Method and system for providing consultation service using chatbot based on large language model

Abstract

The present invention relates to a method and system for providing a consultation service using a chatbot based on a large language model. The method for providing a consultation service using a chatbot based on a large language model according to the present invention comprises the steps of: collecting user-related information related to a user account logged in to a user terminal; specifying similar users having characteristics similar to the characteristics of the user among different users on the basis of the user-related information; generating a user group including the user and the similar users; generating a group data set on the basis of user-related information of a plurality of users included in the user group; constructing a group chatbot model corresponding to the user group using the group data set on the basis of a pre-trained large language model; and generating an answer corresponding to a user query received from the user terminal using the group chatbot model and providing the generated answer to the user terminal.

Description

Method and System for Providing Consultation Services Using Large Language Model-Based Chatbots

[0001] The present invention relates to a method and system for providing consultation services to a user using a chatbot based on a large language model.

[0002] Recently, with the rapid advancement of artificial intelligence (AI) technology, generative language models (e.g., ChatGPT) capable of interacting naturally with humans are gaining attention. In particular, chatbots based on Large Language Models (LLMs) are being widely utilized in areas such as information provision, consultation, and recommendations, based on advanced language processing technology that understands context and freely generates responses, unlike existing chatbots that only performed simple responses within limited scenarios.

[0003] Such technological advancements are presenting new possibilities in the field of rehabilitation treatment and management. During the rehabilitation process, patients experience various physical and psychological difficulties, such as pain, limited range of motion, and anxiety regarding repetitive movements; therefore, providing accurate information and emotional support tailored to these situations serves as crucial elements of the rehabilitation process. However, receiving real-time consultation from medical professionals whenever patients need it is often difficult due to realistic limitations, such as time constraints, a shortage of medical personnel, and financial burdens.

[0004] To address these issues, there has recently been a surge in demand for non-face-to-face consultation using large language model-based chatbots. Large language model-based chatbots go beyond simply conveying information to partially replace the role of a counselor and are also receiving positive evaluations for their ability to empathize with emotions.

[0005] However, while large language models are trained on vast amounts of general-purpose data and possess excellent generality capable of responding to various topics and situations, they have limitations in providing personalized information that reflects the context of individual users or specific user groups. In other words, by generating generalized responses without considering the user's state, background, or preferences, there is a possibility that information unsuitable for individual needs or unnecessary information may be provided.

[0006] To overcome these challenges, there is a growing demand for technologies that improve the precision of user-customized responses while maintaining the generality of large language models. In particular, active research is being conducted on techniques that data-drivenly cluster user groups with similar characteristics—such as health status, behavioral history, medical history, and interests—and fine-tune large language models based on the groups' conversation history or question-and-answer data.

[0007] As such, there is a need to provide customized consultation services to users by building a chatbot model with response tendencies specialized for specific groups based on a large language model.

[0008] The present invention is intended to provide a method and system for providing user-customized consultation services using a large language model-based chatbot.

[0009] Specifically, the present invention aims to provide a method and system for providing a consultation service using a large language model-based chatbot capable of clustering users similar to a user to create a user group, building a chatbot corresponding to the user group, and providing a customized answer to a user query.

[0010] Furthermore, the present invention aims to provide a method and system for providing a consultation service using a large language model-based chatbot, which allows a user to continue a conversation with a user in the same group and the chatbot by utilizing question-and-answer information performed between the user and the chatbot by users in a user group that includes the user.

[0011] Furthermore, the present invention aims to provide a method and system for providing a consultation service using a large language model-based chatbot that can perform conversations with other users based on a group chat room, and can provide customized answers to user queries by reflecting the content of the conversation performed in the group chat room in a prompt.

[0012] Furthermore, the present invention aims to provide a method and system for providing a consultation service using a large language model-based chatbot that can update an exercise program provided to a user account according to a user query through the approval of medical staff.

[0013] To solve the problem described above, the present invention proposes a method for interacting with a user remotely and providing customized consultation services to the user by using a large language model-based chatbot specialized for a user group that includes users similar to the user. The method for providing consultation services using a large language model-based chatbot according to the present invention may include the steps of: collecting user-related information related to a user account logged into a user terminal; identifying similar users among different users who have characteristics similar to the user based on the user-related information; creating a user group including the user and the similar users; creating a group data set based on user-related information of a plurality of users included in the user group; constructing a group chatbot model corresponding to the user group using the group data set based on a previously trained large language model; and generating an answer corresponding to a user query received from the user terminal using the group chatbot model and providing the generated answer to the user terminal.

[0014] Furthermore, the step of building the group chatbot model may include the step of fine-tuning the previously trained large language model using the group data set corresponding to the user group, and the step of building the group chatbot model that generates an answer to the user query based on the group data set using the fine-tuned large language model.

[0015] Furthermore, the step of providing the above answer to the user terminal may include: generating a prompt to generate the above answer to the user query based on at least one of the user-related information and the group data set; processing the above prompt as input to the finely tuned large language model to obtain the above answer to the user query through the large language model; and providing the above answer to the group chatbot model to provide the above answer to the user query to the user terminal through the group chatbot model.

[0016] Furthermore, the step of generating the prompt may include the step of extracting keywords related to the user query from the group data set, the step of identifying specific user-related information related to the keyword among a plurality of user-related information included in the group data set, and the step of generating the prompt to generate an answer to the user query using the specific user-related information.

[0017] Furthermore, the step of identifying the similar users may include the step of searching for similar user-related information in a database based on similarity with the user-related information, and the step of identifying a specific user corresponding to the similar user-related information as a similar user based on the similarity.

[0018] Furthermore, the step of searching for similar user-related information may include the step of embedding the user-related information to obtain a target user vector corresponding to the user-related information, the step of calculating the similarity between the target user vector and the registered user vector corresponding to each of the different users stored in the database, and the step of specifying the registered user-related information corresponding to at least one registered user vector that satisfies a pre-set similarity condition as the similar user-related information.

[0019] Furthermore, the step of providing the above answer to the user terminal may further include the step of extracting the above question-and-answer list, which includes a plurality of question-and-answer items related to the user query from the group data set, and the step of providing the user terminal with a consultation page capable of performing subsequent consultation related to at least one question-and-answer item included in the above question-and-answer list.

[0020] Furthermore, the step of providing the consultation page to a user terminal may include the step of specifying one of the plurality of question-and-answer items constituting the question-and-answer list according to a first user input entered into the user terminal, and the step of providing the consultation page to the user terminal, which allows the user to input a user query, so as to perform the subsequent consultation related to the specified question-and-answer item.

[0021] Furthermore, regarding the above user query, the method further includes the step of creating a group chat room capable of question-and-answer exchange between the user included in the user group and the plurality of users, and providing this to the user terminal; and in the step of generating the prompt, the method may generate a prompt that generates the answer to the user query by utilizing question-and-answer information including conversation content performed in the group chat room.

[0022] Furthermore, the step of providing the group chat room to the user terminal may include: outputting a list of group members for each of a plurality of user groups to the user terminal; generating an invitation icon corresponding to each of the group members included in the list of group members in a part area of ​​the list of group members; transmitting an invitation request to a guest user terminal where the guest user's user account is logged in, based on receiving a second user input for a specific invitation icon from the user terminal, so that the guest user corresponding to the specific invitation icon among the group members can perform a question and answer in the group chat room; and providing the group chat room that the guest user can participate in to the user terminal based on the occurrence of an approval event for the invitation request.

[0023] Furthermore, the user-related information may include at least one of medical information, exercise history information, question and answer information, and exercise program information related to the user.

[0024] Furthermore, the method may further include a step of updating an exercise program set in a user account according to a user query received from the user terminal, wherein the step of updating the exercise program may include: generating an update prompt to update the exercise program assigned to the user account using the user-related information; inputting the update prompt into a large language model to generate an updated exercise program according to the user query through the large language model; transmitting an approval request for the updated exercise program to a medical staff terminal; and setting the updated exercise program in the user account based on the occurrence of an approval event corresponding to the approval request from the medical staff terminal.

[0025] Meanwhile, a consultation service provision system using a large language model-based chatbot according to the present invention includes a communication unit that receives a user query from a user terminal and a control unit that collects user-related information related to a user account logged into the user terminal. The control unit identifies similar users among different users who have characteristics similar to the user based on the user-related information, creates a user group including the user and the similar users, creates a group data set based on user-related information of a plurality of users included in the user group, constructs a group chatbot model corresponding to the user group using the group data set based on a previously trained large language model, generates an answer corresponding to the user query received from the user terminal using the group chatbot model, and provides the answer to the user terminal.

[0026] Meanwhile, the program is executed by one or more processes in an electronic device and is stored on a computer-readable recording medium, and the program may include instructions for performing the steps of: collecting user-related information related to a user account logged into a user terminal; identifying similar users among different users who have characteristics similar to said user based on said user-related information; creating a user group including said user and said similar users; creating a group data set based on user-related information of a plurality of users included in said user group; building a group chatbot model corresponding to said user group using said group data set based on a pre-trained large language model; and generating an answer corresponding to a user query received from said user terminal using said group chatbot model and providing said answer to said user terminal.

[0027] The method and system for providing consultation services using a large language model-based chatbot according to the present invention can improve response precision by providing an optimized response to user queries through the construction of a group chatbot based on user groups created by clustering users similar to the user.

[0028] Furthermore, the method and system for providing a consultation service using a large language model-based chatbot according to the present invention links a conversation with the chatbot based on existing question-and-answer information of a user group including the user, thereby allowing the user to continue the consultation by referring to or continuing the consultation flow of similar past users, and thus receive a continuous and context-aware consultation service.

[0029] Furthermore, the method and system for providing a consultation service using a large language model-based chatbot according to the present invention configures a chat room in the form of a user group and reflects the content of multi-party conversations within the group chat room in real time on a prompt, thereby enabling the chatbot to generate a high-precision response that considers not only the user's query but also the conversation context of others within the group, and the user can experience a more meaningful consultation service through mutual communication based on a sense of shared identity.

[0030] Furthermore, the method and system for providing a consultation service using a large language model-based chatbot according to the present invention can flexibly provide a customized exercise program suitable for the user's current health condition by updating the exercise program provided to the user account with the approval of medical staff based on user queries, thereby ensuring medical reliability and safety.

[0031] FIG. 1 is a conceptual diagram illustrating a consultation service provision system using a large language model-based chatbot according to the present invention.

[0032] FIG. 2 is a flowchart illustrating a method for providing a consultation service using a large language model-based chatbot according to the present invention.

[0033] FIG. 3 is a conceptual diagram illustrating user-related information according to the present invention.

[0034] Figure 4 is a conceptual diagram illustrating the process of optimizing a large language model according to a user group according to the present invention.

[0035] FIG. 5 is a conceptual diagram illustrating the process of generating a prompt to be input into a large language model according to the present invention.

[0036] FIGS. 6a and FIGS. 6b are conceptual diagrams for explaining the process of providing an exercise guide by analyzing image data according to the present invention.

