Method and system for providing digital human providing customized conversation

Abstract

To solve the problem described above, the method for providing a digital human that provides a user-customized conversation according to the present invention comprises the steps of: receiving a search term for information retrieval from an electronic device; inferring the user's intention to input the search term based on a user intention inference model learned by an artificial intelligence algorithm; identifying associated information corresponding to the intention to input the search term among the search result information of the search term using the inference result; and providing a digital human that performs a conversation based on the associated information on the electronic device. The associated information may be different information based on the fact that the intention to input the search term is inferred differently even if the search term is the same.

Description

Technology Field

[0001] The present invention relates to a method and system for providing a digital human that provides conversational information matching the user's search term input intent based on learned data, simply by the user presenting a search term. Background Technology

[0002] With the advancement of technology, the widespread adoption of electronic devices (e.g., smartphones, tablet PCs, automated devices, etc.) has become commonplace, and consequently, dependence on electronic devices is gradually increasing in many aspects of daily life.

[0003] In particular, with the advancement of display technology, users can view not only flat images but also three-dimensional images and three-dimensional spaces through various means. Accordingly, virtual reality and augmented reality services are being provided to offer users a three-dimensional visual user experience. As such, experiencing three-dimensional space has become a new trend in user experience, allowing users to simulate in virtual spaces, receive visual and other sensory experiences, and interact with virtual images.

[0004] Furthermore, the metaverse is a representative service that encompasses all of the above concepts and provides a mixed reality environment. The term metaverse is a compound word formed from 'meta,' meaning fabrication or abstraction, and 'universe,' meaning the real world, signifying the connection between three-dimensional virtual space and real space. As a concept more advanced than the existing term 'virtual reality environment,' the metaverse provides the aforementioned mixed reality environment in which virtual worlds such as the web and the internet are absorbed into the real world.

[0005] The metaverse allows for infinite expansion based on user participation, and enables the free selection of characters, environments, and spaces. For example, users can create terrain and place desired natural elements and facilities within it.

[0006] Meanwhile, a digital human is a method of graphic representation of a user that other users in a virtual space can see and interact with, and users in the real space can use digital humans to interact with the digital humans of other users in the virtual space, or to perform various economic, cultural, and educational activities in the virtual space.

[0007] In Korean Public Patent No. 10-2022-0112099 (Method and System for Creating a Realistic Digital Human Tutor), a technology is disclosed for giving a lecture using a digital human created using a professor's face image.

[0008] Recently, there has been an increasing need for digital humans capable of engaging in personalized conversations with users. The problem to be solved

[0009] The present invention is intended to provide a method and system for providing a digital human capable of having a customized conversation with a user.

[0010] In particular, the present invention is intended to provide a method and system for providing a digital human capable of inferring the input intent of a search term entered by a user and performing a customized conversation with the user. means of solving the problem

[0011] To solve the problem described above, the method for providing a digital human that provides a user-customized conversation according to the present invention comprises the steps of: receiving a search term for information retrieval from an electronic device; inferring the user's intention to input the search term based on a user intention inference model learned by an artificial intelligence algorithm; identifying associated information corresponding to the intention to input the search term among the search result information of the search term using the inference result; and providing a digital human that performs a conversation based on the associated information on the electronic device. The associated information may be different information based on the fact that the intention to input the search term is inferred differently even if the search term is the same.

[0012] Furthermore, the user intent inference model performs machine learning on the search term and context information related to the search term to infer the user's intent to input the search term when the user inputs the specific search term in a specific situation, and the context information may include at least one of date information, place information, weather information, season information, user schedule information, conversation information, and user history information.

[0013] Furthermore, the conversation information includes conversation content centered on the search term between the digital human and the user, and the intention to input the search term can be inferred based on the conversation topic extracted from the conversation information.

[0014] Furthermore, the user history information includes information about the content selected by the user among the search result information for the specific search term, and the search term input intent can be inferred based on the content of the content selected by the user.

[0015] Furthermore, the above-mentioned related information includes first related information and second related information corresponding to the search term input intention, the first related information is provided in a conversational format by the digital human, and the second related information can be provided in a keyword format on an area of ​​the screen where the digital human is output.

[0016] Furthermore, the method further includes a step of specifying at least one function based on the search term input intent, and the digital human can provide guide information corresponding to the specified function in a virtual space where the specified function is processed.

[0017] Furthermore, the user intent inference model performs machine learning on the user's behavior information taking place in the virtual space, and the behavior information may include user feedback information regarding the guide information.

[0018] Furthermore, the appearance of the digital human may be determined based on at least one of the person search term and person image included in the associated information.

[0019] Meanwhile, a digital human providing a user-customized conversation according to the present invention comprises a communication unit that receives a search term for information retrieval from an electronic device and a control unit that infers the user's intention to input the search term based on a user intention inference model learned by an artificial intelligence algorithm, wherein the control unit identifies associated information corresponding to the intention to input the search term among the search result information of the search term using the inference result and provides a digital human on the electronic device that performs a conversation based on the associated information, and the associated information may be different information based on the fact that the intention to input the search term is inferred differently even if the search term is the same.

[0020] Meanwhile, the program according to the present invention is executed by one or more processes in an electronic device and is a program that can be stored on a computer-readable medium, and includes instructions for performing the steps of: receiving a search term for information retrieval from an electronic device; inferring the user's intention to input the search term based on a user intention inference model learned by an artificial intelligence algorithm; identifying associated information corresponding to the intention to input the search term among the search result information of the search term using the inference result; and providing a digital human on the electronic device that performs a conversation based on the associated information. The associated information may be different information based on the fact that the intention to input the search term is inferred differently even if the search term is the same. Effects of the invention

[0021] The method and system for providing a digital human that offers a user-customized conversation according to the present invention infers the user's intent to input a search term based on a user intent inference model learned by an artificial intelligence algorithm, and uses the inference result to identify related information corresponding to the intent to input the search term among the search result information. Through this, the user can conveniently receive accurate related information corresponding to their keyword input intent simply by entering a keyword. Furthermore, the business operator can efficiently provide a customized information provision service according to the intent of each user.

