System for providing artificial intelligence-based counseling chatbot

Abstract

A system for providing an artificial intelligence-based counseling chatbot may comprise a server and an administrator electronic device having management authority over the server. The server may comprise: a database including content and member information; a wireless communication circuit configured to provide wireless communication with an external electronic device; and a processor configured to train a large language model-based artificial intelligence model on the basis of the content. The server may be configured to provide a virtual space to provide a counseling service, the virtual space being accessible by an external electronic device. The processor may be configured to: receive a question text from the external electronic device connected to the server; extract, from the question text, a prompt to be input to the large language model-based artificial intelligence model; generate a response message based on the content stored in the database, by inputting the prompt to the large language model-based artificial intelligence model; and provide the response message to the external electronic device through the virtual space of the server.

Description

A system that provides a consulting chatbot based on artificial intelligence

[0001] The present disclosure relates to a system that provides a consulting chatbot based on artificial intelligence.

[0002] For service industries providing counseling services, companies are providing counseling chatbots using AI based on large language models (LLMs) with proprietary content. These LLM-based AI models are used for counseling services utilizing reinforcement learning with human feedback (RLHF) techniques, such as ChatGPT4 and Gemini. These AI-based chatbots leverage natural language processing and machine learning technologies to provide human-like conversations, enabling high work efficiency and diverse data collection.

[0003] The above information is provided as background information to aid in understanding the invention disclosed herein. No claim or determination is made as to whether any of the above is applicable as prior art in connection with the present disclosure.

[0004] Chatbots offering consultation services can struggle to provide accurate answers to questions if they don't reflect the specifics of the service area being offered. Furthermore, offering unlimited consultations can lead to persistent, free consultations without a clear revenue stream, making it difficult to generate revenue. Furthermore, increased traffic can lead to problems. Furthermore, responses unrelated to the content can provide users (customers) with incorrect information.

[0005] A system providing a consultation chatbot based on an artificial intelligence model according to one embodiment can enable accurate and rapid consultation by operating a consultation chatbot using an artificial intelligence model trained based on content.

[0006] A system for providing an artificial intelligence-based counseling chatbot according to one embodiment may include a server and an administrator electronic device having management authority over the server. The server may include a database including content and member information uploaded by the administrator electronic device, a wireless communication circuit configured to provide wireless communication with an external electronic device, and a processor configured to train an artificial intelligence model based on a large-scale language model based on the content included in the database. The server may be configured to provide a virtual space to which an external electronic device can access through the wireless communication circuit to provide a counseling service. The processor may be configured to receive a question text from the external electronic device connected to the server through the wireless communication circuit, extract a prompt to be input into the artificial intelligence model based on the large-scale language model from the question text, input the prompt into the artificial intelligence model based on the large-scale language model, thereby generating a response message based on the content stored in the database, and providing the response message to the external electronic device through the virtual space of the server.

[0007] According to one embodiment, the processor may be configured to count the amount or number of times the response message is provided, and collect and analyze a communication log of the external electronic device based on identifying that at least one of the accumulated amount or number of times the response message is provided exceeds a threshold, identify whether the external electronic device is a member based on the analysis result, and if it is identified that the external electronic device is a member, continuously provide the response message, and if it is identified that the external electronic device is not a member, stop providing the response message, and provide a first guidance message for guiding membership to the external electronic device through the virtual space of the server.

[0008] In one embodiment, the first guidance message may include a link that can be accessed by a website address that provides membership registration.

[0009] According to one embodiment, the processor may be configured to connect the external electronic device to the website based on identifying that a first amount of time has elapsed from a first time point at which the first guidance message was provided.

[0010] According to one embodiment, the processor may be configured to provide, to the external electronic device via the wireless communication circuit, first information for rendering the virtual space, second information for rendering an avatar displayed within the virtual space, and third information for rendering the response message to be displayed adjacent to the avatar included within the virtual space.

[0011] According to one embodiment, the processor may be configured to generate the avatar based on the content contained within the database.

[0012] According to one embodiment, the processor may be configured to provide the first information and the second information based on the question text.

[0013] According to one embodiment, the processor may be configured to provide a second guidance message to the external electronic device through the virtual space of the server to guide that the response message is being generated based on identifying that a second time has elapsed from the second time point at which the question text was received.

[0014] According to one embodiment, the processor may be configured to identify a time elapsed from the second point in time and provide a first visual object representing the elapsed time to the external electronic device through the virtual space of the server together with the second guidance message.

[0015] According to one embodiment, the processor may be configured to calculate a time required to generate the response message and provide a second visual object representing the required time to the external electronic device through the virtual space of the server together with the second guidance message.

[0016] According to one embodiment, the manager electronic device can store history data corresponding to information related to the provision history of the response message in the database.

[0017] According to one embodiment, the processor may be configured to identify a similarity between the question text and the content, and provide the response message to the external electronic device based on identifying the similarity that is greater than or equal to a first similarity, identify one or more keywords within the content based on the question text based on identifying the similarity that is less than the first similarity, generate an asking back text corresponding to the question text using the one or more keywords, and provide the asking back text to the external electronic device through the virtual space of the server.

[0018] According to one embodiment, the processor may be configured to provide a third guidance message to the external electronic device through the virtual space of the server to guide that the response message for the question text cannot be provided based on identifying the similarity that is lower than or equal to a second similarity that is lower than the first similarity.

[0019] A system providing a consultation chatbot based on an artificial intelligence model according to one embodiment can enable accurate and rapid consultation by operating a consultation chatbot using an artificial intelligence model trained based on content.

[0020] FIG. 1 is a block diagram illustrating components of a system that provides an artificial intelligence-based consulting chatbot according to one embodiment.

[0021] Figure 2 is a flowchart illustrating the operation of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0022] FIG. 3 illustrates an example of consultation content provided to an external electronic device through a system that provides an artificial intelligence-based consultation chatbot according to one embodiment.