[0037] FIGS. 7a and 7b are conceptual diagrams illustrating the process of providing answers to user queries through a group-based chatbot according to the present invention.

[0038] FIG. 8a is a conceptual diagram illustrating an embodiment in which a user continues a conversation with a group member of a user group, such as a user, according to the present invention.

[0039] FIG. 8b is a conceptual diagram illustrating an embodiment of performing a conversation between group members and generating a prompt based on the conversation content through a group chat room of a user group according to the present invention.

[0040] FIGS. 9a and 9b are conceptual diagrams illustrating an embodiment of updating an exercise program according to a user query according to the present invention.

[0041] FIG. 10 is a block diagram illustrating a computing system in which the present invention can be implemented.

[0042] FIGS. 11 and FIGS. 12 are block diagrams illustrating an embodiment of a computing device according to the present invention.

[0043] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components are assigned the same reference number regardless of the drawing symbols, and redundant descriptions thereof will be omitted. The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably solely for the ease of drafting the specification and do not have distinct meanings or roles in themselves. Furthermore, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art could obscure the essence of the embodiments disclosed in this specification, such detailed description will be omitted. Additionally, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification; the technical concept disclosed in this specification is not limited by the attached drawings, and it should be understood that they include all modifications, equivalents, and substitutions that fall within the spirit and technical scope of the present invention.

[0044] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but said components are not limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.

[0045] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0046] A singular expression includes a plural expression unless the context clearly indicates otherwise.

[0047] In this application, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0048] The present invention relates to a method and system for providing consultation services using a chatbot based on a large language model. More specifically, the present invention relates to a method and system for creating a user group that includes similar users with attributes similar to a user, and constructing a group chatbot model corresponding to the user group to provide customized answers to user queries. Here, a large language model (LLM) may refer to an artificial intelligence model capable of understanding and generating natural language by learning a vast amount of data.

[0049] Unlike simple rule-based chatbots and retrieval-based chatbots, a large language model-based chatbot according to the present invention may refer to a chatbot model that can understand the context of a user query and generate new sentences based on learned knowledge to provide a natural and flexible response.

[0050] In the present invention, user-related information related to a user account logged into a user terminal (10) is collected from a pre-established database, and similar users having attributes similar to the user among different users can be identified. Here, “user-related information” may include various information related to the user, such as the user’s medical information, exercise history information, question and answer information, and exercise program information. The user-related information according to the present invention may serve as a criterion for creating a user group, and a large language model-based chatbot model may be used to generate an answer that reflects the user’s current state.

[0051] A “user group” according to the present invention may refer to a group of users having similar characteristics based on specific criteria such as age group, gender, type of indication, and exercise purpose. In this case, users included in the user group may be referred to as group members.

[0052] Furthermore, according to the present invention, the “indication” refers to a symptom or clinical situation requiring specific treatment or examination, which can be understood as the user’s disease or symptoms. For the convenience of explanation, the present invention focuses on “indications related to musculoskeletal diseases,” but is not necessarily limited thereto. As an example, the counseling service described in the present invention may relate to exercise for the treatment of users requiring rehabilitation treatment for musculoskeletal diseases or users suffering from various diseases (e.g., cancer, diabetes, hypertension, etc.).

[0053] Furthermore, the present invention may provide counseling services necessary for health promotion in daily life, rather than rehabilitation exercises for therapeutic purposes related to the user's indications. For example, the exercise according to the present invention is not limited to any specific purpose and may be an exercise performed for various purposes, such as rehabilitation exercises, fitness exercises, ball sports, or dance exercises, for therapeutic, health promotion, or cosmetic purposes.

[0054] In the present invention, by using a group-based chatbot model corresponding to a specific user group among different user groups, a customized answer to a user query from a user included in the specific user group can be provided, thereby providing a consultation service related to the user's indications. According to the present invention, a user (or patient) can run an application on a user terminal (10) and receive a consultation service related to the user's indications through a group chatbot model.

[0055] Furthermore, in the present invention, a user group-based group chat room is created, and through conversation among group members in the group chat room, similar symptoms or rehabilitation experiences are shared, or by utilizing question and answer with a past group chatbot model, a counseling service related to the user's indications can be provided.

[0056] In the foregoing, a consultation service using a large language model-based chatbot according to the present invention has been generally described, and this can be implemented by a consultation service provision system described below. Below, with reference to FIG. 1, a consultation service provision system using a large language model-based chatbot according to the present invention will be described in detail. FIG. 1 is a conceptual diagram for explaining a consultation service provision system using a large language model-based chatbot according to the present invention.

[0057] As illustrated in FIG. 1, a consultation service provision system using a large language model-based chatbot according to the present invention (hereinafter referred to as the “consultation service provision system,” 100) may include at least one of a communication unit (110), a storage unit (120), and a control unit (130). At this time, the consultation service provision system (100) according to the present invention is not limited to the components described above and may further include components that perform the same or similar roles as the functions described in the specification.

[0058] Meanwhile, the counseling service providing system (100) according to the present invention may be implemented as an application or software. The counseling service providing system (100) implemented as software may be downloaded via a program (e.g., Play Store) that allows downloading applications on a user terminal (10), or implemented via an initial installation program on the user terminal (10). In this case, the communication unit (110), storage unit (120), and control unit (130) according to the present invention may be utilized as components of the user terminal (10). In the present invention, the user terminal (10) can be understood as referring to an application installed on the user terminal (10). Such an application (or software) can be understood as a component of the counseling service providing system (100) according to the present invention.

[0059] In the present invention, the user terminal (10) may also be named a 'mobile terminal' or 'electronic device', and the user terminal (10) described in this specification may include a mobile phone, a smartphone, a smart TV, a laptop computer, a digital broadcasting terminal, a PDA (personal digital assistants), a PMP (portable multimedia player), a navigation device, a slate PC, a tablet PC, an ultrabook, a wearable device (e.g., a smartwatch, a smart glass, a head-mounted display), etc.

[0060] More specifically, the user terminal (10) according to the present invention is not limited to an electronic device in which an application is activated, but may refer to an electronic device connected to an electronic device in which an application is activated. As an example, based on the fact that the user terminal (10) according to the present invention is a smartphone, the user terminal (10) may refer to a smart TV connected to said smartphone.

[0061] Meanwhile, the counseling service providing system (100) may exist inside a server (hereinafter referred to as a server) built to perform a specific purpose (e.g., providing counseling services), or it may exist as a separate device from said server. When the counseling service providing system (100) exists inside a server, the counseling service providing system (100) according to the present invention may provide a user-customized counseling service through at least one component among a communication unit (110), a storage unit (120), and a control unit (130) located inside the server, or through a module that performs a function similar to each of said components.

[0062] In this case, the application can provide a consultation service on a user terminal (10) on which the application is installed through communication with a server. Furthermore, the consultation service providing system (100) according to the present invention can provide a consultation service providing method using a large language model-based chatbot according to the present invention to the user terminal (10) by linking with a plurality of different external servers.

[0063] A user (U, or patient) of the present invention may perform consultation through a large language model-based chatbot regarding the indications of the user (U) via an application or webpage provided by the consultation service provision system (100) according to the present invention. At this time, the user (or patient, U) of the present invention may possess a user account registered with the consultation service provision system (100) according to the present invention. For convenience of explanation, the account of the user who is a patient in this specification is referred to as a 'user account (or patient account)'. The 'account' described above may be created through a page linked to the consultation service provision system (100).

[0064] Alternatively, an ‘account’ may be created on at least one other server (e.g., a medical staff server) linked to the counseling service providing system (100) according to the present invention. Accordingly, in this specification, without distinguishing the server where the account was issued, all accounts based on the counseling service providing system (100) according to the present invention are referred to as “accounts already registered in the counseling service providing system (100) according to the present invention.”

[0065] Meanwhile, medical staff may grant approval to the user (U) for prescriptions related to indication treatment (or rehabilitation treatment) and exercise program updates through a medical staff terminal. The “medical staff” described in this invention refers to a person employed at a medical institution (e.g., a hospital) and may include, for example, at least one of a doctor, a nurse, or a physical therapist. For convenience of explanation, doctors and physical therapists are used as examples of medical staff in this invention. However, medical staff are not limited thereto, and any user employed at a medical institution to provide user-customized counseling services may be considered medical staff according to this invention.

[0066] A medical professional according to the present invention may possess a medical professional account already registered in the consultation service providing system (100) according to the present invention. In this specification, a user terminal logged in with a medical professional account is referred to as a medical professional terminal. As an example, the consultation service providing system (100) according to the present invention may receive medical information including prescription information prescribed by a medical professional (D) to a user (U) by linking with a medical professional server.

[0067] Meanwhile, according to the present invention, the communication unit (110) may be connected via a wireless or wired network to a user terminal (10), a medical staff terminal, an LLM server (140), a central server, a device, and at least one network, and may be configured to receive or transmit overall data and information necessary for the operation of the consultation service provision system (100) according to the present invention. Specifically, the communication unit (110) may receive medical information related to the indications of the user (U) from the medical staff terminal so that the user (U) can perform customized consultation through a large language model-based chatbot model.

[0068] Furthermore, the communication unit (110) may receive a user query from the user terminal (10). Here, “receiving a user query” may mean receiving an input signal (or selection signal) corresponding to the user query based on the user query being input by the user through the user terminal (10). According to the present invention, the user query may include at least one of a document, text, an image (or video), and voice. In this case, the consultation service providing system (100) may further include a module that converts voice into text. For example, the consultation service providing system (100) may include a voice recognition model capable of analyzing voice data corresponding to voice received through a microphone provided in the user terminal (10). For example, a speech recognition model can convert speech into text based on at least one of the following: a STT (Speech-to-Text) algorithm, a HMM (Hidden Markov Model), a HMM-GMM (Hidden Markov Model-Gaussian Mixture Model), a CTC (Connectionist Temporal Classification) based model, a Beam Search based model, a DNN (Deep Neural Network) based model, an RNN (Recurrent Neural Network) based model, a Seq2Seq (Sequence-to-Sequence) model, and a Transformer based model.

[0069] Here, the user terminal (10) may include at least one of a mobile phone, a smartphone, a notebook computer, a laptop computer, a slate PC, a tablet PC, an ultrabook, a desktop computer, a digital broadcasting terminal, a PDA (personal digital assistants), a PMP (portable multimedia player), a navigation device, and a wearable device (e.g., a smartwatch, a smart glass, a head-mounted display).

[0070] The communication unit (110) may include at least one communication module capable of wireless communication and wired communication between the counseling service providing system (100) and the communication target. Additionally, the communication unit (110) may include a communication module that connects the counseling service providing system (100) to at least one network.

[0071] Meanwhile, the communication unit (110) can support various communication methods depending on the communication standard of the communicating device. For example, the communication unit (110) may be configured to perform communication using at least one of the following technologies: WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), HSDPA (High Speed ​​Downlink Packet Access), HSUPA (High Speed ​​Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), 5G (5th Generation Mobile Telecommunication), Bluetooth (Bluetooth™ Frequency Identification), Infrared Communication (Infrared Data Association; IrDA), UWB (Ultra-Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus).