[0022] Furthermore, the method and system for providing a digital human that provides user-customized conversation according to the present invention provides a digital human that performs a conversation based on related information on an electronic device, thereby allowing the user to receive related information more intuitively and enjoyably than receiving related information through text or images. Brief explanation of the drawing

[0023] FIG. 1 is a conceptual diagram illustrating a digital human provision system according to the present invention. FIG. 2 is a flowchart illustrating a method for providing a digital human according to the present invention. Figures 3a and 3b are conceptual diagrams illustrating a user intent inference model trained with an artificial intelligence algorithm. FIGS. 4, FIGS. 5, and FIGS. 6 are conceptual diagrams for explaining a method of providing related information corresponding to a user's search term input intent. FIGS. 7a, FIGS. 7b, FIGS. 8a, and FIGS. 8b are conceptual diagrams for explaining a method of providing a digital human in a virtual space according to the present invention. Figure 9 is a conceptual diagram for explaining the user's emotional information. Specific details for implementing the invention

[0024] 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.

[0025] 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.

[0026] 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.

[0027] A singular expression includes a plural expression unless the context clearly indicates otherwise.

[0028] 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.

[0029] The present invention relates to a digital human provision method and system capable of providing information in a conversational format that matches the user's search term input intent based on learned data, simply by the user presenting a search term.

[0030] Specifically, the digital human provision method system according to the present invention can provide customized conversations through a digital human by inferring the user's search term input intent based on an artificial intelligence model and big data when the user inputs a search term for information search. This provides relevant information that matches the user's search term input intent.

[0031] For example, as illustrated in FIG. 1, the present invention can provide warning information (2) about wildfires through a digital human (H) when a user inputs a search term (1) for “mountain”, and the search term is entered in March, when wildfires frequently occur.

[0032] That is, the digital human provision method system (100) according to the present invention can provide different information through the digital human depending on the situation in which the search term is entered, even if the search term is the same.

[0033] Meanwhile, in the present invention, the digital human (10) can be understood as a virtual character (or avatar) provided in a virtual space. The digital human (10) can be named as a character or avatar.

[0034] Hereinafter, a method and system for providing a digital human that provides a user-customized conversation will be described in detail together with the attached drawings. Fig. 1 is a conceptual diagram for explaining a digital human provision system according to the present invention. Fig. 2 is a flowchart for explaining a digital human provision method according to the present invention, Fig. 3 is a conceptual diagram for explaining a method for inferring a user's search term input intent, Figs. 4, 5, and 6 are conceptual diagrams for explaining a method for providing related information corresponding to a user's search term input intent, and Fig. 7 is a conceptual diagram for explaining a user intent inference model learned by an artificial intelligence algorithm.

[0035] As illustrated in FIG. 1, the digital human providing system (100) according to the present invention may be configured to include at least one of a communication unit (110), a storage unit (120), an information acquisition unit (130), and a control unit (140). At this time, the digital human providing 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 described in the description of the present specification.

[0036] Meanwhile, the digital human provision system (100) according to the present invention may operate in conjunction with an electronic device (10). Here, the electronic device (10) is an electronic device (10) for providing a digital human, and there is no restriction on the type thereof. For example, the electronic device (10) may include at least one of a smartphone, a mobile phone, a tablet PC, a head-mounted display, XR glasses, a computer, a laptop, a digital broadcasting terminal, a PDA (Personal Digital Assistants), a PMP (Portable Multimedia Player), and a Kiosk.

[0037] Meanwhile, the digital human provision system (100) according to the present invention may be implemented as an application or software. At this time, the application or software according to the digital human provision system (100) may be installed in an electronic device (10) and may perform a process for providing a service using the digital human described below.

[0038] In this case, the communication unit (110), storage unit (120), information acquisition unit (130), and control unit (140) of the digital human provision system (100) according to the present invention can perform a series of functions for a digital human provision service by utilizing the hardware configuration of the electronic device (10).

[0039] For example, the communication unit (110) can transmit and receive data with at least one external server (e.g., database (DB)) by using the configuration of a communication module (e.g., mobile communication module, short-range communication module, wireless internet module, location information module, broadcast reception module, etc.) provided in the communication unit of the electronic device (10).

[0040] Furthermore, the storage unit (120) may be configured to store data related to the digital human provision service according to the present invention by using a memory provided (or inserted) in the electronic device (10).

[0041] The information acquisition unit (130) can acquire various training data related to a user intent inference model learned by an artificial intelligence algorithm. For example, the information acquisition unit (130) can collect (acquire) situational information surrounding the input search term (age, gender, time zone, location, schedule, and SNS (Social Network Service) information of the user who entered the search term).

[0042] Meanwhile, the information acquisition unit (130) is part of the control unit (140) and can perform the work functions performed by the control unit (140). Accordingly, in the present invention, the information acquisition unit (130) and the control unit (140) are not distinguished separately, and are described as the control unit (140) without distinction.

[0043] Meanwhile, the control unit (140) can perform control of each component and overall control related to the present invention by utilizing the CPU provided in the electronic device (10). In order to perform the present invention, the control unit (140) may have control authority over at least one of the components of the electronic device (10).

[0044] Furthermore, the control unit (140) may utilize information received or acquired (collected) from the components of the electronic device (10) for the present invention. For example, the control unit (140) may generate a user inference model based on artificial intelligence learning that infers the user's search term input intent by performing learning on the learning data.

[0045] The control unit (140) can output a digital human that performs a user-customized conversation using an electronic device (10). The control unit (140) can control the digital human to perform a conversation with the user based on related information corresponding to a search term entered by the user.

[0046] In addition, the control unit (140) can perform various processes related to the digital human provision service based on the user's purpose of visiting the space according to the present invention by using other components provided in the electronic device (10) (e.g., proximity sensor, infrared sensor, touch screen, input unit, etc.).

[0047] Meanwhile, the digital human provision system (100) according to the present invention may exist not only in the form of a program or application, but also in various forms capable of performing the configuration and functions described below.

[0048] Accordingly, the digital human provision system (100) according to the present invention will be described below without imposing strict limitations thereon.

[0049] As illustrated in FIG. 1, the digital human providing system (100) according to the present invention receives a search term entered by a user (e.g., “mountain, 10”) from an electronic device (10) and can infer the user’s search term input intent based on a user intent inference model learned by an artificial intelligence algorithm. Furthermore, the digital human providing system (100) according to the present invention can control the digital human (H) to perform a conversation based on related information (20) corresponding to the user’s search term input intent.