[0023] FIG. 4 is a flowchart illustrating an operation of providing a first guidance message of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0024] FIG. 5 illustrates an example of an external electronic device accessing a website via a first guidance message of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0025] FIG. 6 illustrates an example of an external electronic device displaying a virtual space and an avatar of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0026] FIGS. 7A and 7B illustrate an example of an external electronic device displaying a second guidance message of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0027] FIG. 8 is a flowchart illustrating an operation of providing a response text of a system providing a consultation chatbot based on artificial intelligence according to one embodiment.

[0028] FIGS. 9A, 9B, and 9C illustrate examples of external electronic devices displaying response text of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0029] FIG. 10 illustrates an external electronic device using a department change service of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0030] FIG. 11 illustrates an example of a history of providing a response message registered through an electronic device of an administrator of a system providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0031] The following descriptions, with reference to the attached drawings, are provided to facilitate a comprehensive understanding of various embodiments of the present disclosure as defined by the claims and their equivalents. While various details are provided below to facilitate understanding, they should be considered merely exemplary. Accordingly, those skilled in the art will recognize that various modifications and variations can be made to the various embodiments described herein without departing from the scope and spirit of the present disclosure. Furthermore, descriptions of known functions and configurations may be omitted for clarity and conciseness.

[0032]

[0033] FIG. 1 is a block diagram showing components of a system (100) that provides an artificial intelligence-based consulting chatbot according to one embodiment.

[0034] A system (100) providing a consultation chatbot based on artificial intelligence according to one embodiment is provided. The system (100) providing a consultation chatbot based on artificial intelligence may be briefly referred to as "system (100)" in the following description. The system (100) according to one embodiment may be configured to receive a question text from an external electronic device (200), identify the intent of the question of a user of the external electronic device (200) from the question text, and then automatically provide a consultation chatbot based on a database (121) uploaded to a server (120). The consultation chatbot may be referred to as a consultation service that receives a question text written in natural language through the external electronic device (200), understands the question text, and provides a response message corresponding to the question text based on artificial intelligence based on a large-scale language model.

[0035] A system (100) according to one embodiment may include an administrator electronic device (110), a server (120), and a website (130).

[0036] According to one embodiment, the administrator electronic device (110) may be configured to manage the overall operation of the system (100). The administrator electronic device (110) may have management authority over the server (120). For example, the administrator electronic device (110) may instruct the processor (124) to upload content stored in the database (121) of the server (120) and train an artificial intelligence model (122) based on a large-scale language model based on the uploaded content. The administrator electronic device (110) may be configured to manage information provided through the website (130) or approve membership requests received through the website (130). The administrator of the system (100) may manage the overall operation of the system (100) through the administrator electronic device (110).

[0037] According to one embodiment, the server (120) may include a database (121), an artificial intelligence model (122), a wireless communication circuit (123), and a processor (124).

[0038] According to one embodiment, the database (121) may include content and member information. In the present disclosure, the content may include text data as a data source used to train an artificial intelligence model (122) based on a large language model. The artificial intelligence model (122) based on a large language model (LLM) is an artificial intelligence model trained with text data and may be used to generate a response message for a question text. According to one embodiment, the processor (124) may be configured to train the artificial intelligence model (122) based on a large language model based on content included in the database (121). In the case of a member registered on the website (130), the member information may be referenced as information entered upon member registration on the website (130).

[0039] A large-scale language model is a language model comprised of an artificial neural network pre-trained using text data. Compared to conventional general language models, a large-scale language model can contain more than ten times more parameters. Large-scale language models can be used to perform various natural language processing tasks, such as text generation, language translation, question answering, and summarization, and can operate using deep learning techniques.

[0040] According to one embodiment, the manager electronic device (110) can upload content into the database (121) of the server (120). The content may be based on the type of service provided through the server (120). For example, if the service provided through the server (120) is a medical consultation service, the content may be composed of content in the medical field. For example, if the service provided through the server (120) is a legal consultation service, the content may be composed of content in the legal field. The content is uploaded into the database (121) of the server (120) through the manager electronic device (110), and the type and nature of the consultation service provided through the server (120) may be determined based on the content registered through the manager electronic device (110).

[0041] Training of artificial intelligence based on a large-scale language model can be referred to as an artificial intelligence model (122) understanding human language and generating response messages based on a large amount of text data. A large amount of content uploaded via an administrator electronic device (110) can be uploaded to a database (121). As described above, the content can be based on the type and nature of the counseling service to be provided through the system (100) according to one embodiment. The uploaded content can be organized through a preprocessing process, such as removing unnecessary words or sentences and standardizing the format. A large-scale language model to be used for training is selected (e.g., GPT-3, Jurassic-1 Jumbo, Megatron-Turing NLG, etc.), and the hyperparameters of the large-scale language model can be set. The hyperparameters can affect the learning speed and accuracy of the model, etc.

[0042] In one embodiment, the processor (124) may train an artificial intelligence model (122) based on a large-scale language model based on the uploaded content. The artificial intelligence model (122) based on the large-scale language model may learn text contained within the content and, based on the text, generate new text or learn how to answer questions.

[0043] In one embodiment, the processor (124) can evaluate the performance of an artificial intelligence model (122) based on a trained large-scale language model. For example, the performance of the artificial intelligence model (122) can be evaluated using automated evaluation metrics, and based on the evaluation results, the performance of the artificial intelligence model (122) can be improved by adjusting hyperparameters or adding content.

[0044] According to one embodiment, the wireless communication circuit (123) may be configured to provide wireless communication with an external electronic device (200). For example, the wireless communication circuit (123) may preprocess information processed by the processor (124), up-convert information to a frequency to be transmitted to the external electronic device (200), and transmit the information to the external electronic device (200). For example, the wireless communication circuit (123) may down-convert information transmitted from the external electronic device (200) to a frequency that can be read by the processor (124), and provide the information to the processor (124). According to one embodiment, the external electronic device (200) may connect to the server (120) through the wireless communication circuit (123). A question text transmitted from the external electronic device (200) may be provided to the processor (124) through the wireless communication circuit (123). The processor (124) can use the artificial intelligence model (122) to generate a response message for the question text and transmit the response message to an external electronic device (200) through a wireless communication circuit (123).