[0072] Next, the storage unit (120) may be configured to store various information related to the present invention. In the present invention, the storage unit (120) may be provided in the counseling service providing system (100) itself, or alternatively, at least a part of the storage unit (120) may mean a database (Database: DB, 200).

[0073] The storage unit (120) may include one or more non-transient computer-readable storage media that can be read and / or accessed by at least one processor. One or more computer-readable storage media may include volatile and / or non-volatile storage components such as optical, magnetic, organic, or other memory or disk storage devices. In some examples, the storage unit (120) may be implemented using a single physical device (e.g., one optical, magnetic, organic, or other memory or disk storage device), whereas in other examples, the storage unit (120) may be implemented using multiple physical devices.

[0074] The storage unit (120) may include computer-readable instructions and additional data. The storage unit (120) may include a storage necessary to perform at least some of the methods and techniques described herein and / or at least some of the functions of the device and network.

[0075] Furthermore, at least a portion of the storage unit (120) may be a cloud storage or a cloud server. That is, the storage unit (120) is sufficient as long as it is a space where information necessary for the operation of the counseling service provision system (100) according to the present invention is stored, and it can be understood that there are no restrictions on the physical space. Accordingly, the storage unit (120) and the database (200) may be used interchangeably without being separately distinguished below.

[0076] Data and commands necessary for the operation of the consultation service provision system (100) according to the present invention may be stored in the storage unit (120). Specifically, commands for the operation of the prompt generation unit (131) may be stored in the storage unit (120). The prompt generation unit (131) according to the present invention may refer to a module that generates a prompt to be input into a large language model based on user-related information. There may be a wide variety of methods for generating a prompt to be input into a large language model according to the present invention, and the present specification is not limited to any type or method as long as it is a module capable of generating a prompt to be input into a large language model.

[0077] The counseling service providing system (100) according to the present invention may further include at least one of a module and an algorithm that perform the same function as the prompt generation unit (131). At this time, the prompt generation unit (131) may generate a prompt to be input into a large language model (140) through at least one artificial intelligence model or module. For example, the prompt generation unit (131) may include at least one of a large language model based on T5 (Text-to-Text Transfer Transformer), BART (Bidirectional and Auto-Regressive Transformer), GPT (Generative Pre-trained Transformer), or LLaMA (Language Model for Many Applications), a rule-based template matching algorithm, a conditional prompting technique, a contextual embedding selection module, or a few-shot prompt generator. At this time, the prompt generation unit (131) according to the present invention is not limited to the model, algorithm, or module described above, and the consultation service providing system (100) according to the present invention may further include a model, algorithm, or module that performs the same function as the prompt generation unit (131).

[0078] Furthermore, the storage unit (120) according to the present invention may store commands for the operation of a large language model-based chatbot model (132). Here, the chatbot model (132) according to the present invention may refer to an artificial intelligence model capable of receiving a user query in relation to the user's indication, generating an answer using a large language model, receiving the answer from a large language model (140), and outputting it to a user terminal.

[0079] Specifically, the chatbot model (132) can provide answers to user queries to the user terminal (10) by linking with a large language model. For example, the chatbot model (132) can provide answers to user queries to the user terminal (10) by linking with at least one large language model based on at least one of GPT (Generative Pre-trained Transformer), T5 (Text-to-Text Transfer Transformer), Vision-Language Models (e.g., CLIP, Flamingo), Multimodal Generative Models (GPT-4 Multimodal, PaLM-E), BERT (Bidirectional Encoder Representations from Transformers), and LaMDA (Language Model for Dialogue Applications).

[0080] Meanwhile, the storage unit (120) may store user-related information including at least one of medical information, exercise history information, question and answer information, and exercise program information. As an example, the medical information according to the present invention may include at least one of the user's age, the user's gender, the user's medical history, prescription information, and treatment plan. Additionally, the exercise history information according to the present invention may include at least one of the date of exercise performance, the composition of exercise movements included in the exercise program performed, the results of motion analysis for each exercise movement, and the results of exercise history analysis.

[0081] Furthermore, the question-and-answer information according to the present invention may refer to information related to at least one of a user query entered by a user and an answer from a chatbot model when a user performs consultation using a chatbot model. Specifically, the question-and-answer information may include at least one of the content, topic, and log records of past user queries, the chatbot model's answer to the user query, and information regarding user feedback (or user response) to the answer.

[0082] In addition, the exercise program information according to the present invention may include at least one of the name of each of a plurality of exercise items constituting an exercise program assigned to a user account, the number of exercises, the timing of the exercises, and information on the difficulty of the exercises, as well as an exercise video and an exercise description corresponding to each of the plurality of exercise items.

[0083] Meanwhile, the database (DB, 200) may be configured to store various information related to the provision of counseling services. Specifically, user-related information for each of multiple users may be stored in the database (200). For example, at least one of medical information, exercise history information, question and answer information, and exercise program information corresponding to each of multiple users may be stored in the database (200).

[0084] Furthermore, the database (200) may store exercise motion content (e.g., exercise name, number of exercises, timing of exercises, difficulty of exercises, etc.) corresponding to each of multiple body parts (e.g., shoulder, elbow, wrist & hand, hip & pelvis, knee, ankle & foot, neck, back, waist, abdomen) related to the indication. Furthermore, the database (200) may store at least one of an exercise video and an exercise description corresponding to each exercise motion content.

[0085] In the present invention, the storage unit (120) is described as existing separately from the database (200), but is not limited thereto, and the storage unit (120) may include the database (200).

[0086] Meanwhile, user authentication information may be stored in the storage unit (120). Here, “user authentication information” may refer to information used in a user authentication process performed to log in to a user account on an electronic device (10). As an example, user authentication information may be various, such as i) ID, ii) password, iii) password pattern, iv) user’s fingerprint authentication information, v) face authentication information, vi) voice authentication information, vii) iris authentication information, viii) vein authentication information, etc., set by the user.

[0087] Next, the control unit (130) may be configured to control the overall operation of the consultation service provision system (100) related to the present invention. Specifically, the control unit (130) may include at least one of a prompt generation unit (131) and a chatbot model (132). The control unit (130) may process signals, data, information, etc. that are input or output through the components described above, or provide or process appropriate information and functions to the user.

[0088] The control unit (130) can control the output of a service page for providing a consultation service through a display unit (or touchscreen) provided in the user terminal (10) and the medical staff terminal. Such a service page may be output on the user terminal (10) and the medical staff terminal through an application or web page installed on the user terminal (10) and the medical staff terminal. The service page is a page linked to the consultation service providing system (100) according to the present invention and is configured to be controlled by the consultation service providing system (100) according to the present invention.

[0089] Furthermore, when a service page is provided in the form of an application, the service page may be controlled by a CPU (Central Processing Unit) of a user terminal (10) on which the application is installed. In this case, the CPU of the user terminal (10) may provide a customized consultation service to the user based on information provided by the consultation service providing system (100) according to the present invention.

[0090] Meanwhile, the control unit (130) collects user-related information from the storage unit (120, or database (200)) and, based on the user-related information, can identify similar users among different users who have characteristics similar to the user. The method for identifying similar users who have characteristics similar to the user according to the present invention may be very diverse, and the present specification does not limit the method for identifying similar users who have characteristics similar to the user. For example, the control unit (130) can embed the collected user-related information and, based on the similarity between the embedding vectors, identify similar users who have characteristics similar to the user from the database. Specifically, the control unit (130) can embed the collected user-related information to generate target user vectors corresponding to each of the multiple characteristics (e.g., age, gender, purpose of exercise, pain area, exercise history, question and answer records, etc.) included in the user-related information.

[0091] Furthermore, the control unit (130) can calculate the similarity between a registered user vector and a target user vector corresponding to user-related information of each of the plurality of users stored in the database (200), and identify similar users having characteristics similar to the user based on the calculated similarity. As an example, the control unit (130) can calculate a similarity score based on the cosine similarity between the target user vector and the registered user vector, and determine users whose calculated similarity score is above a specific threshold as similar users. At this time, the specific threshold for the similarity score according to the present invention is not limited to a specific value and can be set in various ways by at least one of the user terminal, the medical staff terminal, and the consultation service providing system (100).

[0092] Furthermore, the control unit (130) can create a user group that includes users and similar users. The control unit (130) can create a group data set based on user-related information of each of the multiple users (or group members) included in the user group. Here, the “group data set” can be understood as including user-related information of each of the group members of a specific user group.

[0093] The control unit (130) according to the present invention includes a chatbot model (132) for building a group chatbot model, and the chatbot model (132) may be implemented to receive a user query in relation to the user's indication, generate an answer using a large language model, receive the answer from the large language model (140), and output it to a user terminal. For example, the “chatbot model (132)” according to the present invention may include at least one of a rule-based chatbot model, an artificial intelligence chatbot model, a natural language processing chatbot model, and a hybrid chatbot model.

[0094] In the following, it can be understood that the control unit (130) builds a group chatbot model based on the chatbot model (132), and in this specification, the chatbot model (132) and the group chatbot model may be used interchangeably.

[0095] The control unit (130) can fine-tune (or tune, optimize) a large language model using a group data set corresponding to a user group. Specifically, the control unit (130) can build a group chatbot model corresponding to the user group by fine-tune (or tune, optimize) the response tendencies and expression methods of the large language model to suit the characteristics of the user group based on user-related information included in the group data set. At this time, the large language model according to the present invention may be included in an existing LLM server (140). For example, the LLM server (140) may include at least one large language model based on at least one of GPT (Generative Pre-trained Transformer), T5 (Text-to-Text Transfer Transformer), Vision-Language Models (e.g., CLIP, Flamingo), Multimodal Generative Models (GPT-4 Multimodal, PaLM-E), BERT (Bidirectional Encoder Representations from Transformers), and LaMDA (Language Model for Dialogue Applications).

[0096] In the present invention, the LLM server (140) is described as existing separately from the counseling service providing system (100), but is not limited thereto, and the counseling service providing system (100) may be configured to include the LLM server (140). That is, the counseling service providing system (100) and the LLM server (140) according to the present invention may exist separately, or the LLM server (140) may be included in the counseling service providing system (100). For convenience of explanation, the LLM server (140) and the large language model are used interchangeably below, and the use of the large language model by the control unit (130) can be understood as using at least one large language model included in the LLM server (140).

[0097] The control unit (130) can build a group chatbot model corresponding to a user group using a large language model fine-tuned with a group data set, and provide a consultation service for a user query input from a user terminal (10). Specifically, the control unit (130) can generate a prompt to be input into a large language model (140) based on a user query input from a user terminal (10) and user-related information using a prompt generation unit (131). At this time, the prompt generation unit (130) can generate a prompt to be input into a large language model (140) by considering user-related information of the group members constituting the user group. In addition, although the present invention describes the control unit (130) as including a prompt generation unit (131), it is not limited thereto, and the control unit (130) and the prompt generation unit (131) may exist separately. At this time, the control unit (130) can generate a prompt in conjunction with the prompt generation unit (131).