[0050] The communication unit (110) may be configured to communicate with at least one of an electronic device (10) and an external server (not shown) via wired or wireless communication. The communication unit (110) may receive a search term entered into the electronic device (10) from the electronic device (10) using wired or wireless communication.

[0051] Here, the “search term” can be entered into the electronic device (10) in various ways. For example, the user can enter the search term into the electronic device (10) as text or voice. Additionally, the user can enter conversation information including the search term into the electronic device (10) (for example, “My area of ​​interest these days is ‘mountains’.”). In this case, the control unit (140) can extract the search term from the user’s conversation information.

[0052] The storage unit (120) may be configured to store data related to the digital human provision service according to the present invention.

[0053] In the storage unit (120), i) user information (user name, age, history information, interest information, etc.), ii) training data for inferring the user's search term input intent (e.g., conversation information, past search term input intent information, social information, etc.), and iii) digital human person information (e.g., digital human's age, gender, appearance, tone, intonation, conversation speed, conversation style, personality, etc.) may exist.

[0054] The control unit (140) can identify related information corresponding to the user's search history intent among the search results corresponding to the search term.

[0055] In this case, the control unit (140) can identify different information among the search results corresponding to the search term as related information based on the fact that even if the search term is the same, the user's search term input intent is inferred (or identified) differently.

[0056] For example, if a first user intent is inferred as the user's search term input intent, the control unit (140) can identify first information corresponding to the first user intent among the search term search results as associated information. On the other hand, if a second user intent different from the first user intent is inferred as the user's search term input intent, the control unit (140) can identify second information corresponding to the second user intent among the search term search results as associated information.

[0057] The control unit (140) can generate conversation information based on specific association information. And, the control unit (140) can conduct a user-customized conversation by outputting a digital human providing conversation information onto an electronic device (10).

[0058] Below, we will examine in more detail the method for providing a digital human that performs a user-customized conversation based on the user's search term input intent according to the present invention, together with a flowchart.

[0059] In the present invention, a process of receiving a search term for information retrieval from an electronic device may be carried out (S210, see FIG. 2). A communication unit (110) may receive a search term from an electronic device (10).

[0060] The control unit (140) can control the communication unit (110) to receive a search term entered on the electronic device (10).

[0061] In the present invention, the “search term” can be input into the electronic device (10) in various ways.

[0062] For example, a user can input a search term into the electronic device (10) as text or voice. Additionally, the user can input conversation information including the search term into the electronic device (10) (e.g., “My area of ​​interest these days is ‘mountains’.”). In this case, the control unit (140) can extract the search term from the user’s conversation information.

[0063] Meanwhile, in the present invention, a process of inferring the user's search term input intent based on a user intent inference model trained with an artificial intelligence algorithm may be performed (S220, see FIG. 2).

[0064] As illustrated in FIG. 3a, the control unit (140) can infer different search term input intentions (310, 320, 330) based on a user intention inference model (300) trained with an artificial intelligence algorithm, even if the search term (1) received from the electronic device (10) is the same.

[0065] Here, “a user intent inference model trained with an artificial intelligence algorithm” may refer to an artificial intelligence algorithm model that has performed training on training data (conversation information, information on past search intent, social information, etc.) to infer the user’s search term input intent.

[0066] The control unit (140) can learn what information the user wants to receive when the user has entered a search term, based on the search term and situation information related to the search term.

[0067] That is, the control unit (140) can learn what search term was entered to search for information in a specific situation.

[0068] Here, context information may include information about various situations related to the search term. For example, context information may include at least one of the following: date information on the date the search term was entered, information on the content of the conversation centered on the search term, gender information of the users who performed the conversation, age information, occupation information, interest information, history information (history of past search term entries, past search history, history of content selected by the user among search result information for a specific past search term, history of past conversations with digital humans, etc.), location information where the search term was entered (or where the user is located) (country information, region information, etc.), the user's address information, weather information (or season information) of the location where the search term was entered (or where the user is located), and the user's schedule information.

[0069] When the control unit (140) receives a search term from the electronic device (10), it can extract situational information related to the received search term. Then, the control unit (140) can infer the search term input intent of the user who entered the search term by performing learning on the extracted situational information and the search term.

[0070] The control unit (140) can infer different search term input intentions based on the fact that the extracted situation information is different even if the input search terms are the same.

[0071] The control unit (140) can infer a first user intent as the user's search term history intent when first situation information is extracted. On the other hand, the control unit (140) can infer a second user intent as the search term input intent when second situation information different from the first situation information is extracted.

[0072] Specifically, the control unit (140) can extract a conversation topic from conversation information containing conversation content centered on a search term between a digital human and a user. And, based on the extracted conversation topic, the control unit (140) can infer the user's intention to input a search term.

[0073] Furthermore, the control unit (140) can check information about content selected by the user among search result information for a specific search term in the past from user history information. And, the control unit (140) can infer the search term input intent based on the content of the content selected by the user.

[0074] For example, let us assume the case where the search term “restaurant” is entered. The control unit (140) can infer that the user’s search term input intent is “Korean food restaurant search” based on the fact that the user selected information (or content) corresponding to “Korean food” among the past restaurant search result information.

[0075] In this case, the control unit (140) can specify a specific category (e.g., “Korean food”) corresponding to the search result information most frequently selected by the user during a preset period among a plurality of categories (e.g., Korean food, Western food, etc.) corresponding to a specific search term, based on the user’s search term input intent.

[0076] Furthermore, the control unit (140) can infer the user's intention to input a search term based on issue information that occurs on the date the search term was entered.

[0077] For example, let us assume the case where the search term “mountain” is entered. Based on the fact that the date the search term was entered is March, when wildfires frequently occur, the control unit (140) can infer the user’s intention to enter the search term as “searching for wildfire-related information.” On the other hand, based on the fact that the date the search term was entered is October, the control unit (140) can infer the user’s intention to enter the search term as “searching for autumn hiking spots.”

[0078] Furthermore, the control unit (140) can infer the user's search term input intention based on the user's schedule included in the user schedule information.

[0079] For example, let us assume the case where the search term “National Tax Service” is entered. Based on the user’s schedule information, the control unit (140) can specify the search term input intent as “search for tax invoice information” in relation to a user who has a scheduled tax payment work schedule. On the other hand, based on the user’s schedule information, the control unit (140) can specify the search term input intent as “search for year-end tax settlement information” in relation to a user who has a scheduled year-end tax settlement schedule.