[0045] According to one embodiment, the server (120) may be configured to provide a virtual space accessible to an external electronic device (200) via a wireless communication circuit (123). For example, the virtual space may be referred to as a virtual space for providing a consultation service via a consultation chatbot. For example, the virtual space may be displayed on the display of an external electronic device (200) accessing the virtual space, and consultation content provided by the consultation chatbot may be displayed through the virtual space.

[0046] In one embodiment, the website (130) may be used to receive a membership request from an external electronic device (200). For example, the system (100) according to one embodiment may provide a certain amount or number of consultation services through a consultation chatbot. After a certain amount or number of consultation services have been provided, the user of the external electronic device (200) may be required to register as a member in order to use additional consultation services. The external electronic device (200) may input information for membership registration through the website (130), and the administrator electronic device (110) may be configured to approve the membership registration based on the information entered through the website (130). If the consultation chatbot is provided indefinitely, only free consultation services that are not actually monetized will be continuously provided, making it difficult to generate actual revenue. Therefore, the system (100) according to one embodiment may provide consultation services only a certain number of times or a certain amount through the consultation chatbot, and provide additional consultation services to an external electronic device (200) that has registered after requesting membership.

[0047] Below, the operation of a system (100) providing a consulting chatbot based on artificial intelligence according to one embodiment is described.

[0048]

[0049] FIG. 2 is a flowchart illustrating the operation of a system (100) providing an artificial intelligence-based consultation chatbot according to one embodiment. FIG. 3 illustrates an example of consultation content provided to an external electronic device (200) through a system (100) providing an artificial intelligence-based consultation chatbot according to one embodiment.

[0050] The operations illustrated in FIG. 2 may be referred to as operations performed by the processor (124) of the server (120).

[0051] Referring to FIG. 2, in operation 201, the processor (124) may be configured to receive a question text (e.g., question text (301) of FIG. 3) from an external electronic device (200) connected to the server (120).

[0052] According to one embodiment, an external electronic device (200) may connect to a server (120) to utilize the system (100) providing a consultation chatbot. The server (120) may be configured to provide a virtual space to which the external electronic device (200) may connect via a wireless communication circuit (123). For example, an application capable of connecting to the virtual space of the server (120) may be installed on the external electronic device (200). The external electronic device (200) may connect to the virtual space of the server (120) by executing the application. FIG. 3 illustrates a screen displayed on an external electronic device (200) connected to the virtual space of the server (120). An avatar (310) may be displayed within the virtual space. The server (120) may provide the avatar (310) displayed within the virtual space.

[0053] According to one embodiment, the processor (124) may generate information for generating a virtual space and an avatar (310). The processor (124) may generate the avatar (310) and the virtual space. The processor (124) may be configured to transmit, to the external electronic device (200) via the wireless communication circuit (123), first information for rendering the virtual space, second information for rendering the avatar (310), and third information for rendering the response message described below. The first information, the second information, and the third information will be described below with reference to FIG. 6.

[0054] According to one embodiment, the processor (124) may be configured to generate an avatar (310) and a virtual space based on content included in the database (121). The processor (124) generating the avatar (310) and the virtual space based on the content may be referred to as generating the avatar (310) by reflecting the characteristics indicated by the content. For example, the processor (124) may analyze content (e.g., text data) included in the database (121) and generate the avatar (310) and the virtual space by reflecting the characteristics indicated by the content. For example, if the database (121) includes content in the medical field (e.g., text content such as disease, healthcare, hospital, medicine, etc.), the processor (124) may be configured to identify the characteristics of the content as being in the medical field and generate an avatar (310) of an image of a doctor or nurse including medical supplies such as a stethoscope and a gown. For example, if the database (121) includes content in the legal field (e.g., text content such as statutes, precedents, penalty provisions, legal requirements, and legal effects), the processor (124) may be configured to identify the characteristics of the content as being in the legal field and generate an avatar (310) of an image of a judge or lawyer including legal supplies such as a robe and gavel. In addition, the processor (124) may be configured to generate an avatar (310) by considering the characteristics of the fields of various counseling services provided through the counseling chatbot.

[0055] An external electronic device (200) connected to a server (120) can transmit a question text. For example, a user of the external electronic device (200) can input a question text regarding a matter to be consulted through the system (100) into the external electronic device (200). The external electronic device (200) can provide the question text to the server (120) via a wireless communication circuit (123). The processor (124) can receive the question text via the wireless communication circuit (123) and display the received question text within a virtual space. For example, referring to FIG. 3, the question text (301) displayed within the virtual space can be displayed via the external electronic device (200).

[0056] In operation 203, the processor (124) may be configured to extract a prompt from the question text to be input into a large-scale language model-based artificial intelligence model (122).

[0057] According to one embodiment, the processor (124) may be configured to extract a prompt from a question text received from an external electronic device (200). For example, the processor (124) may be configured to identify keywords within the question text and extract a prompt to be input into an artificial intelligence model (122) based on the identified keywords. Inputting the prompt into the artificial intelligence model (122) based on the large-scale language model may refer to the output information of the last neural network layer of the large-scale language model obtained when the prompt is input into an inference engine based on the large-scale language model. For example, as illustrated in FIG. 3, if the question text (301) is a text such as "Please introduce Director Hong Gil-dong," the processor (124) may extract a prompt from the question text (301). In the above example, the entire text "Please introduce Director Hong Gil-dong" may be extracted as the prompt.

[0058] In operation 205, the processor (124) may be configured to generate a content-based response message by inputting a prompt into a large-scale artificial intelligence model (122).