[0098] Furthermore, the large language model (140) according to the present invention can generate an answer corresponding to a prompt and provide it to the group chatbot model. At this time, the group chatbot model can provide different answers for each user group according to the group data set used to fine-tune the large language model (140).

[0099] Meanwhile, the control unit (130) can generate a question-and-answer list containing question-and-answer information between the members of a user group and a group chatbot model, and provide the question-and-answer list to the members. Furthermore, the control unit (130) can provide a consultation page to the user terminal (10) that allows for a continuation of the question-and-answer corresponding to any one of the question-and-answer items in the question-and-answer list provided to the members. For example, the control unit (130) can provide a chat room to the user terminal (10) that allows for a continuation of the question-and-answer with the group chatbot model corresponding to the specific question-and-answer item, based on user input for a specific question-and-answer item.

[0100] Furthermore, the control unit (130) may provide a group chat room to the user terminal (10) in which members of a user group can participate. Specifically, the control unit (130) may provide a group chat room to a service page displayed on the user terminal (10) in which members of a specific user group can perform a conversation. There may be many different methods for providing a group chat room to which members of a user group can perform a conversation according to the present invention, and the present specification is not limited to any type or method as long as it is a module capable of providing a group chat room to which members of a user group can perform a conversation.

[0101] At this time, the control unit (130) can selectively connect members of a specific user group and other user groups to a group chat room corresponding to the specific user group, thereby creating a group chat room where members of different groups can converse with each other. For example, the control unit (130) can display an invitation icon in a part of the service page that can invite guest users of a specific user group and other user groups. Furthermore, based on the input of a specific user input regarding the invitation icon, the control unit (130) can transmit an invitation request to the guest user terminal that enables the guest user corresponding to the specific invitation icon to perform a question and answer in the group chat room.

[0102] The control unit (130) can invite guest users of a specific user group and a different user group to a group chat room corresponding to the specific user group based on the occurrence of an approval event for an invitation request.

[0103] Furthermore, the control unit (130) may use question and answer information from a group chat room during the prompt generation process. For example, the control unit (130) may input the chat room question and answer information from the group chat room, along with user-related information and user queries, into the prompt generation unit (131) to generate a prompt that includes the question and answer information from the group chat room.

[0104] Meanwhile, the control unit (130) can update the exercise program set in the user account according to the user query (10) by using a large language model (140). The method for updating the exercise program set in the user account according to the present invention may be very diverse, and the present specification is not limited to the type and method as long as it is a module capable of updating the exercise program set in the user account. Here, “exercise program update” may mean changing at least one of the exercise movements, the number of repetitions of the movements, and the total exercise time that constitute the exercise program.

[0105] For example, the control unit (130) may, based on receiving a user query related to an exercise program update, transmit the updated exercise program to a medical staff terminal and request approval for the updated exercise program from the medical staff. Furthermore, the control unit (130) may update the exercise program set in the user account based on receiving an approval event for the updated exercise program from the medical staff terminal.

[0106] Meanwhile, the counseling service providing system (100) may include one or more processors, and such processors may include one or more general-purpose processors and / or one or more special-purpose processors (e.g., digital signal processors, tensor processing units (TPUs), graphics processing units (GPUs), neural network processing units (NPUs), application-specific integrated circuits, application-specific semiconductors (ASICs), etc.). One or more processors may be configured to execute instructions, computer-readable instructions, and / or other instructions described herein that are stored (or included) in the storage unit (120). The counseling service providing system (100) may perform data processing described below in cooperation with memory and at least one processor. The processor may perform a series of operations and data processing using data and information stored in memory. Here, “memory” may be a component of the storage unit (120), and “processor” may be used interchangeably with the control unit (130).

[0107] In the above description, the consultation service provision system (100) of the present invention has been described, and it can be implemented based on the method of providing consultation services using a large language model-based chatbot described below.

[0108] Hereinafter, with reference to FIG. 2 together with FIG. 3, FIG. 4, FIG. 5, FIG. 6a, FIG. 6b, FIG. 7a, and FIG. 7b, a method for providing a consultation service using a large language model-based chatbot according to the present invention will be described in more detail. FIG. 2 is a flowchart for explaining a method for providing a consultation service using a large language model-based chatbot according to the present invention, and FIG. 3 is a conceptual diagram for explaining user-related information according to the present invention. FIG. 4 is a conceptual diagram for explaining a process of optimizing a large language model according to a user group according to the present invention, and FIG. 5 is a conceptual diagram for explaining a process of generating a prompt to be input into a large language model according to the present invention. FIG. 6a and FIG. 6b are conceptual diagrams for explaining a process of providing an exercise guide by analyzing video data according to the present invention, and FIG. 7a and FIG. 7b are conceptual diagrams for explaining a process of providing an answer to a user query through a group-based chatbot according to the present invention.

[0109] In the present invention, a process of collecting user-related information related to a user account logged into a user terminal may be performed (S210, see FIG. 2).

[0110] The control unit (130) can verify the user account logged into the user terminal (10). Furthermore, the control unit (130) can collect user-related information corresponding to the user account from at least one of the storage unit (120), the database (200), and an external server (e.g., a medical staff server). Here, “user-related information” may include at least one of medical information, exercise history information, question and answer information, and exercise program information related to the user. The method for collecting user-related information corresponding to the user account according to the present invention may be very diverse, and the present specification is not limited to the type and method as long as it is a module capable of collecting user-related information corresponding to the user account.

[0111] For example, the control unit (130) can verify user identification information corresponding to a user account in order to collect user-related information. At this time, the user identification information may include various identification information such as a unique identifier (ID) of the user account, a session token, and a login token, and the control unit (130) can identify (or verify) the user account logged into the user terminal using the user identification information.

[0112] Based on the identification of a user account, the control unit (130) can retrieve user-related information corresponding to the identified user account from at least one of a storage unit (120) that stores user-related information corresponding to the identified user account information, a database (200), and an associated external server. Furthermore, the control unit (130) can collect the retrieved user-related information as user-related information corresponding to the user account.

[0113] As an example, the control unit (130) may collect medical information including at least one of the user's age, the user's gender, the user's medical history, prescription information, and treatment plan. Additionally, the control unit (130) may collect exercise history information including at least one of the user's exercise performance date, the composition of exercise movements included in the exercise program performed, motion analysis results for each exercise movement, and exercise history analysis results.

[0114] Furthermore, the control unit (130) may collect question-response information including at least one of a user query entered by a user into the chatbot model (132) and an answer from the chatbot model (132) to the user query. Specifically, the question-response information may include at least one of the content, topic, and log records of past user queries, an answer from the chatbot model (132) to the user query, and information regarding user feedback (or user response) to the answer.

[0115] Additionally, the control unit (130) can collect exercise program information including at least one of the name of each of a plurality of exercise items constituting an exercise program assigned to a user account, the number of times exercise, the time of exercise, and the difficulty of exercise, as well as an exercise video and an exercise description corresponding to each of the plurality of exercise items.

[0116] As illustrated in FIG. 3, the database (200) according to the present invention may store user-related information (310) including at least one of medical information (320), exercise history information (330), question and answer information (340), and exercise program information (350) collected from a user terminal (10) and a medical staff terminal (20). In this way, user-related information for each of a plurality of different users may be stored in the database (200).

[0117] Next, in the present invention, based on user-related information, a process of identifying similar users having characteristics similar to the user among different users may be carried out (S220, see FIG. 2).

[0118] The control unit (130) can collect similar user-related information (360) from the database (200) based on the collected user-related information. Specifically, the control unit (130) can search for similar user-related information in the database based on the similarity with the user-related information. For example, the control unit (130) can obtain a target user vector corresponding to the user-related information by embedding the collected user-related information. Specifically, the control unit (130) can obtain a target user vector corresponding to the user-related information by embedding the collected user-related information using a pre-prepared embedding model. Here, the “embedding model” may refer to an artificial intelligence model that converts high-dimensional data into a low-dimensional vector.

[0119] For example, the embedding model according to the present invention may include at least one of text embedding models such as Word2Vec, GloVe (Global Vectors), and BERT; code embedding models such as Code BERT, GraphCodeBERT, and Codex; and image embedding models such as VGG (Visual Geometry Group), RESNET (Residual Networks), and CLIP. The embedding model according to the present invention is not limited to the models described above and may further include models that perform the same function as the embedding model according to the present invention.

[0120] Furthermore, the control unit (130) can identify similar user-related information in the database (200) using the acquired target user vector. At this time, it can be understood that the database (200) stores registered user vectors in which user-related information corresponding to each of the different users is embedded.

[0121] Specifically, the control unit (130) can calculate the similarity between a registered user vector and a target user vector corresponding to each of the different users stored in the database. For example, the control unit (130) can calculate the similarity based on at least one of the cosine similarity and the Euclidean distance between a user query vector and a registered user vector stored in the database (200).

[0122] Furthermore, the control unit (130) may specify as similar user-related information that corresponds to at least one registered user vector whose calculated similarity satisfies a preset similarity condition. Here, the preset similarity condition may refer to a condition satisfied based on the similarity between the target user vector and the user vector registered in the database (200) being greater than or equal to a threshold value. At this time, the threshold value is not limited to a specific value, but can be set to various values ​​and can be changed by the system.

[0123] The control unit (130) can identify a specific user corresponding to similar user-related information as a similar user based on similarity. Specifically, the control unit (130) can identify the account of the specific user corresponding to the identified similar user-related information and identify the specific user associated with the specific user account as a similar user having characteristics similar to the user. At this time, although the control unit (130) describes identifying similar users among different users, it can be understood that only one similar user may be identified based on the similarity criteria.

[0124] That is, in the present invention, the number of similar users may mean at least one, and below, the control unit (130) is described as specifying a plurality of similar users having characteristics similar to the user.

[0125] Next, in the present invention, a process of creating a user group including users and similar users may be carried out (S230, see FIG. 2).

[0126] The control unit (130) can create multiple different user groups that include users with similar characteristics. Specifically, the control unit (130) can create a specific user group that includes a specific user and users similar to the specific user. At this time, the users included in the specific user group can be named as members of the specific user group.

[0127] The method for creating a user group according to the present invention may be very diverse, and the present specification is not limited to any type or method as long as it is a module capable of creating a user group. For example, the control unit (130) may create a common group identifier (Group ID, GID) corresponding to the user group to create and identify the user group, and may assign the created GID to each user account included in the user group. At this time, the GID is a unique identification value and may be generated by a hash value, a serial number, and a group attribute value. Furthermore, the control unit (130) may index the user group using the GID as a key value and map the GID to the user account corresponding to each user account, thereby setting users having the same GID to belong to the same user group.

[0128] Additionally, the control unit (130) may store group metadata corresponding to a user group in a storage unit (120) or a database (200), and the group metadata may include information such as a GID, a list of group members, and the time of group creation.

[0129] Next, in the present invention, a process of generating a group data set based on user-related information of a plurality of users included in a user group may be carried out (S240, see FIG. 2).