[0080] Meanwhile, as illustrated in FIG. 3b, the control unit (140) can collect various training data (301) to perform machine learning on a user intention inference model.

[0081] The training data may include at least one of various internal data (301a) existing within the system (100), user information (gender information, age information, etc., 301b), context information (keyword information, conversation information, user selection information, etc., 301c), and social information (or social data, 301d).

[0082] The control unit (140) can perform data learning (302) on the collected learning data.

[0083] Specifically, the control unit (140) can perform learning as learning data by extracting situation information related to the search term based on the reception of a specific search term, and the search term and the situation information.

[0084] Furthermore, when the control unit (140) infers the user's search term input intention based on the search term and situation information, it can perform learning using the inferred search term input intention as learning data.

[0085] Furthermore, the control unit (140) provides a plurality of search term input intentions inferred based on the search term input to the electronic device (10) and can receive a user selection for a specific search term input intention. The control unit (140) can perform learning using the user selection received from the electronic device (10) as learning data.

[0086] The control unit (140) can infer (304) the user's search term input intent based on machine learning on the training data.

[0087] Meanwhile, in the present invention, a process of identifying related information corresponding to the user's search term input intent among the search result information of a search term using the inference result may be performed (S230, see FIG. 2).

[0088] The control unit (140) can perform a search for a search term in various ways. The control unit (140) can perform a search for a search term from information already stored in the storage unit (120). In addition, the control unit (140) can perform a search for a search term through communication with an external server (e.g., a search server).

[0089] As illustrated in FIG. 4, the control unit (140) can identify related information (410, 420) corresponding to the user's search term input intention among the search result information (400) based on the search result for the search term.

[0090] Meanwhile, the control unit (140) can identify different search information among the search result results (400) as related information even if the search term is the same (e.g., “mountain, 1”), if the user’s search term input intention is different.

[0091] When the first user intent (310) is inferred from the user's search term input intent, the control unit (140) can identify the first search information corresponding to the first user intent (310) among the search term search results (400) as the associated information corresponding to the first user intent (e.g., “Uljin Samcheok wildfire information”, 410).

[0092] On the other hand, when the second user intent (320) is inferred from the user's search term input intent, the control unit (140) can identify the second search information corresponding to the second user intent (320) among the search term search results as related information (e.g., “October autumn hiking spot recommendation information”, 420).

[0093] In the present invention, for convenience of explanation, the associated information corresponding to the first user intention (310) may be named the first associated information, and the associated information corresponding to the second user intention (320) may be named the second associated information.

[0094] Meanwhile, in the present invention, a process of providing a digital human that performs a conversation based on related information on an electronic device may be carried out (S230, see FIG. 2).

[0095] The control unit (140) can generate conversation information based on specific association information. And, the control unit (140) can provide a digital human conducting a conversation on the electronic device (10) based on the conversation information.

[0096] Meanwhile, the control unit (140) can generate different conversation information even if the same search term (e.g., “mountain, 1”) is used, depending on whether the user’s search term input intent is different.

[0097] As illustrated in FIG. 5(a), the control unit (140) can generate first conversational information based on first related information (410) ("We must pay special attention to wildfire prevention and control from now on.", 510) when a first user intent (310) is inferred from the user's search term input intent.

[0098] On the other hand, as illustrated in FIG. 5(b), the control unit (140) can generate second conversational information based on second related information (420) when a second user intent (320) is inferred from the user's search term input intent (520).

[0099] That is, the control unit (140) can control the digital human to conduct a customized conversation according to the user's intention by generating conversation information based on the user's search term input intention.

[0100] Meanwhile, the control unit (140) can provide first related information and second related information corresponding to the user's search term input intention on the electronic device (10).

[0101] The control unit (140) provides the first associated information in a conversational format by a digital human, and

[0102] The digital providing a user-customized conversation is characterized in that the second associated information is provided in keyword format on a specific area of ​​the screen where the digital human is output.

[0103] As illustrated in FIG. 6(a), the control unit (140) can provide the first conversation information (610) generated based on the first association information to the digital human (H) in a conversational format. Then, while the digital human (H) is having a conversation with the user based on the conversation information (610) generated based on the first association information, the control unit (140) can output a keyword of the second association information (e.g., “recommendation of autumn hiking spots”, 620) on an area of ​​the screen of the electronic device (10) where the digital human (H) is outputting.

[0104] As illustrated in FIG. 6(a), the control unit (140) can control the digital human (H) to proceed with a conversation with the user according to the second conversation information (630) generated based on the second association information, based on the termination of the provision of the first conversation information (610) generated based on the first association information. That is, the control unit (140) can provide the second association information to the user through the second conversation information (630).

[0105] Meanwhile, the control unit (140) can identify lower-priority related information in various ways.

[0106] The control unit (140) can identify a first association and a second association corresponding to a specific search term input intention (e.g., a first user intention, see reference numeral 310 in FIG. 3) among the search result information for a search term.

[0107] The control unit (140) can calculate an association score (or reliability score) for each of the first association information and the second association information.

[0108] Here, the relevance score (or relevance score or confidence score) can be calculated based on the degree of association between the user's specific search term input intent and specific information.

[0109] The control unit (140) can calculate an association score for each of the first association information and the second association information based on situation information related to the search term.

[0110] For example, the control unit (140) can calculate a higher correlation score than the second correlation information for the first correlation information corresponding to the first content selected first among the first content and the second content included in the past specific search result information, based on the user's history information.

[0111] Furthermore, the control unit (140) can assign a rank to each of the first association information and the second association information based on the association score calculated for each of the first association information and the second association information.

[0112] For example, if the correlation score of the first correlation information is higher than the correlation score of the second correlation information, the control unit (140) may assign a first rank to the first correlation information and assign a second rank lower than the first rank to the second correlation information.

[0113] And, the control unit (140) can provide the first association information assigned a first rank to the user with priority over the second association information assigned a second rank.

[0114] Meanwhile, the control unit (140) can identify multiple search term input intentions based on a user intention inference model learned by an artificial intelligence algorithm and calculate an accuracy score (or confidence score) for the multiple search term input intentions.