[0059] In one embodiment, the processor (124) may input the extracted prompt into a large-scale artificial intelligence model (122). The artificial intelligence model (122) may convert the prompt from an input language into an expression that can be processed within the artificial intelligence model (122). The expression that can be processed internally may be a data structure that the artificial intelligence model (122) uses to understand and process the prompt. The artificial intelligence model (122) may perform model inference based on the expression, thereby remembering information related to the prompt and generating a response to the prompt. During the inference process, the artificial intelligence model (122) may access a database (121) containing content and utilize pre-trained information.

[0060] As described above, since the artificial intelligence model (122) is trained based on the content included in the database (121), it can be used to generate a response message based on the content according to a prompt. The processor (124) can generate a response message using the artificial intelligence model (122). For example, referring to FIG. 3, the processor (124) can input “Please introduce Director Hong Gil-dong” into the artificial intelligence model (122) based on a large-scale language model, and generate a response message (302) related to information about Director Hong Gil-dong output from the artificial intelligence model (122).

[0061] In operation 207, the processor (124) may be configured to provide a response message to an external electronic device (200) through the virtual space of the server (120).

[0062] According to one embodiment, the processor (124) may be configured to provide a response message to an external electronic device (200) through a virtual space. The external electronic device (200) may obtain the response message provided through the virtual space. For example, referring to FIG. 3, the response message (302) may be displayed through a screen of the external electronic device (200) connected to the virtual space. The response message (302) may be positioned adjacent to an avatar (310). For example, a visual object, such as a speech bubble, may be associated with the avatar (310), and the response message (302) may be displayed within the visual object.

[0063] According to one embodiment, the system (100) provides a consultation chatbot that utilizes an artificial intelligence model (122) based on a large-scale language model trained on content contained in a database (121), thereby providing accurate and prompt information. Since the response message is provided using the artificial intelligence model (122), errors resulting from subjective judgment can be prevented, and problems arising from reliance on the counselor's personal feelings or subjectivity can be resolved. Furthermore, customer satisfaction can be improved by providing accurate and prompt consultation through an automated consultation chatbot without the emotional distress of the counselor caused by the user of an external electronic device (200).

[0064]

[0065] FIG. 4 is a flowchart showing an operation of providing a first guidance message (510) of a system (100) that provides an artificial intelligence-based consulting chatbot according to one embodiment.

[0066] For example, if a consulting chatbot using the system (100) is provided without limitation, it may be difficult to provide a service that leads to tangible results. To attract customers, a consulting service using a consulting chatbot can be provided in a certain quantity or frequency, and if the quantity or frequency of response messages provided exceeds a certain range, membership growth can be expected by inducing membership registration. The operations depicted in Figure 4 may be referenced as operations performed by the processor (124) of the server (120).

[0067] Referring to FIG. 4, in operation 401, the processor (124) may be configured to provide a response message. Operation 401 may be referred to as the operations of FIG. 2.

[0068] In operation 403, the processor (124) may be configured to count the amount or number of times a response message is provided.

[0069] According to one embodiment, the processor (124) may count the amount or number of times the response messages are provided using an artificial intelligence model based on a large-scale language model. For example, the processor (124) may count all response messages generated by the artificial intelligence model (122) and provided to the external electronic device (200), write a script for counting response messages, and analyze the script to count the amount or number of times the response messages are provided.

[0070] In operation 405, the processor (124) may be configured to identify whether the amount or number of times the response message is provided exceeds a threshold.

[0071] According to one embodiment, the processor (124) may compare the counted and accumulated provision amount or provision number of response messages with a threshold. For example, the threshold may include a first threshold corresponding to the accumulated provision amount of response messages and / or a second threshold corresponding to the accumulated provision number of response messages. The processor (124) may identify whether at least one of the provision amount or provision number of response messages exceeds the threshold. If the provision amount or provision number of response messages exceeds the threshold, operation 407 may be performed. If the provision amount or provision number of response messages does not exceed the threshold, operation 401 may be performed again.

[0072] In operation 407, the processor (124) may be configured to identify whether a member has registered based on a communication log of the external electronic device (200).

[0073] According to one embodiment, the processor (124) may collect and analyze communication logs of an external electronic device (200) connected to the server (120) based on identifying that the amount or number of response messages provided exceeds a threshold. The communication logs record communication activities between the external electronic device (200) and the server (120), and may indicate communication time of the external electronic device (200), a connected IP address, a port number, the amount of data transmitted, the protocol used, etc. The processor (124) may be configured to identify whether the external electronic device (200) has registered as a member of a website through the collection and analysis of the communication logs. Identification of whether the external electronic device (200) has registered as a member of the website may be identified by identifying whether the external electronic device (200) has logged in as a registered member of the website. For example, if an external electronic device (200) logs in through a website and then accesses the server (120) to receive a response message, the processor (124) can identify the external electronic device (200) as a member because a login record of the external electronic device (200) is left in the communication log. If the external electronic device (200) does not log in through the website but accesses the server (120) to receive a response message, the processor (124) can identify the external electronic device (200) as not a member because no login record of the external electronic device (200) is left in the communication log. However, the present invention is not limited thereto, and the processor (124) can identify whether the external electronic device (200) is a member through various methods. For example, the processor (124) can identify whether the external electronic device (200) is a member by using member information included in the database (121). If the external electronic device (200) corresponds to any one of the terminals registered in the member information, the processor (124) can also identify that the external electronic device (200) is a member. If the external electronic device (200) does not correspond to the terminals registered in the member information, the processor (124) may identify that it is a non-member.

[0074] In operation 409, the processor (124) may be configured to continuously provide a response message if it identifies a member.

[0075] According to one embodiment, the processor (124) may be configured to continuously provide response messages based on member identification at step 407. That is, even if the number or frequency of response messages provided exceeds a threshold, the processor (124) may continue to provide messages. Since members are likely to be key customers of the service, continuously providing services to members via a consultation chatbot may increase the likelihood of generating actual revenue. Therefore, because members have a high potential for revenue generation, the processor (124) may encourage use of the service by continuously providing response messages.