[0130] The control unit (130) can generate multiple different user groups and extract user-related information for each group member of each user group. Furthermore, the control unit (130) can generate a group data set using user-related information of group members belonging to the same user group. The method for generating a group data set according to the present invention can be very diverse, and the present specification is not limited to any type or method as long as it is a module capable of generating a group data set.

[0131] As illustrated in FIG. 4, the control unit (130) may create a plurality of user groups, including a first user group and a second user group, based on different characteristics. For example, the control unit (130) may create a first user group (410) including a first user (411) and a third user (413), which is a similar user having characteristics similar to the first user (411), based on the fact that the user account logged into the user terminal (10) is the first user account of the first user (411). As another example, the control unit (130) may create a second user group (420) including a second user (421) and a fourth user (422), which is a similar user having characteristics similar to the second user (421), based on the fact that the user account logged into the user terminal (10) is the second user account of the second user (412).

[0132] Furthermore, the control unit (130) can generate user group-related information by using user-related information of group members belonging to the user group. For example, the control unit (130) can generate first user group-related information (430) by extracting user-related information corresponding to each of the first user (411) and the third user (412) belonging to the first user group from the database (200). At this time, the first user group-related information (430) may include user-related information of all group members belonging to the first user group (410).

[0133] As another example, the control unit (130) may generate second user group related information (440) by extracting user-related information corresponding to each of the second user (421) and the fourth user (422) belonging to the second user group (420) from the database (200). At this time, the second user group related information (440) may include user-related information of all group members belonging to the second user group (420).

[0134] Furthermore, the control unit (130) can generate a group data set using user group-related information. Here, the “group data set” may refer to a data set processed into a preset format by performing preprocessing on user group-related information. At this time, the “preset format” according to the present invention may refer to at least one format among a question-response pair (QA Pair) format consisting of a user query and an answer corresponding to the user query, a command-response pair format consisting of a user's objective instruction and an input and expected response, a line-delimited JSON format based on JSON or JSONL, a nested JSON format, a natural language prompt-based template format, a tokenization-based text sequence format processed using a predefined tokenizer, and a training data format for training a large language model.

[0135] The method for generating a group data set according to the present invention may be very diverse, and the present specification is not limited to any type or method as long as it is a module capable of generating a group data set. At this time, the control unit (130) can generate a group data set through at least one of an algorithm, an artificial intelligence model, and a module. For example, the control unit (130) can generate a group data set corresponding to a user group by performing preprocessing, filtering, normalization, removal of sensitive information, text cleaning, tokenization, and question-answer pair (QA) formatting based on user group-related information.

[0136] Next, in the present invention, a process of constructing a group chatbot model corresponding to a user group using a group dataset based on a pre-trained large language model may be carried out (S250, see FIG. 2).

[0137] As previously explained, the control unit (130) can provide an answer to a user query received from a user terminal by linking the chatbot model (132) with a pre-trained large language model (140). At this time, the control unit (130) can fine-tun the pre-trained large language model using a group data set corresponding to a user group. There may be a wide variety of methods for fine-tuning the pre-trained large language model according to the present invention, and the present specification is not limited to any type or method as long as it is a module capable of fine-tuning the pre-trained large language model.

[0138] For example, the control unit (130) can retrain some of the parameters of a large language model using a group dataset to adjust the model to reflect response tendencies that match the characteristics of the user group. In this process, the control unit (130) may include a validation procedure to prevent overfitting, perform iterative learning (epoch) to improve model performance, and adjust hyperparameters (e.g., learning rate, batch size, etc.).

[0139] Furthermore, the control unit (130) can perform fine tuning by selectively using a group data set to selectively fine-tune only some constituent layers of a pre-trained large model (partial tuning or parameter-efficient tuning).

[0140] Referring again to FIG. 4, when the user according to the present invention is the first user (411), the control unit (130) can fine-tune the parameters of the large language model (140) so that the pre-trained large language model (140) generates a response that matches the characteristics of the first user group (410) by using the first group data set (450) corresponding to the first user group (410). Alternatively, when the user according to the present invention is the second user (421), the control unit (130) can fine-tune the parameters of the large language model (140) so that the pre-trained large language model (140) generates a response that matches the characteristics of the second user group (420) by using the second group data set (460) corresponding to the second user group (420).

[0141] Furthermore, the control unit (130) can build a group chatbot model that generates answers to user queries based on a group data set using the finely tuned large language model. There may be many different methods for building a group chatbot model according to the present invention, and the present specification is not limited to any type or method as long as it is a module capable of building a group chatbot model.

[0142] For example, the control unit (130) can build a group chatbot model that utilizes a large language model finely tuned based on a group data set corresponding to a specific user group as a response engine, and enables the large language model (140) to output an answer (or response) that reflects the group data set (or user group related information) in response to a user query.

[0143] Hereinafter, the operation process of the group chatbot model according to the present invention is briefly described, and a detailed description of the group chatbot model is provided together with the relevant drawings.

[0144] The group chatbot model according to the present invention can provide a natural language-based answer to a query received from a user terminal (10) by linking with a finely tuned large language model. Specifically, the group chatbot model can generate a natural language response (or answer) corresponding to a user query by considering past response examples reflected in a group data set or the question-response patterns of the group members of the user group to which the user belongs.

[0145] As illustrated in FIG. 5, the control unit (130) inputs a user query (500) received from a user terminal (10) into a group chatbot model, and the group chatbot model can generate an answer to the user query by linking with a large language model to generate an answer to the user query. For example, when the control unit (130) receives a first user group query entered by a member of a first user group, the answer to the first user group query can be generated through the first group chatbot model.

[0146] Specifically, the control unit (130) can input the first user query (511) corresponding to the first user and the third user query (512) corresponding to the third user into the first group chatbot model (531). Furthermore, the control unit (130) can generate a first answer (541) for each of the first user query (511) and the third user query (512) by using the first group chatbot model (531) and a large language model (140) finely tuned with the first group data set. The large language model finely tuned with the first group data set transmits the first answer (541) to the first group chatbot model (531), and the first group chatbot model (531) can provide the first answer (541) to the user terminal.

[0147] Alternatively, the control unit (130) may input the second user query (521) corresponding to the second user and the fifth user query (522) corresponding to the fourth user into the second group chatbot model (532). Furthermore, the control unit (130) may generate a second answer (542) for each of the second user query (521) and the fourth user query (522) by using the first group chatbot model (531) and a large language model (140) fine-tuned with the second group data set. The large language model fine-tuned with the second group data set transmits the second answer (542) to the second group chatbot model (532), and the second group chatbot model (532) may provide the second answer (542) to the user terminal.

[0148] Next, in the present invention, a process may be performed to generate an answer corresponding to a user query received from a user terminal using a group chatbot model and to provide the generated answer to the user terminal (S260, see FIG. 2).

[0149] As previously explained, the control unit (130) can generate an answer to a user query by linking the group chatbot model with a large language model. Specifically, the control unit (130) can generate a prompt to be input into the large language model using user queries and user-related information received from a user terminal.

[0150] As illustrated in FIG. 6a, the control unit (130) can use the prompt generation unit (131) to generate a prompt (600) to be input into a pre-trained large language model (140) based on at least one of a user query (500) and user-related information (310). Specifically, the prompt generation unit (131) can generate a prompt (600) including at least one of a user query prompt (610), a medical information prompt (620), an exercise history prompt (630), a question-and-answer prompt (640), and an exercise program prompt (650) corresponding to the user-related information (310).

[0151] Furthermore, the control unit (130) can generate a prompt that generates an answer to a user query based on at least one of user-related information and a group data set. Specifically, the control unit (130) can use the prompt generation unit (131) to generate a prompt that reflects user-related information of the group members belonging to the user group, using the group data set of the user group to which the user account logged into the user terminal (10) belongs. More specifically, the control unit (130) can extract keywords related to the user query from the group data set.

[0152] The method for extracting keywords related to a user query from a group data set according to the present invention may be very diverse, and in this specification, any module capable of extracting keywords related to a user query from a group data set is not limited to any specific type or method. The control unit (130) according to the present invention can extract keywords related to a user query from a group data set through at least one algorithm. For example, the control unit (130) can extract keywords related to a user query from a group data set by using a keyword extraction algorithm stored in the storage unit (120). Here, the “keyword extraction algorithm” may refer to an algorithm that extracts keywords related to a user query from a group data set based on at least one of Named Entity Recognition (NER), KeyBERT, GPT, and Transformer. The keyword extraction algorithm according to the present invention is not limited to the algorithms described above and may further include algorithms that perform the same function as the keyword extraction algorithm according to the present invention.

[0153] The control unit (130) can identify specific user-related information related to the keyword among a plurality of user-related information included in the group data set. Furthermore, the control unit (130) can use the specific user-related information to generate a prompt that generates an answer to a user query.

[0154] As illustrated in FIG. 6b (a), the control unit (130) can extract keywords related to a user query from a first group data set (450) corresponding to the first user group based on the fact that the user of the user account logged into the user terminal (10) is included in the first user group. As an example, the control unit (130) can identify user-related information corresponding to the user's age, gender, body part with pain, and exercise purpose among a plurality of user-related information included in the first group data set (450) based on the extracted keywords. Furthermore, the control unit (130) can generate a first prompt (601) to obtain an answer to the user query (10) using specific user-related information (e.g., female in her 40s or 50s, knee pain, rehabilitation exercise, etc.).

[0155] Alternatively, as illustrated in (b) of FIG. 6b, the control unit (130) may extract keywords related to the user query (10) from a second group data set (460) corresponding to the second user group based on the fact that the user of the user account logged into the user terminal (10) is included in the second user group. As an example, the control unit (130) may specify user-related information corresponding to the user's age, gender, body part with pain, and exercise purpose among a plurality of user-related information included in the second group data set (460) based on the extracted keywords. Furthermore, the control unit (130) may generate a second prompt (602) to obtain an answer to the user query (10) using specific user-related information (e.g., male in his 20s or 30s, shoulder pain, strength training exercise, etc.).

[0156] That is, the control unit (130) can generate a user-customized prompt by using a group data set corresponding to the user group to which the user belongs, depending on which user group the user of the user account logged into the user terminal (10) belongs to, extracting keywords related to the user query (10), and reflecting different user-related information.

[0157] The control unit (130) processes the prompt as input to a finely tuned large language model, thereby obtaining an answer to the user query through the large language model. Specifically, the control unit (130) can input the generated prompt into a finely tuned large language model (140) that corresponds to a group data set to which the user of the user account logged into the user terminal (10) belongs. Furthermore, the control unit (130) can generate an answer to the user query received from the user terminal (10) from the finely tuned large language model (140) that corresponds to the group data set.

[0158] The control unit (130) can provide an answer to a group chatbot model, and through the group chatbot model, provide an answer to a user query to a user terminal. As illustrated in FIG. 7a, if the first user of a user account logged into the user terminal (10) belongs to a first user group, the control unit (130) can input a first user query (511) and first user-related information (311) received from the user terminal (10) into a prompt generation unit (131). Furthermore, the control unit (130) can use the prompt generation unit (131) to generate a first prompt (601) that generates an answer to the first user query (511).