[0115] Here, the accuracy score can be understood as being calculated based on the accuracy of the search term input intent.

[0116] The control unit (140) can calculate an accuracy score for each of the multiple search term input intentions based on situational information related to the search term. And, the control unit (140) can assign (assign) a rank to the search term input intentions based on the accuracy score for each of the multiple search term input intentions.

[0117] Specifically, when the control unit (140) infers a first user intent (e.g., “search for information related to forest fires”) and a second user intent (e.g., “search for autumn hiking spots”) with an intention to input a search term related to a search term (e.g., “mountain”), the control unit (140) can calculate an accuracy score for each of the first user intent and the second user intent based on situational information.

[0118] The control unit (140) may assign a first rank to the first user intention and assign a second rank lower than the first rank to the second user intention based on the fact that the accuracy score for the first user intention is higher than the accuracy score for each of the second user intentions.

[0119] The control unit (140) can specify information corresponding to a first user intent among the search result information for a search term as first associated information, and information corresponding to a second user intent as second associated information.

[0120] In this case, it can be understood that a first rank is assigned to the first associated information corresponding to the first user intent, and a second rank is assigned to the second associated information corresponding to the second user intent.

[0121] Meanwhile, the control unit (140) can set the person information of the digital human differently depending on the user's search term input intent.

[0122] In the personal information of the digital human described in the present invention, at least one of the digital human's appearance, age, gender, occupation, clothing, voice, intonation, tone, conversational style, conversation speed, conversation pattern, movement (or gesture), and facial expression may differ.

[0123] The personal information of such digital humans may exist in the storage unit (120).

[0124] The control unit (140) can generate different digital humans based on the user's search term input intent being different, even if the same search term is input, by referencing person information of the digital human existing in the storage unit (120), and provide them on the electronic device (10).

[0125] As illustrated in FIG. 5(a), when a first user intent (310) is inferred from a user's search term input intent, the control unit (140) can generate a digital human (H1) corresponding to person information corresponding to the first user intent (310) and provide it on the electronic device (10). For example, the digital human (H1) corresponding to the first user intent (310) may be in the form of an announcer or weather caster providing wildfire prevention warning information.

[0126] On the other hand, as illustrated in FIG. 5(b), when a second user intent (320) is inferred from the user's search term input intent, the control unit (140) can generate a digital human (H2) corresponding to the person information corresponding to the second user intent (320) and provide it on the electronic device (10). For example, the digital human (H2) corresponding to the second user intent (320) can be in the form of a "professional mountaineer" recommending autumn hiking spots.

[0127] Meanwhile, the control unit (140) can create a digital human by specifying the person information of the digital human based on the related information corresponding to the user's search term input intention.

[0128] The control unit (140) can determine the person information of the digital human based on the person search terms included in the associated information. For example, the control unit (140) can create a digital human based on the person information corresponding to the person search terms, based on the fact that the associated information includes person search terms that can identify a person, such as "announcer" or "professional mountaineer."

[0129] Furthermore, the control unit (140) can determine the appearance of the digital human as an appearance corresponding to a person image included in the associated information. For example, if the associated information includes a person image corresponding to a “firefighter,” the control unit (140) can create a digital human using the firefighter person image.

[0130] Meanwhile, the control unit (140) can provide guide information through a digital human that converses in a different conversational style based on the user's search term input intent.

[0131] The control unit (140) can set the conversation style of the digital human differently by setting the intonation, tone, speed, conversational style, dialogue, movements (or gestures), and facial expressions of the digital human.

[0132] In the present invention, it can be understood that there are multiple conversation methods of the digital human. These multiple conversation methods may differ in any one of the elements described above.

[0133] The control unit (140) can control the digital human to provide related information by any one of a plurality of conversation methods based on the user's search term input intent.

[0134] Furthermore, the control unit (140) can control the digital human to provide related information according to one of a plurality of conversation methods, taking into account the user's emotional information.

[0135] The control unit (140) can extract emotional information of the user from the user's face image. The control unit (140) can extract emotional information of the user from the user's face image based on an emotional analysis model learned by an artificial intelligence algorithm.

[0136] To this end, the control unit (140) may conduct a test to extract the user's emotional information. For example, the control unit (140) may output guidance information instructing the user's face to face the front of the camera equipped in the electronic device (10) for a certain period of time (e.g., “3 seconds”). As another example, to extract the user's emotional information, a test conversation may be conducted through a digital human.

[0137] The control unit (140) can extract (or detect) first emotion information corresponding to “positive” based on the user’s facial expression included in the user’s face image.

[0138] In this case, the control unit (140) can control the digital human to provide associated information in a first conversational style (e.g., “business conversational style”) corresponding to the first emotion information, based on the extraction (detection or detection) of the first emotion information (e.g., “positive”).

[0139] In contrast, the fisherman (140) can extract (detect) second emotion information corresponding to “negation” based on the user’s facial expression included in the user’s face image.

[0140] In this case, the control unit (140) can control the digital human to provide associated information in a second conversational method (e.g., “kind conversational method”) corresponding to the second emotion information, based on the extraction (detection or detection) of the second emotion information (e.g., “negative”).

[0141] Meanwhile, the control unit (140) can sense a user's face image from a camera equipped in the electronic device (10) while providing related information in a specific conversational manner through a digital human.

[0142] And, the control unit (140) can maintain or change the conversation method of the digital human based on the user's emotional information extracted from the user's face image.

[0143] The control unit (140) can extract first emotion information corresponding to “positive” based on the user’s facial expression included in the user’s face image while the digital human providing related information in a first conversational manner is provided through the electronic device (10).

[0144] In this case, the control unit (140) can maintain the first conversation method of the digital human.

[0145] In contrast, the control unit (140) can extract second emotion information corresponding to “negation” based on the user’s facial expression included in the user’s face image while the digital human providing related information in the first conversational manner is provided through the electronic device (10).

[0146] In this case, the control unit (140) can control the digital human, which was providing related information in a first conversational manner, to provide related information in a second conversational manner different from the first conversational manner, based on the extraction of second emotion (e.g., “negative”) information. That is, the control unit (140) can change the conversational manner of the digital human.

[0147] Meanwhile, the control unit (140) can extract age information or gender information of the user from the user's face image and control the digital human to provide related information according to one of a plurality of conversation methods based on at least one of the extracted age information and gender information.