[0076] In operation 411, the processor (124) may be configured to stop providing a response message and provide a first guidance message (510) if it identifies that the user is a non-member.

[0077] According to one embodiment, the processor (124) may be configured to, based on identifying a non-member in operation 409, stop providing a response message and provide a first guidance message (510) for guiding membership registration to the external electronic device (200) through the virtual space of the server (120).

[0078] FIG. 5 illustrates an example of an external electronic device (200) accessing a website through a first guidance message (510) of a system (100) that provides an artificial intelligence-based consulting chatbot according to one embodiment.

[0079] The first screen (500a) of FIG. 5 illustrates an external electronic device (200) on which a first guidance message (510) is displayed. Referring to the first screen (500a), if the amount or number of times counseling messages are provided exceeds a threshold, the processor (124) may stop providing counseling messages and provide the external electronic device (200) with a first guidance message (510) for guiding membership registration. The first guidance message (510) may be displayed on the first screen (500a) of the external electronic device (200) through a virtual space. The first guidance message (510) may include text such as, "For additional counseling, please sign up for membership and receive counseling." According to one embodiment, the first guidance message (510) may include a link (511) that can access a website address that provides membership registration. When user input is provided for the above link (511), the external electronic device (200) can access the website through the link (511).

[0080] The second screen (500b) of FIG. 5 illustrates an external electronic device (200) connected to a website (520). For example, a user of the external electronic device (200) may connect the external electronic device (200) to the website (520) by providing a user input (e.g., a touch input) to a link (511) included in a first guidance message (510). When the external electronic device (200) is connected to the website (520), a website (520) providing membership registration may be displayed through the external electronic device (200). The website (520) may receive information for membership registration, and when a user enters the information and requests membership registration, the website may provide an approval request for membership registration to an administrator terminal. The administrator terminal may be configured to receive a request received through the website (520) and approve membership registration according to the request. After registering, users can provide a question text and receive a response message provided by an AI-based consultation chatbot. For example, after registration is complete, after step 413, the processor (124) can re-identify the user as a member and, accordingly, continue to provide response messages.

[0081] According to one embodiment, the processor (124) may be configured to connect the external electronic device (200) to the website (520) based on identifying that a first period of time has elapsed from the first time point at which the first guidance message (510) was provided. For example, if the amount or number of times response messages are provided exceeds a threshold and the first guidance message (510) is provided through the virtual space, the processor (124) may record the first time point at which the first guidance message (510) was provided. The processor (124) may induce membership registration by forcibly connecting the external electronic device (200) to the website (520) even without a user input for the link (511) of the first guidance message (510), based on identifying that a first period of time has elapsed from the first time point. For example, the processor (124) may be configured to use a location field in a response header from the server (120) to specify a URL of a website, and to cause the external electronic device (200) to automatically move to the URL of the website (520) when a first time elapses from a first point in time.

[0082] A system (100) providing an artificial intelligence-based counseling chatbot according to one embodiment can provide a content-based response message in response to a question text received from an external electronic device (200), and can recruit members naturally by forcing membership registration when the provided response messages exceed a certain amount or number of times. For example, if the system (100) is a system (100) providing counseling services in the medical field, the profitability of the revenue model can be improved as the number of members increases. Since revenue generation through the system (100) can increase with the number of members, such as by receiving advertisements from hospitals or pharmaceutical companies or providing promotions for directly operated hospitals, the above method can smoothly generate revenue by increasing the number of members.

[0083]

[0084] FIG. 6 illustrates an example of an external electronic device (200) displaying a virtual space and an avatar (310) of a system (100) providing a consulting chatbot based on artificial intelligence according to one embodiment.

[0085] According to one embodiment, the processor (124) may generate a virtual space (610) and an avatar (310) displayed within the virtual space (610). The processor (124) may generate the virtual space (610) and the avatar (310) based on content within the database (121). For example, if the content includes medical content (e.g., text data), the processor (124) may identify attributes of the medical field and, based on the attributes, generate a virtual space (610) of an examination room image and an avatar (310) of a doctor image.

[0086] According to one embodiment, the processor (124) may be configured to transmit information for displaying a virtual space (610) and an avatar (310) to the external electronic device (200) connected to the server (120). The processor (124) may be configured to transmit, to the external electronic device (200) via the wireless communication circuit (123), first information for rendering the virtual space (610), second information for rendering the avatar (310) displayed within the virtual space (610), and third information for rendering a response message to be displayed adjacent to the avatar (310) included within the virtual space (610). The external electronic device (200) may render the virtual space (610) using the first information. For example, the external electronic device (200) may render the virtual space (610) of an image of a treatment room using the first information. The external electronic device (200) can render an avatar (310) using the second information. For example, the external electronic device (200) can render an avatar (310) of a pseudo-image using the second information. The external electronic device (200) can render a response message adjacent to the avatar (310) displayed in the virtual space (610) using the third information. The third information can include a response message generated through an artificial intelligence model (122) based on a large-scale language model.

[0087] In one embodiment, the processor (124) may be configured to generate an avatar (310) based on question text. For example, if the question text is a question about a specific disease or a specific legal issue, the processor (124) may determine the intent of the question and provide an avatar (310) (a medical professional or lawyer) and a virtual space (610) (a medical office or a courtroom) corresponding to the question.

[0088] According to one embodiment, a user of an external electronic device (200) can have an experience similar to receiving counseling in a real-world environment when receiving counseling services through a counseling chatbot using the system (100). While counseling provided through the artificial intelligence model (122) may be stiff and dry, the user experience can be enhanced when counseling is provided through a content-based virtual space (610) and avatar (310).

[0089]

[0090] FIGS. 7A and 7B illustrate an example of an external electronic device (200) displaying a second guidance message of a system (100) providing an artificial intelligence-based consulting chatbot according to one embodiment.