[0159] The control unit (130) can process the generated first prompt (601) as input to a large language model (or a pre-trained large language model, 140) that is fine-tuned to a first group data set corresponding to the first user group. Furthermore, the control unit (130) can obtain a first answer as an answer to the first user query (511) from the large language model (140) that is fine-tuned to the first group data set. The control unit (130) provides the obtained first answer to the first group chatbot model (531), and the first group chatbot model (531) can output the first answer (541) as an answer to the first user query on a consultation page (700) that provides consultation services to the user.

[0160] At this time, in the present invention, even if user queries of the same content are input, different answers can be output by reflecting the group data set of each user group according to the user group to which the user belongs. As illustrated in FIG. 7b (a), the control unit (130) can input a first user query (511) into a first group chatbot model (531) based on the fact that the user of the user account logged into the user terminal (10) is included in the first user group. Furthermore, the first group chatbot model (531) can output a first answer (541) as an answer to the first user query (511) on a consultation page by linking with a large language model (140) that is finely tuned with a first group data set corresponding to the first user group.

[0161] Alternatively, as illustrated in FIG. 7b (b), the control unit (130) may input a second user query (512) that is identical to the first user query (511) into the second group chatbot model (532) based on the fact that the user of the user account logged into the user terminal (10) is included in the second user group. Furthermore, the second group chatbot model (532) may be linked with a large language model (140) that is finely tuned with a second group data set corresponding to the second user group, and may output a second answer (542) containing content different from the first answer (541) as an answer to the second user query (512) on the consultation page.

[0162] In this way, the control unit (130) can build a group-based chatbot model based on the user group to which the user account logged into the user terminal (10) belongs, and provide a user-customized consultation service using the group-based chatbot. There may be many different methods for providing a user-customized consultation service using the group-based chatbot according to the present invention, and the present specification is not limited to any type or method as long as it is a module capable of providing a user-customized consultation service using the group-based chatbot. For example, the consultation service according to the present invention may be conducted in the form of a conversation on the consultation page described above, or it may be implemented in the form of providing feedback information and exercise programs in response to user queries.

[0163] That is, the counseling service according to the present invention is not limited to the described examples and can be modified and implemented to be applicable to various user interfaces and service scenarios.

[0164] In the foregoing, the method for providing a generative model-based counseling service according to the present invention has been described in detail. Below, with reference to FIGS. 8a, 8b, 9a, and 9b, an embodiment using the method for providing a generative model-based counseling service will be described in detail. FIGS. 8a is a conceptual diagram illustrating an embodiment in which a user continues a conversation among group members of a user group, such as a user, according to the present invention; FIGS. 8b is a conceptual diagram illustrating an embodiment in which a conversation is performed between group members through a group chat room of a user group according to the present invention, and a prompt is generated based on the content of the conversation; and FIGS. 9a and 9b are conceptual diagrams illustrating an embodiment in which an exercise program is updated according to a user query according to the present invention.

[0165] Meanwhile, the control unit (130) can provide a follow-up consultation service that allows the consultation performed by another group member to continue using a group data set of the user group to which the user account logged into the user terminal (10) belongs.

[0166] As illustrated in FIG. 8a (a) and (b), the control unit (130) can extract a question-and-answer list (810) containing a plurality of question-and-answer items (811 to 813) related to a user query from a group data set. Here, "question-and-answer item" refers to an item generated based on the consultation history performed by group members of a user group with a group chatbot model, and may mean information including a user query and an answer thereto. Furthermore, the control unit (130) can provide a consultation page (800) to a user terminal (10) capable of performing a follow-up consultation related to at least one question-and-answer item included in the question-and-answer list (810). Here, "follow-up consultation" may refer to a consultation process based on a group-based chatbot model, in which an additional user query regarding a user query related to at least one question-and-answer item is entered, or a supplementary explanation regarding an existing answer and a new question-and-answer related to the existing answer are performed.

[0167] Specifically, the control unit (130) can specify one of the multiple question-and-answer items constituting the question-and-answer list according to the first user input input to the user terminal. Here, “first user input” may mean a user input that specifies one of the multiple question-and-answer items included in the question-and-answer list. For example, “first user input” may mean a user input regarding an icon related to a specific question-and-answer item, and such first user input may be performed in at least one of the following methods: tap, double tap, long press, click, swipe, drag, pinch in, pinch out, and rotate on the icon.

[0168] The control unit (130) may provide a consultation page to a user terminal that can input a user query, based on receiving a first user input that specifies one of the multiple question-and-answer items included in the question-and-answer list, so as to perform subsequent consultation related to the specified question-and-answer item. Here, “receiving input” may mean receiving an input signal (or selection signal) corresponding to the user’s input based on input being made by the user through the input unit configuration provided in the user terminal (10).

[0169] In addition, in the present invention, the input unit does not necessarily refer to a hardware means, but can be understood as a channel for receiving input from a user. The input unit may also be referred to as a user interface module. The input unit may include a touch screen, touch input means (e.g., virtual key, soft key, visual key, touch key, etc.), a computer mouse, a keyboard, a keypad, a touchpad, a trackball, a joystick, a dome switch, a jog wheel, a jog switch, a voice recognition module, or other similar devices. However, the present invention does not limit the type of input unit.

[0170] Meanwhile, the control unit (130) can create a chat room where questions and answers (or conversations) are possible among the group members of the user group to which the user account logged into the user terminal (10) belongs. Specifically, the control unit (130) can create a group chat room where questions and answers are possible between the user included in the user group and multiple users in relation to a user question received from the user terminal (10), and provide it to the user terminal.

[0171] As illustrated in FIG. 8b, the control unit (130) can create a group chat room (820) where a user (e.g., Kim Young-hee) of a user account logged into the user terminal (10) belongs to a first user group, and where a plurality of users belonging to the first user group can engage in question-and-answer (or conversation).

[0172] Furthermore, the control unit (130) can output a group member list (831, 832) to a consultation page including a group chat room (820). At this time, the control unit (130) can output a group member list (831, 832) for each of a plurality of user groups to a user terminal. For example, the control unit (130) can provide a first user group list (831) containing the group members of the first user group and a second user group list (832) containing the group members of a second user group different from the first user group to a group chat room (820) related to the first user group.

[0173] The control unit (130) can implement an invitation function so that a member of a second user group, different from the first user group, can participate in a group chat room (820) associated with the first user group. There may be a wide variety of methods for implementing the invitation function according to the present invention, and the present specification is not limited to any type or method as long as it is a module capable of implementing the invitation function.

[0174] For example, the control unit (130) can generate an invitation icon corresponding to each of the group members included in the group member list in a part of the group member list. Specifically, the control unit (130) can generate an invitation icon (833) corresponding to a specific group member (e.g., user 04) of a second user group different from the first user group in a group chat room (820) associated with the first user group.

[0175] Furthermore, based on receiving a second user input for a specific invitation icon (833) from a user terminal, the control unit (130) may transmit an invitation request to a guest user terminal, which enables a guest user (e.g., user 04) among the group members corresponding to the specific invitation icon (833) to perform a question and answer in a group chat room. At this time, the guest user according to the present invention may refer to a group member among the group members of a specific user group whom one wishes to invite to a group chat room. Furthermore, the guest user terminal may refer to a user terminal in which the guest user's user account is logged in.

[0176] The “second user input” according to the present invention may refer to user input related to an invitation request for a specific guest user among the members of a specific user group. For example, the “second user input” may refer to user input for a specific invitation icon, and such second user input may be performed in at least one of the following ways: tap, double tap, long press, click, swipe, drag, pinch in, pinch out, and rotate on said icon.

[0177] Furthermore, the control unit (130) can provide a group chat room that the guest user can participate in to the user terminal based on the occurrence of an approval event for an invitation request at the guest user terminal.

[0178] Meanwhile, the control unit (130) can generate a prompt to generate an answer to a user query by using question-and-answer information that includes conversation content performed in a group chat room. Referring again to FIG. 8b, the control unit (130) can generate a prompt (600) to generate an answer to a user query (500) by using the prompt generation unit (131) to generate a prompt to generate an answer to a user query (500) by using question-and-answer information that includes conversation content (840) in a group chat room that performs a question-and-answer related to the user query, together with user-related information (310) and user query (500).

[0179] Meanwhile, the control unit (130) can update the exercise program set in the user account according to a user query received from the user terminal. At this time, the exercise program according to the present invention is configured to include a plurality of exercise modules, and each of the plurality of exercise modules is matched to a different exercise type and may include at least one exercise motion content related to the exercise type matched to each exercise module. For example, the different exercise types may mean at least one of a warm-up exercise type, a main exercise type, and a cool-down exercise type.

[0180] As previously described, the control unit (130) can update the exercise program set in the user account according to the user query (10) by using a pre-trained large language model (140). The method for updating the exercise program set in the user account according to the present invention may be very diverse, and the present specification is not limited to any type or method as long as it is a module capable of updating the exercise program set in the user account. Here, “exercise program update” may mean changing at least one of the exercise movements, the number of repetitions of the movements, and the total exercise time that constitute the exercise program. Below, an exercise program update that changes the number of repetitions of the movements is described as an example, but is not limited thereto, and at least one of the exercise movements and the total exercise time may also be changed.

[0181] As illustrated in FIG. 9a, when the control unit (130) performs a question and answer (or conversation) on a consultation page (910) with a group chatbot model, if it receives a user query related to an exercise program update, it can update the exercise program (900) set in the user account according to the user query.

[0182] Specifically, the control unit (130) can update at least one of the plurality of exercise modules (901 to 904) that constitute the exercise program (900) set in the user account according to a user query. As an example, when the control unit (130) performs a question and answer (or conversation) on the consultation page (910) with the group chatbot model, it may receive a user query related to updating the exercise program (e.g., “Please increase the difficulty of the exercise by adjusting the number of repetitions and sets of squat movements!”). Furthermore, the control unit (130) may change the number of repetitions of a specific exercise movement (e.g., squat) that constitutes a specific exercise module (e.g., main exercise, 902) among the plurality of exercise modules that constitute the exercise program (900) set in the user account according to the received user query.

[0183] More specifically, the control unit (130) can generate a prompt for updating an exercise program using a pre-trained large language model (140). The control unit (130) can generate an update prompt for updating the exercise program assigned to the user account using user-related information.

[0184] The control unit (130) can extract text of a pre-set topic related to an exercise program update from a user query and include the text of the pre-set topic in an update prompt. Here, “pre-set topic” may mean at least one of adjusting the difficulty of the exercise, selecting the type of exercise, and changing the type of exercise.

[0185] Furthermore, the control unit (130) can input an update prompt into a large language model (or a pre-trained large language model, a finely tuned large language model, 140) and, through the large language model, generate an updated exercise program according to the user's query.

[0186] Meanwhile, the control unit (130) may perform a medical staff approval process for the updated exercise program before setting the updated exercise program to the user account. Specifically, the control unit (130) may transmit an approval request for the updated exercise program (900) to the medical staff terminal (20).

