, for example, items corresponding to role-playing, such as age, gender, occupation, family structure, residential area, and insurance information. The customer's speaking style includes, for example, first person, second person, and catchphrases. The customer's personality traits include, for example, neuroticism, extroversion, openness, conscientiousness, and cooperativeness. The persona generation unit 211 determines whether or not there is input for other items, etc. from the user terminal device 300 (step S108). If input is received, the process of step S106 is repeated; if no input is received, the persona generation prompt is confirmed. The persona generation unit 211 stores the created persona generation prompt as a variable in the storage unit 220. Fig. 9 is a diagram showing an example of a variable created from a persona generation prompt in an embodiment. The variable created from the persona generation prompt is information predicted based on the output of the persona generation LLM after inputting customer-defined information into the persona generation LLM.

[0044] The persona generation unit 211 may input existing items including the persona's gender, age, speaking style, and personality as persona information (customer-defined information) into a persona generation LLM (first large-scale language model), and further input at least one of items specified based on user operation, items related to the characteristics of customers in a specific industry, and items related to the characteristics of customers in a specific generation into the persona generation LLM, and create a persona generation prompt based on the output of the persona generation LLM. In this way, the persona generation unit 211 realizes Retrieval-Augmented Generation (RAG) by inputting various information into the persona generation LLM, and can create highly accurate persona generation prompts that meet the user's requests.

[0045] The persona generation unit 211 may input items related to issues customers have with respect to a specific product into a persona generation LLM (first large-scale language model) and create a persona generation prompt based on the output of the persona generation LLM. Items related to issues customers have with respect to a specific product are information empirically assumed in a specific field or industry, or information such as customer issues related to the purchase of a product assumed based on market analysis and survey results. Items related to issues customers have with respect to a specific product may be added or modified as variables by the user.

[0046] Next, the persona generation unit 211 creates a persona specification prompt (step S110). FIG. 10 is a diagram showing an example of a persona specification prompt in an embodiment. The persona generation unit 211 may input variables corresponding to the persona generation prompt into the LLM and create the persona specification prompt based on the output of the LLM. The variables corresponding to the persona generation prompt may be variables selected based on a user operation, or may be variables extracted from the persona generation prompt randomly or according to a predetermined rule. The persona generation unit 211 transmits persona specification information such as that shown in FIG. 11 to the processing server device 100 based on the persona specification prompt. FIG. 11 is a diagram showing an example of persona specification information in an embodiment.

[0047] Next, the dialogue control unit 120 controls an avatar corresponding to the persona based on the persona designation information, and controls the dialogue between the avatar and the user (step S112). As a result, the dialogue control unit 120 performs role-playing through dialogue between the user and the avatar. The utterance acquisition unit 121 acquires utterance information indicating the user's utterances, and the conversation history generation unit 125 stores the conversation history (step S114).

[0048] Next, the processing server device 100 determines whether to evaluate the user (step S116). If the processing server device 100 does not evaluate the user, the processing server device 100 repeats the processes of steps S112 and S114. For example, when the processing server device 100 detects a user utterance such as "evaluate the role-play," the processing server device 100 determines to evaluate the user and transmits the conversation history to the generation server device 200. The generation server device 200 evaluates the user based on the utterance information acquired by the utterance acquisition unit 121 (step S118). At this time, the generation server device 200 generates an evaluation prompt based on the conversation history acquired from the processing server device 100. The generation server device 200 may input the conversation history to an LLM that generates an evaluation prompt and generate the evaluation prompt based on the output of the LLM. For example, when a user explains a product in a role-play, the generation server device 200 may input the conversation history and information stored in a product information database to an LLM that generates an evaluation prompt and generate the evaluation prompt based on the output of the LLM that generates the evaluation prompt. FIG. 12 is a diagram showing an example of an evaluation prompt in an embodiment. The evaluation prompt includes, for example, evaluation items, evaluation criteria, evaluation points, customer understanding evaluation, product knowledge evaluation, and communication evaluation for conversations for sales activities.

[0049] The generation server device 200 outputs evaluation information to the processing server device 100 as a result of evaluating the user's conversation based on the evaluation prompt (step S120). As a result, the processing server device 100 transmits the evaluation information to the user terminal device 300, and the user terminal device 300 can present the evaluation to the user.