[0091] Referring to FIG. 7a, a system (100) providing an artificial intelligence-based consulting chatbot according to one embodiment may provide a second guidance message (711) to an external electronic device (200) to inform that a response message (713) is being generated when the time for providing a response message (713) is delayed.

[0092] If it takes too long to generate the response message (713), it may be difficult for the user of the external electronic device (200) to determine whether the response message (713) is being generated or whether an error in the system (100) has occurred in the operation. To solve this problem, the processor (124) may be configured to provide a second guidance message (711) to the external electronic device (200) through the virtual space of the server (120) to inform that the response message (713) is being generated when a second time has elapsed from the second time point of receiving the question text (701). As illustrated in FIG. 7A, the second guidance message (711) may include text such as "The AI ​​counselor is searching for data, so please wait a moment." The user can identify that a response message (713) is being generated through a second guidance message (711) provided through an external electronic device (200) and wait for the provision of the response message (713). Once the response message (713) is generated, the response message (713) can be provided to the external electronic device (200) through a virtual space. As illustrated in FIG. 7A, the question text (701), the first guidance message (510), and the response message (713) can be sequentially displayed. As described above, an avatar (310) can be displayed on the external electronic device (200). The second time can be set to a time at which the provision of the response message (713) is determined to be delayed.

[0093] According to one embodiment, the processor (124) may be configured to identify the elapsed time from a second point in time and provide a first visual object (712) indicating the elapsed time to the external electronic device (200) through the virtual space of the server (120) together with a second guidance message (711). Referring to FIG. 7A, the first visual object (712) may include the elapsed time together with an icon indicating that an operation for generating a response message (713) is being performed. A user of the external electronic device (200) may recognize the elapsed time and confirm the waited time through the first visual object (712).

[0094] Referring to FIG. 7B, according to one embodiment, the processor (124) may be configured to calculate the time required to generate a response message (713) and provide a second visual object (714) indicating the required time to the external electronic device (200) through the virtual space of the server (120) together with a second guidance message (711). For example, the second visual object (714) may include the required time along with text such as “estimated waiting time.” A user of the external electronic device (200) may recognize the expected time required to wait to receive the response message through the second visual object (714) and confirm the waiting time. The second visual object (714) may include an image such as a scroll bar indicating the elapsed time compared to the required time. Through the scroll bar image, the required time, the elapsed time, and the remaining time may be intuitively recognized.

[0095] As described above, the system (100) providing an artificial intelligence-based counseling chatbot according to one embodiment can provide time information to users receiving counseling services through the counseling chatbot by providing a second guidance message (711), a first visual object (712), and a second visual object (714). Users of an external electronic device (200) can wait to receive a response message (713) through the second guidance message (711), the first visual object (712), and the second visual object (714) or use the system (100) at another time, thereby enhancing the user experience.

[0096]

[0097] FIG. 8 is a flowchart illustrating an operation of providing a response text of a system (100) providing an artificial intelligence-based consultation chatbot according to one embodiment. FIGS. 9a, 9b, and 9c illustrate examples of an external electronic device (200) displaying a response text of a system (100) providing an artificial intelligence-based consultation chatbot according to one embodiment.

[0098] The operations illustrated in FIG. 8 may be referred to as operations performed by the processor (124) of the server (120).

[0099] If the question text received from an external electronic device (200) is unrelated or has low relevance to the content, it may be difficult to generate a response message. In this case, even if a response message is generated and provided, there is a high probability that it will not actually be a response to the question text. Therefore, the system (100) providing an artificial intelligence-based consultation chatbot according to one embodiment can clarify the intent of the question through a counter-question (reverse question) when the similarity between the question text and the content is low.

[0100] Referring to FIG. 8, in operation 801, the processor (124) may be configured to identify similarity between question text and content.

[0101] According to one embodiment, the processor (124) may identify the similarity between the question text and the content to determine whether the question text received from the external electronic device (200) is related to the content. For example, the processor (124) may utilize a vector space model (VSM). The processor (124) may represent the content and the text as a set of words and generate a vector indicating the importance of each word. The processor (124) may calculate the similarity of the vectors to identify the similarity between the content and the text. In addition, various other methods may be utilized.

[0102] In operation 803, the processor (124) may be configured to identify whether the identified similarity is less than a first similarity.

[0103] According to one embodiment, the first similarity may be referred to as a similarity between the content and the question text that is sufficiently high to generate a response message for the question text. If the identified similarity is greater than or equal to the first similarity, operation 805 may be performed. If the identified similarity is less than the first similarity, operation 807 may be performed.

[0104] In operation 805, the processor (124) may be configured to provide a response message to the external electronic device (200) based on identifying a similarity greater than or equal to the first similarity.

[0105] According to one embodiment, since a similarity greater than or equal to the first similarity indicates that the question text is a question about content, the processor (124) may generate a response message and provide the response message to the external electronic device (200). Operation 805 may be referred to as the operations illustrated in FIG. 2.

[0106] In operation 807, the processor (124) may be configured to identify whether the identified similarity exceeds a second similarity that is lower than the first similarity.

[0107] In one embodiment, the second similarity may be referred to as a similarity that is sufficient to determine the intent of the question and generate a response message through the response text, although it is not sufficient to generate a response message. If the identified similarity is less than or equal to the second similarity, operation 809 may be performed. If the identified similarity exceeds the second similarity, operation 811 may be performed.

[0108] In operation 809, the processor (124) may be configured to provide a third guidance message to the external electronic device (200) through the virtual space of the server (120) to inform that a response message for the question text cannot be provided based on identifying a similarity lower than or equal to a second similarity that is lower than the first similarity. The third guidance message may be referred to as a message to inform that the question is incorrect and thus a response message cannot be generated. The user of the external electronic device (200) may recognize, through the third guidance message, that the question text is incorrect or that a response message for the question text cannot be obtained.