[0148] Based on at least one of the age information and gender information extracted from the face image being different from each other, the control unit (140) can control the digital human to communicate in different conversation styles.

[0149] Based on the gender information (e.g., female, 女) extracted from the first face image and the gender information (e.g., male, 男) extracted from the second face image being different from each other, the control unit (140) can set different conversation styles for the digital human.

[0150] That is, the control unit (140) can provide a digital human that provides association information in different conversation styles to female users and male users respectively.

[0151] Furthermore, even if the gender information (e.g., male, 男) extracted from each of the second face image and the third face image is the same, based on the age information (e.g., 20s, 80s) being different from each other, the control unit (140) can set different conversation styles for the digital human.

[0152] For example, when the extracted age information corresponds to the second age information (e.g., 70s) rather than the first age information (e.g., 20s), the control unit (140) can control the digital human to provide association information in a conversation style with a slower conversation speed and a higher decibel (dB).

[0153] Furthermore, although not shown, when the age information extracted from the face image corresponds to below a preset age (e.g., “under 13 years old” or “under 19 years old”), the control unit (140) can control the digital human to provide a guidance message for reconfirming the user's age (e.g., “Your age confirmation is required”), and a guidance message for notifying the user of the inability to use a specific function (e.g., “For you, services that require a guardian such as loan consultations are restricted”).

[0154] Meanwhile, the control unit (140) can control the provision of digital human association information according to one of a plurality of conversation methods based on user preference information.

[0155] Here, user preference information may include information regarding the preferred method of providing related information. For example, user preference information may include various information such as a preference for being guided only on the key points of related information, a preference for being reminded of summarized past tasks, preferred speech (conversation) speed, tone, intonation, a preference for a friendly type, or a preference for a business-like type.

[0156] The control unit (140) can extract user tendency information from user information. And, the control unit (140) can control the digital human to provide related information according to a conversation method corresponding to the user tendency information.

[0157] The control unit (140) can cause the digital human to provide related information in a conversational manner corresponding to the first type of tendency if the user's tendency corresponds to the first type of tendency. And, if the second user's tendency corresponds to the second type of tendency, the control unit can control the digital human to provide related information in a conversational manner corresponding to the second type of tendency.

[0158] This tendency information can be set by the user or by the control unit (140).

[0159] The control unit (140) can analyze the user's tendencies based on the user's history information. That is, the control unit (140) can analyze the user's tendencies based on past conversations between the user and the digital human, and store the analyzed tendencies as tendency information in the storage unit (120).

[0160] Furthermore, the control unit (140) may conduct a test conversation between the digital human and the user to determine the user's tendency information. Based on the test conversation, the control unit (140) may extract (determine) the user's tendency analysis.

[0161] Meanwhile, the control unit (140) can control the digital human to conduct a test conversation with the user in order to extract the user's emotional information or tendency information.

[0162] The control unit (140) can conduct a test conversation by controlling the digital human in a conversational manner based on dialogue, intonation, tone, gestures, etc., which can determine the user's emotional information or tendency information.

[0163] The control unit (140) can determine the user's emotional information or tendency information based on the test conversation results. In this case, the control unit (140) can finally determine the user's emotional information or tendency information based on the test conversation results, the user's face image, and the tendency information included in the user information.

[0164] The control unit (140) can control the digital human to proceed with the conversation according to one of the multiple conversation methods, based on the user's emotional information or tendency information determined based on the test conversation results.

[0165] This test conversation can be performed at the stage before providing guide information through the digital human, or at the stage where guide information is provided through the digital human.

[0166] The control unit (140) can provide guide information through the digital human by conducting a test conversation between the digital human and the user before providing guide information through the digital human, and by using a conversation method corresponding to the user's emotional information and tendency information extracted based on the test conversation.

[0167] Furthermore, the control unit (140) may conduct a test conversation to determine the user's true emotional information when the user's emotional information extracted from the user's face image corresponds to specific emotional information (e.g., second emotional information corresponding to "negation") while the digital human is providing guide information in a specific conversational manner.

[0168] Specifically, the control unit (140) can control the digital human to attempt a test conversation with the user when the emotional information extracted from the user's face image corresponds to "negative (second emotional information)" while the digital human is providing guide information in the first conversational manner.

[0169] If the control unit (140) determines that the user’s emotional information corresponds to “negative (second emotional information)” as a result of the test conversation, it can change the conversation method of the digital human to a second conversation method different from the first conversation method and provide guide information.

[0170] On the other hand, if the user’s emotional information corresponds to “positive (first emotional information)” as a result of the test conversation, the control unit (140) can maintain the conversation method of the digital human as the first conversation method.

[0171] That is, the control unit (140) can analyze the user's emotional state through conversation with the user in addition to the emotional information extracted from the user's face image, and control the digital human to respond in a conversational manner appropriate to the user's emotions.

[0172] Meanwhile, in the present invention, a virtual space corresponding to a search term entered by a user can be provided on an electronic device (10).

[0173] The control unit (140) can infer the search term input intent for the search term entered by the user and specify a virtual space corresponding to the search term input intent.

[0174] The control unit (140) can create a virtual space corresponding to the spatial search term and provide it on the electronic device (10) when the search term is a spatial search term that refers to a specific space (e.g., “National Tax Service”).

[0175] Furthermore, if the search term is a function search term corresponding to a specific function (e.g., “tax payment”), the control unit (140) can create a virtual space corresponding to a space for processing said function (e.g., “National Tax Service”) and provide it on the electronic device (10).

[0176] That is, the control unit (140) can provide a virtual space corresponding to a real space capable of performing specific functions on the electronic device (10).

[0177] The control unit (140) can provide guide information on functions that can be processed in the virtual space to the electronic device (10) through a digital human existing in the virtual space.

[0178] Meanwhile, the control unit (140) can specify at least one function corresponding to the user's search term input intent. Additionally, the control unit (140) can create a virtual space corresponding to the space where the specified function is processed.

[0179] As illustrated in (a) of FIG. 7a, the control unit (140) can specify a first function (e.g., “pay property tax”, 710) and a second function (e.g., “pay health insurance premium”, 720) based on the user’s search term input intent.

[0180] As illustrated in Fig. 7a (b), the control unit (140) can create a virtual space (730) corresponding to a space (e.g., “bank”) where the first function (710) and the second function (720) are processed.