[0050] As described above, the search data processing system according to the embodiment stores multiple types of source information with different data formats, extracts topic information indicating the subject of the source information from the stored source information, classifies the stored source information based on the extracted topic information, and registers the classified source information as search data by category. This search data processing system extracts and classifies topics from the source information of the search data, enabling information to be classified by topic even when duplicate information is distributed across multiple sources, thereby enabling information retrieval in a short time. For example, text information from multiple sources with different data formats, such as books and videos, can be extracted and classified and registered by category. Furthermore, even when statements or writings by a specific person or organization are found across multiple documents, effectively constructing data to be used in RAG significantly speeds up information retrieval. Furthermore, even when polysemous terms are used in various sources, classifying each data by topic allows accurate output of the location of the subject in each data. Furthermore, the search data processing system can be used to create manuals for specific products, for example, by classifying data from multiple sources by topic regarding the use of a specific product.

[0051] The functions of the processing server device 100, the generation server device 200, and the user terminal device 300 in the above-described embodiments may be implemented by a computer. In this case, a program for implementing the functions may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed. Note that the term "computer system" as used herein includes hardware such as an operating system and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into a computer system. Furthermore, the term "computer-readable recording medium" may also include devices that dynamically store programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or telephone lines, or devices that store programs for a fixed period of time, such as volatile memory within the computer system serving as the server or client. The program may be designed to implement only a portion of the above-described functions, or may be capable of implementing the above-described functions in combination with a program already stored in the computer system, or may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0052] Although each embodiment and variant example has been described, these are merely examples and are not intended to limit the scope of the present invention. For example, one of the embodiments or variant examples, or a part of each embodiment or a part of each variant example, may be combined with one or more other embodiments or one or more other variant examples to realize one aspect of the present invention. [Explanation of symbols]

[0053] 1...dialogue support system, 100...processing server device, 110...customer generation unit, 111...reception unit, 112...customer definition unit, 120...dialogue control unit, 121...utterance acquisition unit, 122...emotion parameter processing unit, 123...response prompt generation unit, 124...response text conversion unit, 125...conversation history generation unit, 130...motion control unit, 131...avatar generation unit, 132...voice generation unit, 133...voice tone information processing unit, 134...motion processing unit, 135...emotion processing unit, 136...lip sync processing unit, 140 ...storage unit, 141...customer information, 142...response information, 143...voice information, 144...action information, 200...generation server device, 210...generation unit, 211...persona generation unit, 212...response text generation unit, 214...unique information acquisition unit, 220...storage unit, 221...unique information, 222...LLM information, 230...LLM learning unit, 300...user terminal device, 400...search data processing device, 410...information acquisition unit, 420...storage unit, 430...extraction unit, 432...summarization unit, 440...classification unit, 450...registration unit

Claims

1. A search data processing system for registering search data, a storage unit for storing a plurality of types of original information having different data formats; an extraction unit that extracts subject information indicating a subject of the original information from the original information stored in the storage unit; a classification unit that classifies the original information stored in the storage unit based on the subject information extracted by the extraction unit; a registration unit that registers the raw information classified by the classification unit as the search data for each item; A search data processing system comprising:

2. 2. The search data processing system according to claim 1, wherein the original information is data sent by a specific person or organization and distributed among a plurality of information sources.

3. 2. The search data processing system according to claim 1, wherein the extraction unit standardizes the original information stored in the storage unit into a predetermined format and inputs it into a large-scale language model, and extracts the main topic based on the output of the large-scale language model.

4. 2. The search data processing system according to claim 1, wherein the extraction unit inputs raw information having a plurality of types of data formats stored in the storage unit into a multimodal machine learning model and extracts the main theme based on the output of the multimodal machine learning model.

5. 2. The search data processing system according to claim 1, wherein the classification unit classifies the subject of the original information by repeating a process of determining whether the subject of the original information belongs to one of a plurality of classifications using the Nippon Decimal Classification or the Colon Classification.

6. a summarizing unit that summarizes the subject of the original information extracted by the extracting unit, the classification unit classifies the data based on subject information indicating a subject of the data summarized by the summarization unit; 2. The search data processing system according to claim 1.

7. a step in which a search data processing system for registering search data stores a plurality of types of original information with different data formats in a storage unit; a step in which the search data processing system extracts subject information indicating a subject of the original information from the original information stored in the storage unit; a step in which the search data processing system classifies the original information stored in the storage unit based on the extracted subject information; a step in which the search data processing system registers the classified raw information as the search data for each item; A method for processing data for search, comprising:

8. A computer of a search data processing system that registers search data, storing a plurality of types of raw information having different data formats in a storage unit; extracting subject information indicating a subject of the original information from the original information stored in the storage unit; classifying the original information stored in the storage unit based on the extracted subject information; a step of registering the classified raw information as the search data for each item; A program that executes.

Images