[0109] In operation 811, the processor (124) may be configured to identify one or more keywords within the content based on the question text, based on identifying a similarity less than a first similarity and greater than a second similarity.

[0110] According to one embodiment, the processor (124) may be configured to identify one or more keywords with a high degree of similarity to the question text based on the question text. For example, if the question text contains an error (typo), the processor (124) may search for and identify one or more keywords within the content that are most similar to the word containing the error.

[0111] In operation 813, the processor (124) may be configured to generate asking back text corresponding to the question text using one or more keywords.

[0112] In one embodiment, the processor (124) may be configured to generate a counter-text corresponding to the question text using one or more identified keywords. For example, if a keyword with the highest similarity to a word in error is found, the processor (124) may be configured to generate a counter-text asking whether the word in the question text is the keyword. For example, if the relevance between the content and the keyword in the question text is low, the counter-text may be generated by searching for the keyword with the highest similarity.

[0113] In operation 815, the processor (124) may be configured to provide the response text to an external electronic device (200) through the virtual space of the server (120).

[0114] According to one embodiment, the processor (124) may be configured to provide a response text to an external electronic device (200), and to correct the question text through confirmation by the external electronic device (200) and provide a response message. For example, if it is confirmed through the response text that a word with an error corresponds to the keyword, the processor (124) may replace the word with the keyword to re-identify the question text, and extract a prompt from the question text to generate a response message. Through this, the consultation can proceed smoothly, and if there is a lack of content for generating a response message for the response message, this can be fed back to the manager electronic device (110). For example, if a response message occurs due to a lack of a source registered as content, the processor (124) may notify the manager electronic device (110) of the lack of content for the corresponding response. The manager electronic device (110) may feed this back by additionally registering content.

[0115] Examples of the above operations are described through FIGS. 9a, 9b, and 9c.

[0116] Figure 9a illustrates a case where the similarity between content and question text (911) is less than the first similarity but greater than the second similarity. If the question text (911) is "Tell me about Gardasil," the processor (124) can identify the similarity between the question text (911) and the content. If the content is about a common cold or an infectious disease, the similarity between the question text (911) and the content may be less than the first similarity. The processor (124) can search for a keyword with the highest similarity to the question text (911) within the content and generate a response text (912). For example, the response text (912) may include text such as "Are you curious about Gardasil, a vaccine that prevents human papillomavirus (HPV)-related diseases?" In this case, if a positive response (913) is received, the processor (124) may generate a response message (914) from one or more keywords according to the response text (912).

[0117] Unlike the example illustrated in FIG. 9A, the same operation can be performed even if there is an error in the question text (911). For example, if the content includes content related to obstetrics and gynecology, but there is a typo in the question text (911) (e.g., the question text (911) such as "Tell me about gaddasil"), the processor (124) can search for "gaddasil" in the content using the keyword with the highest similarity to the typo "gaddasil." The processor (124) can generate a response text (912) based on "gaddasil," and accordingly, a response message (914) can be generated.

[0118] Figure 9b illustrates a case where the similarity between content and question text is less than the first similarity but greater than the second similarity. If the question text includes text such as "Tell me about seasonal health management," the processor (124) may search for one or more keywords with the highest similarity to "seasonal health management" within the content. If the content does not include "seasonal health management," but does include content about respiratory infectious diseases, the processor (124) may search for respiratory infectious diseases as a keyword to generate and provide a response text (922). If a negative response (923) is received, the processor (124) may search for allergic diseases as a keyword with the next highest similarity to generate and provide a response text (924). If a positive response (925) is received, the processor (124) may be configured to generate and provide a response message from the content about allergic diseases using the artificial intelligence model (122).

[0119] Figure 9c illustrates a case where the similarity between content and question text is less than or equal to the second similarity. If the content included in the database (121) is medical content and the question text includes text such as "Tell me about BTS," the processor (124) can identify that the similarity between the content and the question text is less than or equal to the second similarity. In this case, the processor (124) can be configured to provide a third guidance message to the external electronic device (200) through the virtual space of the server (120) to inform that a response message for the question text cannot be provided. The third guidance message can be referred to as a message informing that the question is incorrect and thus a response message cannot be generated. The user of the external electronic device (200) can recognize, through the third guidance message, that the question text is incorrect or that a response message for the question text cannot be obtained.

[0120] As described above, the system (100) providing an AI-based consultation chatbot according to one embodiment can accurately identify a question text using a response text or indicate that the question text is unrelated to the content. This allows for an accurate response to the question text.

[0121]

[0122] FIG. 10 illustrates an external electronic device (200) that uses a department change service of a system (100) that provides an artificial intelligence-based consulting chatbot according to one embodiment.

[0123] According to one embodiment, when a consultation with a specialized department is required, the processor (124) may provide information guiding the main contents of the consultation for the department, and when the department is changed during the consultation, the consultation may be processed to proceed in continuation of the existing consultation. For example, referring to the first screen (1000a) of FIG. 10, the external electronic device (200) may perform specialized consultation for a specific field (e.g., a women's plastic surgery clinic). Then, referring to the second screen (1000b) of FIG. 10, the external electronic device (200) may change the field and request specialized consultation for the field of childbirth / delivery. Referring to the third screen (1000c) of FIG. 10, when the field is changed from a women's plastic surgery clinic to the field of childbirth / delivery, the previously performed response message may be displayed as is, thereby providing continuity of the consultation.

[0124]

[0125] FIG. 11 illustrates an example in which a history of providing a response message is registered through an administrator electronic device (110) of a system (100) that provides a consulting chatbot based on artificial intelligence according to one embodiment.