[0181] As illustrated in FIG. 7b(a), the control unit (140) can place a digital human associated with the virtual space (730) in the virtual space (750). For example, if the virtual space (713) corresponds to a “bank,” the control unit (140) can place a “banker” in the virtual space (730).

[0182] When a user (or user avatar or user graphic object, 740) enters the virtual space (730), the control unit (140) can provide first guide information corresponding to the first function (710) and second guide information corresponding to the second function (720) through a digital human (750) placed on the virtual space (730).

[0183] The control unit (140) can sequentially provide first guide information and second guide information based on the priority of the specified first function (710) and second function (720).

[0184] For example, let us assume that the first function (710) is assigned a first priority, and the second function (720) is assigned a second priority lower than the first priority.

[0185] As illustrated in FIG. 7b (a), the control unit (140) can provide first guide information (e.g., “Mr. Im * Hyun. I will guide you on paying property tax.”, 760) through a digital human (750) placed in a virtual space (730).

[0186] And, as illustrated in (b) of FIG. 7b, the control unit (140) can provide a second guide information (e.g., “I will continue to guide you on paying health insurance premiums.”, 770) through a digital human (750) placed in a virtual space (730) based on the termination of the provision of the first guide information.

[0187] That is, the control unit (140) can sequentially provide first guide information (760) corresponding to the first function and second guide information (770) corresponding to the second function through a digital human on a virtual space (730) corresponding to the space, based on the fact that the space where different first functions (710) and second functions (720) are processed is the same.

[0188] Meanwhile, the control unit (140) can specify at least one multiple function processed in different spaces based on the user's search term input intent. Additionally, the control unit (140) can create multiple virtual spaces corresponding to each of the different spaces.

[0189] As illustrated in (a) of FIG. 8a, the control unit (140) can specify a first function (e.g., “payment of property tax”, 811) processed in a first space (e.g., “bank”), a second function (e.g., “issuance of health certificate”, 812) processed in a second space (e.g., “health center”), and a third function (e.g., “issuance of driver’s license”, 813) processed in a third space (e.g., “road traffic authority”) based on the user’s search term input intent.

[0190] As illustrated in (b) of FIG. 8a, the control unit (140) can generate a plurality of virtual spaces (821 to 823) in which each of the first to third functions (811 to 813) is processed, based on the fact that the spaces in which each of the first to third functions (811 to 813) is processed are different.

[0191] For example, the control unit (140) can create a first virtual space (821) corresponding to a space where the first function (811) is processed, ex: “Bank”), a second virtual space (822) corresponding to a space where the second function (812) is processed, ex: “Health Center”), and a third virtual space (823) corresponding to a space where the third function (813) is processed, ex: “Road Traffic Authority”).

[0192] Furthermore, the control unit (140) can sequentially arrange the first virtual space to the third virtual space (821 to 823) based on the priority of the first to third functions (811 to 813).

[0193] For example, let us assume that a first priority is assigned to the first function (811), a second priority lower than the first priority is assigned to the second function (812), and a third priority lower than the second priority is assigned to the third function (813).

[0194] As illustrated in FIG. 8a (a), the control unit (140) can sequentially arrange a first virtual space (821) corresponding to a first priority, a second virtual space (822) corresponding to a second priority, and a third virtual space (823) corresponding to a third priority.

[0195] Meanwhile, when the control unit (140) receives a request to enter a plurality of spaces from the electronic device (10), it can control the user (or user avatar or user graphic object, 830) to enter the first virtual space (821) corresponding to the first priority among the plurality of virtual spaces.

[0196] As illustrated in FIG. 8b (a), the control unit (140) may provide a first virtual space (821) on the electronic device (10). The control unit (140) may place a first digital human (e.g., “banker”, 831) associated with the first virtual space (e.g., “bank”) on the first virtual space (821). The control unit (140) may provide first guide information (e.g., “Mr. Im * Hyun. I will guide you on paying property tax.”, 841) through the first digital human (831) existing on the first virtual space (821).

[0197] Furthermore, the control unit (140) can switch the virtual space provided on the electronic device (10) based on the termination of the provision of the first guide information (814).

[0198] As illustrated in (c) of FIG. 8b, the control unit (140) can control the first virtual space (821) on the electronic device (10) so that it disappears and a second virtual space (822) corresponding to a second priority is provided. The control unit (140) can provide second guide information (e.g., “Mr. / Ms. Im*Hyun. I will guide you on the issuance of a health certificate.”, 842) on the second virtual space (821) through a second digital human (e.g., “Doctor”, 832) associated with the second virtual space (e.g., “Health Center”).

[0199] Meanwhile, the control unit (140) may provide guidance information (e.g., “Banking business is complete. Would you like to move to the health center for the second business?”) that guides the virtual space transition, as shown in (b) of FIG. 8b, before switching the virtual screen provided on the electronic device (10) from the first virtual space (821) to the second virtual space (822).

[0200] That is, the control unit (140) can provide guidance information that guides virtual space switching on the electronic device based on the termination of the first guide information provision. And, the control unit (140) can receive a user selection related to virtual space switching from the electronic device (10).

[0201] The control unit (140) can control the switching of virtual space on the electronic device (10) if the received user selection is a selection corresponding to the switching of virtual space (e.g., “Yes”, 851).

[0202] In contrast, the control unit (140) can restrict the switching of virtual space on the electronic device (10) if the received user selection is a selection corresponding to not switching virtual space (e.g., “No”, 852).

[0203] Meanwhile, as illustrated in (a) of FIG. 8b, the control unit (140) can display a sub-map (860) overlaid on the display (12) of the electronic device (10) where the virtual space (821) is displayed.

[0204] Here, a “sub-map” can be understood as a map of the space where a specific function is processed.

[0205] The control unit (140) can display a position graphic object (870) on a region corresponding to the virtual space displayed on the electronic device (10) among the sub-maps (860).

[0206] For example, as illustrated in (a) of FIG. 8b, when a first virtual space (821) is displayed on the display (12) of the electronic device (10), a location display graphic object (870) can be displayed on the area (861) corresponding to the first virtual space among the sub-maps (860).

[0207] In another example, as illustrated in (c) of FIG. 8b, when a second virtual space (822) is displayed on the display (12) of the electronic device (10), a location display graphic object (870) can be displayed on the area (862) corresponding to the second virtual space among the sub-maps (860).