[0187] As illustrated in FIG. 9b, in the present invention, an approval event for the exercise program approval request (920) may occur based on receiving approval feedback from the medical staff terminal (20). Here, "approval feedback" may refer to response information indicating whether the exercise program approval request (920) received through the medical staff terminal (20) is accepted. The approval feedback according to the present invention may be provided in various ways, such as natural language-based input sentences, clicking approval or rejection icons on a GUI, selecting checkboxes, or voice command input, but is not limited thereto and may be implemented in any input method capable of conveying the medical staff's intention to approve.

[0188] The control unit (130) can set an updated exercise program to a user account based on the occurrence of an approval event in response to an approval request from the medical staff terminal (20). Furthermore, the control unit (130) can output the exercise program (940) for which the update has been approved to the consultation page of the user terminal (10).

[0189] As seen above, the method and system for providing consultation services using a large language model-based chatbot according to the present invention can improve response precision by providing an optimized response to user queries through the construction of a group chatbot based on user groups created by clustering users similar to the user.

[0190] Furthermore, the method and system for providing a consultation service using a large language model-based chatbot according to the present invention links a conversation with the chatbot based on existing question-and-answer information of a user group including the user, thereby allowing the user to continue the consultation by referring to or continuing the consultation flow of similar past users, and thus receive a continuous and context-aware consultation service.

[0191] Furthermore, the method and system for providing a consultation service using a large language model-based chatbot according to the present invention configures a chat room in the form of a user group and reflects the content of multi-party conversations within the group chat room in real time on a prompt, thereby enabling the chatbot to generate a high-precision response that considers not only the user's query but also the conversation context of others within the group, and the user can experience a more meaningful consultation service through mutual communication based on a sense of shared identity.

[0192] Furthermore, the method and system for providing a consultation service using a large language model-based chatbot according to the present invention can flexibly provide a customized exercise program suitable for the user's current health condition by updating the exercise program provided to the user account with the approval of medical staff based on user queries, thereby ensuring medical reliability and safety.

[0193] Furthermore, the consultation service provision system (100) using a large language model-based chatbot according to the present invention can be implemented through a computing device described below and can perform data processing related to the consultation service provision method using the large language model-based chatbot described above.

[0194] Meanwhile, FIG. 10 illustrates an example of a block diagram of a computing system in which the present invention can be implemented.

[0195] Referring to FIG. 10, a computing system (10000) that performs a method for providing a consultation service using a large language model-based chatbot according to one embodiment of the present invention may include at least one computing device. At this time, the at least one computing device may be a single processor or a multi-processor computing device.

[0196] The components of at least one computing device of the present invention may include various hardware components such as one or more processors, memory, other hardware, and a system bus (not shown) that connects various system components so that they can transmit and receive data to and from each other (e.g., telecommutatively connected, physically connected, electrically connected), and the components of at least one computing device are not limited thereto and may be very diverse.

[0197] Meanwhile, at least one computing device included in a computing system (10000) that performs a method of providing consultation services using a large language model-based chatbot may be connected to communicate via a network (1070). For example, at least one computing device included in the computing system (10000) may be clustered or may be part of a local area network (LAN). Additionally, at least one computing device may be part of a wide area network (WAN) or connected to at least one of a client-server network and a peer-to-peer network within the cloud.

[0198] Meanwhile, when at least one computing device is used in at least one of a network environment and a cloud computing environment, the at least one computing device may be connected to at least one of a public and private network through a network interface or adapter. In one embodiment, other communication connection devices, such as a modem, may be used to establish communication through the network. The modem may be at least one of an internal modem and an external modem, and may be connected to a system bus through a network interface or a specific mechanism, etc. A wireless network component consisting of an interface and an antenna may be coupled to the network through a device such as an access point, a peer computer, etc. In the present invention, the method of connecting at least one computing device to communicate through the network (1070) is not limited, and it may be connected to communicate in a manner different from the described example.

[0199] Furthermore, other computer-type devices and / or systems not shown in FIG. 10 may also interact technically with at least one computing device or other system through one or more connections to the network (1070) via a network interface. Here, the network interface may include network interface equipment such as a physical network interface controller (NIC) or a virtual network interface (VIF).

[0200] The network (1070) of the present invention may include various forms such as the Internet, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), HSDPA (High Speed ​​Downlink Packet Access), HSUPA (High Speed ​​Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), 5G (5th Generation Mobile Telecommunication), Bluetooth (Bluetooth™ Frequency Identification), Infrared Communication (Infrared Data Association; IrDA), UWB (Ultra-Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi Direct, Wireless USB (Wireless Universal Serial Bus), etc., and in the present invention, data transmission may be performed based on standard communication protocols such as TCP / IP, HTTP, SSL, etc.

[0201] A computing system (10000) that performs a method for providing a consultation service using a large language model-based chatbot according to the present invention may include at least one of a user computing device (1010, or user computing system), a training computing system (1050, or training computing device), and a server computing system (1030, or server computing device).

[0202] A user computing device (1010) according to the present invention may be understood as a computing device comprising at least one processor (1011) and a memory (1012) for performing a method of providing a consultation service using a large language model-based chatbot. For example, the user computing device (1010) may include at least one computing device among a smartphone, a smart TV, a laptop computer, a desktop computer, a digital broadcasting terminal, a PDA (personal digital assistants), a PMP (portable multimedia player), a navigation device, a slate PC, a tablet PC, an ultrabook, a wearable device (e.g., a smartwatch, a smart glass, and a head-mounted display).

[0203] At least one processor (1011) constituting the user computing device (1010) may include one or more general-purpose processors and / or one or more special-purpose processors. For example, at least one processor (1011) constituting the user computing device (1010) may be composed of at least one or a plurality of electrically connected processors among a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), a neural network processing unit (NPU), an arithmetic logic unit (ALU), a floating-point arithmetic unit (FPU), an application integrated circuit, an application semiconductor (ASIC), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, and / or other electrical units for performing functions.

[0204] Furthermore, at least one processor (1011) may be configured to execute computer-readable instructions contained in memory (1012) and / or other instructions described herein.

[0205] The memory (1012) constituting the user computing system (1010) according to the present invention may include volatile memory, non-volatile memory, fixed media, removable media, magnetic media, optical media, semiconductor media and / or other types of physically durable storage media.

[0206] For example, memory (1012) may include one or more non-transient / transient computer-readable storage media such as RAM, ROM, HDD (Hard Disk Drive), SSD (Solid State Disk), SSD (Silicon Disk Drive), EEPROM, EPROM, flash memory device, magnetic disk, and combinations thereof, and may include web storage of a server that performs the storage function of memory over the internet. Such memory (1012) may store data and instructions necessary for the at least one processor (1011) to perform the operation of an application for providing consultation services using a large language model-based chatbot.

[0207] A user computing device (1010) may include one or more user input components (1021) that detect user input. For example, the user input component (1021) may also be referred to as a user interface module. The user input component (1021) may include a touch screen, a computer mouse, a keyboard, a keypad, a touchpad, a trackball, a joystick, a voice recognition module, or other similar devices. However, the present invention does not limit the type of user input component (1021). In this case, the user input component (1021) in the present invention does not necessarily mean a hardware means, but can be understood as a channel for receiving input from a user. Meanwhile, the user of the present invention may refer to an automated agent, script, playback software, etc., that operates on behalf of one or more people.

[0208] A user can interact with a computing system (10000) including at least one computing device through input text, touch, voice, movement, computer vision, gestures and / or other forms of input / output using a user input component (1021). For example, the user input component (1021) may include one or more of a command line interface (CLI), a graphical user interface (GUI), a natural user interface (NUI), a voice command interface and / or other user interface (UI) representations.

[0209] Between the user input component (1021) and the user computing device (1010), one or more application programming interface (API) calls may be made based on user input received from the user interface and / or network.

[0210] Here, the expression "based on" may be interpreted to include cases where it is based on the use of a specific configuration, modified from, derived from, influenced by, dependent on, or otherwise derived from a specific configuration. In some embodiments, an API call may be configured for a specific API, which may be interpreted or converted into an API call configured for another API. Here, an API may refer to a defined interface or connection between computers or between computer programs.

[0211] In one embodiment, the user computing device (1010) may store at least one machine learning model (1020). For example, the user computing device (1010) may be various machine learning models, such as a plurality of neural networks (e.g., deep neural networks) that provide consultation services using a large language model-based chatbot based on user queries and user-related information, or other types of machine learning models including non-linear models and / or linear models, and may be composed of a combination thereof.

[0212] According to an embodiment of the present invention, a user computing device (1010) may perform a method of providing a consultation service using a large language model-based chatbot by using a local or / and external machine learning model (1020). Alternatively, the user computing device (1010) may perform a method of providing a consultation service using a large language model-based chatbot by using a machine learning model (1040) provided by a server.

[0213] In addition, according to another embodiment of the present invention, a server computing system (1030) communicating with a user computing device (1010) can provide a consultation service using a large language model-based chatbot to the user computing device (1010) on an application or / and the web in accordance with a user's request received through the user computing device (1010).

[0214] In addition, according to another embodiment of the present invention, at least a part of a user computing device (1010) and a server computing system (1030) are interconnected to perform a method of providing a consultation service using a large language model-based chatbot, thereby providing an answer to a user's query to the user.

[0215] Additionally, according to various embodiments of the present invention, a user computing device (1010) and / or a server computing system (1030) can learn machine learning models (1020, 1040) performed in a method for providing consultation services using a large language model-based chatbot through interaction with a training computing system (1050) that is communicatedly connected via a network (1070). In this case, the training computing system (1050) may be a computing system separate from the server computing system (1030). Alternatively, in some embodiments, the training computing system (1050) may be part of the server computing system (1030) or part of the user computing device (1010).

[0216] Meanwhile, the server computing system (1030) may include at least one processor (1031) and memory (1032). Here, the processor (1031) may be composed of at least one or a plurality of electrically connected processors among a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), a neural network processing unit (NPU), an application integrated circuit, an application semiconductor (ASIC), an arithmetic logic unit (ALU), a floating-point arithmetic unit (FPU), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, and / or other electrical units for performing functions. For example, at least one processor (1031) may include a circuit and a transistor configured to execute instructions from memory (1032).

[0217] The memory (1032) constituting the server computing system (1030) according to the present invention may include volatile memory, non-volatile memory, fixed media, removable media, magnetic media, optical media, semiconductor media, and / or other types of physically durable storage media. For example, the memory (1032) may include one or more non-transient / transient computer-readable storage media such as RAM, ROM, HDD (Hard Disk Drive), SSD (Solid State Disk), SSD (Silicon Disk Drive), EEPROM, EPROM, flash memory device, magnetic disk, etc., and combinations thereof, and may include web storage of a server that performs the storage function of memory over the internet. Additionally, the server computing system (1030) may further include a data storage (data store). For example, the data storage may be composed of at least one of a relational database, a NoSQL database, a data warehouse, and a local file system.