[0126] According to one embodiment, the processor (124) may be configured to store history data corresponding to information related to the provision history of response messages in the database (121). Referring to FIG. 11, a server representing history data (1100) is illustrated. The administrator electronic device (110) may manage the provision history of response messages stored in the database (121) and process the registered content to be automatically registered as learning material for the artificial intelligence model (122). In addition, the administrator electronic device (110) may provide management for members by managing the history of response messages provided through consultations as data. For example, the administrator electronic device (110) may check and view the consultation history for each member in a virtual space. This enables continuous management of members, and by training the artificial intelligence model (122) based on a large-scale language model using the response messages for consultations, more specialized and detailed consultations may be possible.

[0127]

[0128] The various embodiments of this document and the terminology used therein are not intended to limit the technical features described in this document to specific embodiments, but should be understood to include various modifications, equivalents, or substitutes of the embodiments. In connection with the description of the drawings, similar reference numerals may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of the items, unless the context clearly indicates otherwise. In this document, each of the phrases "A or B", "at least one of A and B", "at least one of A or B", "A, B, or C", "at least one of A, B, and C", and "at least one of A, B, or C" can include any one of the items listed together in the corresponding phrase among those phrases, or all possible combinations thereof. Terms such as "first," "second," or "first" or "second" may be used merely to distinguish one component from another, and do not limit the components in any other respect (e.g., importance or order). When a component (e.g., a first component) is referred to as "coupled" or "connected" to another component (e.g., a second component), with or without the terms "functionally" or "communicatively," it means that the component can be connected to the other component directly (e.g., wired), wirelessly, or through a third component.

[0129] According to various embodiments, each of the components described above may include one or more entities, and some of the entities may be separated and placed in other components. According to various embodiments, one or more of the components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Alternatively or additionally, the plurality of components may be integrated into a single component. In such a case, the integrated component may perform one or more functions of each of the plurality of components identically or similarly to those performed by the corresponding component among the plurality of components prior to the integration. According to various embodiments, the operations performed by a module, program, or other component may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, omitted, or one or more other operations may be added.

Claims

1. In a system that provides a consulting chatbot based on artificial intelligence, server; and Including an administrator electronic device having management authority of the above server, The above server, A database containing content and member information uploaded by the above administrator electronic device, A wireless communication circuit configured to provide wireless communication with an external electronic device, and A processor configured to train an artificial intelligence model based on a large-scale language model based on the content included in the database, The above server, It is configured to provide a virtual space that can be accessed by an external electronic device through the wireless communication circuit to provide a consulting service, The above processor, Through the wireless communication circuit, a question text is received from the external electronic device connected to the server, From the above question text, extract a prompt to be input to the artificial intelligence model based on the large-scale language model, By inputting the above prompt into the artificial intelligence model based on the large-scale language model, a response message based on the content stored in the database is generated, configured to provide the above response message to the external electronic device through the virtual space of the above server, A system that provides a consulting chatbot based on artificial intelligence.

2. In paragraph 1, The above processor, Counting the amount or number of times the above response message is provided, Collecting and analyzing communication logs of the external electronic device based on identifying that at least one of the accumulated provision amount or the provision number exceeds a threshold, Based on the analysis results, identify whether the external electronic device is a member, If you are identified as a member, we will continue to provide you with the above response message, If it is identified that a non-member is present, the provision of the response message is stopped and a first guidance message for guiding membership registration is provided to the external electronic device through the virtual space of the server. A system that provides a consulting chatbot based on artificial intelligence.

3. In paragraph 2, The above first guidance message is, Including a link to a website address that provides membership registration; A system that provides a consulting chatbot based on artificial intelligence.

4. In paragraph 3, The above processor, Based on identifying that a first time has elapsed from the first time point at which the first guidance message is provided, the external electronic device is configured to connect to the website. A system that provides a consulting chatbot based on artificial intelligence.

5. In paragraph 1, The above processor, configured to provide, through the wireless communication circuit, to the external electronic device, first information for rendering the virtual space, second information for rendering an avatar displayed within the virtual space, and third information for rendering the response message to be displayed adjacent to the avatar included within the virtual space. A system that provides a consulting chatbot based on artificial intelligence.

6. In paragraph 5, The above processor, configured to generate the avatar based on the content included in the above database, A system that provides a consulting chatbot based on artificial intelligence.

7. In paragraph 5, The above processor, Based on the above question text, configured to provide the first information and the second information, A system that provides a consulting chatbot based on artificial intelligence.

8. In paragraph 1, The above processor, A second guidance message is configured to be provided to the external electronic device through the virtual space of the server to guide that the response message is being generated based on identifying that a second time has elapsed from the second time point of receiving the question text. A system that provides a consulting chatbot based on artificial intelligence.

9. In paragraph 8, The above processor, Identify the time elapsed from the second point in time above, configured to provide the first visual object indicating the elapsed time to the external electronic device through the virtual space of the server together with the second guidance message, A system that provides a consulting chatbot based on artificial intelligence.

10. In paragraph 9, The above processor, Calculate the time it will take to generate the above response message, A second visual object indicating the time required is configured to be provided to the external electronic device through the virtual space of the server together with the second guidance message. A system that provides a consulting chatbot based on artificial intelligence.

11. In paragraph 1, The above manager electronic device, Storing history data corresponding to information related to the provision history of the above response message in the above database, A system that provides a consulting chatbot based on artificial intelligence.

12. In paragraph 1, The above processor, Identify the similarity between the above question text and the above content, Providing the response message to the external electronic device based on identifying the similarity greater than or equal to the first similarity, Based on identifying the similarity less than the first similarity, identifying one or more keywords within the content based on the question text, Using one or more of the keywords, generate an asking back text corresponding to the question text, configured to provide the above-mentioned text to the external electronic device through the virtual space of the above-mentioned server, A system that provides a consulting chatbot based on artificial intelligence.

13. In paragraph 12, The above processor, Based on identifying the similarity that is lower than the second similarity that is lower than the first similarity, a third guidance message is provided to the external electronic device through the virtual space of the server to guide that the response message for the question text cannot be provided. A system that provides a consulting chatbot based on artificial intelligence.

Images