[0218] In the memory (1032) constituting the server computing system (1030) according to the present invention, data and instructions necessary for the at least one processor (1031) to perform the operation of an application for providing a consultation service using a large language model-based chatbot may be stored.

[0219] In one embodiment, the server computing system (1030) may be composed of a single device or a plurality of computing devices, and may be configured to operate according to a sequential or parallel computing architecture. Additionally, a distributed processing system may be configured with a plurality of networked devices.

[0220] Meanwhile, the training computing system (1050) may include at least one processor (1051) and memory (1052). The model trainer (1060) is a logical component that executes the training of at least one machine learning model (1020, 1040) and may be implemented in the form of hardware, firmware, or software. For example, the model trainer (1060) may be executed by the processor (1051) after loading training data (1061) stored in a storage device into memory (1052). For example, the model trainer (1060) may be configured to execute one or more operations (e.g., model training, model reconstruction, model validation, model testing) on ​​at least one machine learning model.

[0221] The machine learning model of the present invention may include at least one of a statistical model, an algorithm, a neural network (NN), a convolutional neural network (CNN), a generative neural network (GNN), a Word2Vec model, a Bag of Words model, a TF-IDF (document frequency-inverse document frequency) model, a GPT (Generative Pre-trained Transformer) model (or other autoregressive models), a PPO (Proximal Policy Optimization) model, a nearest neighbor model (e.g., a k-nearest neighbor model), a linear regression model, a K-means clustering model, a Q-learning model, a TD (Temporal Difference) model, a Deep Adversarial Network model, and all other types of models further described herein.

[0222] Specifically, the model trainer (1060) may execute operations to train a machine learning model, and said operations may include at least one of adding, removing, and modifying model parameters. At this time, the training of the machine learning model may be at least one of supervised learning, semi-supervised learning, and unsupervised learning. In one embodiment, the training of the machine learning model may include the step of repeatedly inputting training data (1061) based on epochs and repeatedly performing the machine learning model training process configured in this way. Here, an epoch may refer to a unit in which the entire set of training data (1061) undergoes forward and backpropagation processing once. In some implementations, different levels of training methods (e.g., supervised learning, semi-supervised learning, unsupervised learning) may be used for different epochs.

[0223] The training data (1061) of the present invention may include input data and / or data previously output from at least one machine learning model (e.g., recursive learning feedback). The parameters of at least one machine learning model may include at least one of a seed value, a model node, a model layer, an algorithm, a function, connections between different machine learning models, connections between parameters, machine learning model constraints, and other digital components that influence the output of the machine learning model. In this case, the model connections between different machine learning models may include or represent relationships between model parameters and / or models, which may be dependent or interdependent, hierarchical, and / or static or dynamic. The combinations and configurations of model parameters described herein may be too complex to be maintained or used by human cognitive abilities.

[0224] In the present invention, the machine learning parameters described according to the embodiments are not limited, and a single machine learning model may further include a plurality of model parameters.

[0225] Meanwhile, FIG. 11 illustrates an example of a block diagram of a computing device (1100) that may be included in a user computing device (1010), a server computing system (1030), and a training computing system (1050), as an embodiment of a computing system (10000) in which the present invention can be implemented.

[0226] As illustrated in FIG. 11, the computing device (1100) may include at least one application (e.g., Application 1 to Application N), and each of the at least one application may include a machine learning library and a model execution environment for performing a method of providing consultation services using a chatbot based on a machine learning-based large language model. The at least one application included in the computing device (1100) may communicate with the sensor, context manager, device state manager, or additional component(s) within the computing device (1100) via an Application Programming Interface (API). In one embodiment, the at least one application may interface with device components, such as receiving sensor data or state data or transmitting prediction results to an output device via a public or private API.

[0227] Meanwhile, FIG. 12 illustrates an example of a block diagram in another aspect of a computing device (1200), which is one of the components of a computing system (10000) that performs a method of providing a consultation service using a large language model-based chatbot according to an embodiment of the present invention.

[0228] A computing device (1200) according to the present invention may include at least one application (e.g., Application 1 to Application N), and at least one application may communicate with a central intelligence layer (1210). Each application may interact with a shared model within the central intelligence layer (1210) through an API (e.g., a common API).

[0229] The central intelligence layer (1210) includes one or more machine learning models and may share them among multiple applications or provide them independently to each. In one embodiment, the central intelligence layer (1210) may be integrated as part of an operating system or implemented as a separate logical layer.

[0230] Additionally, the central intelligence layer (1210) can communicate with the central device data layer (1220). The central device data layer (1220) can provide user-related information and user queries stored within the computing device (1200) as input data required for providing consultation services using a large language model-based chatbot. Each device component (e.g., sensor, state manager, etc.) can communicate with the central device data layer (1220) through a private API, etc.

[0231] The technology described in this specification may be composed of a single or multiple computing devices, and the machine learning model performing the method of providing consultation services using a large language model-based chatbot may be executed sequentially or in parallel on one component or multiple distributed components. The data storage, machine learning model, and application may be distributed and operated locally or over a network, and these configurations can be flexibly applied to various system architectures.

[0232] Meanwhile, computer-readable media include all types of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SSD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

[0233] Furthermore, the computer-readable medium may be a server or cloud storage that includes a storage and is accessible to an electronic device via communication. In this case, the computer may download the program according to the present invention from the server or cloud storage via wired or wireless communication.

[0234] Furthermore, in the present invention, the computer described above is an electronic device equipped with a processor, namely a CPU (Central Processing Unit), and no special limitations are placed on its type.

[0235] Meanwhile, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention.

Claims

1. A step of collecting user-related information associated with a user account logged into a user terminal; Based on the above user-related information, a step of identifying similar users among different users who have characteristics similar to the above user; A step of creating a user group including the above user and the above similar users; A step of generating a group data set based on user-related information of a plurality of users included in the above user group; A step of constructing a group chatbot model corresponding to the user group using the group dataset based on a previously trained large language model; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of generating an answer corresponding to a user query received from the user terminal using the group chatbot model above, and providing the generated answer to the user terminal.

2. In Paragraph 1, The step of building the above group chatbot model is, A step of fine-tuning the previously trained large language model using the group data set corresponding to the user group; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of constructing a group chatbot model that generates an answer to a user query based on a group dataset using the finely tuned large language model.

3. In Paragraph 2, The step of providing the above answer to the user terminal is, A step of generating a prompt that generates the answer to the user query based on at least one of the above user-related information and the above group data set; A step of processing the above prompt as input to the above fine-tuned large language model to obtain the above answer to the user query through the above large language model; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of providing the above answer to the group chatbot model and, through the group chatbot model, providing the above answer to the user query to the user terminal.

4. In Paragraph 3, The step of generating the above prompt is, A step of extracting keywords related to the user query from the group data set; A step of identifying specific user-related information related to the keyword among a plurality of user-related information included in the group data set; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of generating a prompt that generates an answer to a user query using the specific user-related information.

5. In Paragraph 1, The step of identifying the aforementioned similar users is, A step of searching for similar user-related information in a database based on similarity with the above user-related information; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of identifying a specific user corresponding to the similar user-related information as a similar user based on the similarity.

6. In Paragraph 5, The step of searching for the above-mentioned similar user-related information is, A step of embedding the above user-related information to obtain a target user vector corresponding to the above user-related information; A step of calculating the similarity between a registered user vector corresponding to each of the different users stored in the database and the target user vector; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of specifying registered user-related information corresponding to at least one registered user vector satisfying a preset similarity condition as similar user-related information.

7. In Paragraph 1, The step of providing the above answer to the user terminal is, A step of extracting a query-response list including a plurality of query-response items related to the user query from the group data set; and A method for providing a consultation service using a large language model-based chatbot, characterized by further including the step of providing a consultation page to a user terminal capable of performing follow-up consultation related to at least one question-and-answer item included in the above question-and-answer list.

8. In Paragraph 7, The step of providing the above consultation page to a user terminal is: A step of specifying one of the plurality of question-and-answer items constituting the question-and-answer list according to a first user input input to the user terminal; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of providing a consultation page to a user terminal where a user query can be entered, so as to perform a follow-up consultation related to any one of the aforementioned specified question-and-answer items.

9. In Paragraph 3, In relation to the above user query, the method further includes the step of creating a group chat room capable of query-answering between the user included in the user group and the plurality of users, and providing it to the user terminal. In the step of generating the above prompt, A method for providing a consultation service using a large language model-based chatbot, characterized by generating a prompt that generates an answer to a user query using question-and-answer information including conversation content performed in the group chat room.

10. In Paragraph 9, In the step of providing the above group chat room to the user terminal, A step of outputting a list of group members for each of multiple user groups to a user terminal; A step of generating an invitation icon corresponding to each of the group members included in the group member list in a part area of ​​the group member list; Based on receiving a second user input for a specific invitation icon from a user terminal, a step of transmitting an invitation request to a guest user terminal where the guest user's user account is logged in, so that the guest user corresponding to the specific invitation icon among the group members can perform a question and answer in the group chat room; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of providing the group chat room that the guest user can participate in to the user terminal based on the occurrence of an approval event for the invitation request.

11. In Paragraph 1, The above user-related information is, A method for providing a consultation service using a large language model-based chatbot, characterized by including at least one of medical information, exercise history information, question and answer information, and exercise program information related to the user.

12. In Paragraph 1, The method further includes the step of updating the exercise program set in the user account according to the user query received from the user terminal, and The step of updating the above exercise program is, A step of generating an update prompt to update the exercise program assigned to the user account using the above user-related information; A step of inputting the above update prompt into the above large language model, and generating an updated exercise program according to the user query through the above large language model; A step of transmitting an approval request for the above-mentioned updated exercise program to a medical staff terminal; and A method for providing a consultation service using a large language model-based chatbot, characterized by including the step of setting the updated exercise program to the user account based on the occurrence of an approval event according to the approval request from the medical staff terminal.

13. A communication unit that receives a user query from a user terminal; and It includes a control unit that collects user-related information related to a user account logged into the user terminal, and The above control unit is, Based on the above user-related information, similar users having characteristics similar to the above user are identified among different users, and Creates a user group including the above user and the above similar users, Based on user-related information of multiple users included in the above user group, a group data set is generated, and Based on a pre-trained large language model, a group chatbot model corresponding to the user group is constructed using the group dataset, and A consultation service provision system using a large language model-based chatbot, characterized by using the group chatbot model above to generate an answer corresponding to the user query received from the user terminal and providing the answer to the user terminal.

14. A program that is executed by one or more processes in an electronic device and stored on a computer-readable recording medium, The above program is, A step of collecting user-related information associated with a user account logged into a user terminal; Based on the above user-related information, a step of identifying similar users among different users who have characteristics similar to the above user; A step of creating a user group including the above user and the above similar users; A step of generating a group data set based on user-related information of a plurality of users included in the above user group; A step of constructing a group chatbot model corresponding to the user group using the group dataset based on a previously trained large language model; and A program stored on a computer-readable recording medium, characterized by including instructions that, using the group chatbot model above, generate an answer corresponding to a user query received from the user terminal and provide the answer to the user terminal.

Images