[0208] Meanwhile, the control unit (140) can collect all user actions performed by the user (or the user's graphic object or the user's avatar) in the virtual space as user action information.

[0209] Specifically, the control unit (140) can collect the following as user behavior information in the virtual space: information about a virtual space visited by the user, information about a function processed by the user in a specific virtual space, information about the time taken by the user to process a specific function in a specific virtual space, information about a conversation between the user and a digital human providing guide information about a specific function, information about the user's feedback to the digital human providing guide information in the virtual space, and information about the user's selection made to process the function in the virtual space.

[0210] This user behavior information may be included in the user's history information. And, the control unit (140) can learn from the user behavior information to infer the user's intention regarding the user's search term input.

[0211] Meanwhile, the control unit (140) can verify whether the search term search intent inferred based on the user's intention inference model learned by an artificial intelligence algorithm based on the user's facial expression information is the user's true search term search intent.

[0212] The control unit (140) can infer at least one search term input intention from a user's intention inference model learned by an artificial intelligence algorithm. For convenience of explanation, multiple search term input intentions are described by naming them as the first search term input intention, the second search term input intention, and the third search term input intention.

[0213] The control unit (140) can calculate an accuracy score for each of the first to third search term input intentions based on situational information related to the search term. And the control unit (140) can assign (assign or set) a rank to each of the first to third search term input intentions based on the accuracy score.

[0214] The control unit (140) may assign a first rank to the first search term input intention based on the fact that the accuracy score of the first search term input intention is the highest, and assign a second rank lower than the first rank to the second search term input intention based on the fact that the accuracy score of the second search term input intention is the second highest. Additionally, the control unit (140) may assign a third rank to the third search term input intention based on the fact that the accuracy score of the third search term input intention is the lowest.

[0215] The control unit (140) can sequentially output the first to third keywords corresponding to each of the first to third search term input intentions onto the electronic device (10).

[0216] The control unit (140) can receive a user's face image sensed (or captured) by a camera equipped in the electronic device (10) from the electronic device (10) while each of the first to third keywords is sequentially output on the electronic device (10).

[0217] As illustrated in FIG. 9, the control unit (140) can extract user emotion information from the user's face image. The control unit (140) can determine whether the extracted user emotion information corresponds to one of the first emotion information (910) corresponding to "positive" and the second emotion information (920) corresponding to "negative".

[0218] The control unit (140) can determine (or confirm) the user's actual search term input intention corresponding to the specific keyword when the extracted user's emotional information corresponds to the first emotional information matched to "positive" while the specific keyword is output on the electronic device (10).

[0219] And, the control unit (140) can provide related information corresponding to the confirmed search term input intention to the user in a conversational format through a digital human.

[0220] For example, while a first keyword (e.g., “wildfire”) corresponding to a first search term input intent (e.g., “search for wildfire-related information”) is being displayed on an electronic device (10), the control unit (140) can extract emotional information from the user’s face image. If the extracted emotional information corresponds to the first emotional information matched to “positive,” the control unit (140) can confirm the first search term input intent as the user’s actual search term input intent. Furthermore, the control unit (140) can provide the first related information corresponding to the first search term input intent to the user in a conversational format through a digital human.

[0221] On the other hand, while a first keyword corresponding to a first search term input intention is being output on the electronic device (10), if the user's emotional information extracted from the user's face image corresponds to the first emotional information matched to "negative," the control unit (140) can control the electronic device (10) so that the first keyword disappears and a second keyword (ex: "autumn hiking spot") corresponding to a second search term input intention (ex: "search for autumn hiking spot") is output.

[0222] Meanwhile, the control unit (140) can extract the user's emotional information while the digital human is performing a conversation based on related information and confirm the user's intention to input a search term.

[0223] Specifically, the control unit (140) can identify first association information corresponding to the first keyword input intention of the first priority. And, the control unit (140) can control the digital human to perform a conversation with the user based on the first association information.

[0224] The control unit (140) can receive a face image of a user sensed (or captured) by a camera equipped in the electronic device (10) from the electronic device (10) while the digital human is having a conversation with the user based on the first association information.

[0225] The control unit (140) can extract user's emotional information from the user's face image. The control unit (140) can determine whether the extracted user's emotional information corresponds to either the first emotional information corresponding to "positive" or the second emotional information corresponding to "negative".

[0226] Upon verification, the control unit (140) can extract (or detect) first emotion information corresponding to “positive” based on the user’s facial expression included in the user’s face image.

[0227] The control unit (140) can control the digital human to continue performing a conversation based on the first association information based on the extraction (detection or detection) of the first emotion (e.g., “positive”) information.

[0228] That is, the control unit (140) can determine that the inferred first keyword input intention corresponds to the user's actual keyword input intention based on the extraction of first emotion information from the user's face image.

[0229] In contrast, the control unit (140) can extract (detect) second emotion information corresponding to “negation” based on the user’s facial expression included in the user’s face image.

[0230] The control unit (140) can control the digital human to terminate the conversation based on the first association information and to perform the conversation based on the second association information based on the extraction (detection or detection) of the second emotion (e.g., “negative”) information.

[0231] In this case, the control unit (140) can confirm the user's intention to provide the second related information before performing a conversation based on the second related information. For example, the control unit (140) can control the digital human to speak guidance information (e.g., “Is the information currently being provided not the information you are looking for? Would you like me to guide you to other information?”) to confirm the user's intention.

[0232] And, the control unit (140) can control the digital human to perform a conversation based on the second related information based on receiving a request for the provision of second related information (e.g., “Please provide other information”) as feedback to the digital human’s utterance of guidance information.

[0233] Meanwhile, the method and system for providing a digital human that offers a user-customized conversation according to the present invention infers the user's intent to input the search term based on a user intent inference model learned by an artificial intelligence algorithm, and uses the inference result to identify related information corresponding to the intent to input the search term among the search result information of the search term. Through this, the user can conveniently receive accurate related information corresponding to their keyword input intent simply by entering a keyword. Furthermore, the business operator can efficiently provide a customized information provision service according to the intent of each user.

[0234] Furthermore, the method and system for providing a digital human that provides user-customized conversation according to the present invention provides a digital human that performs a conversation based on related information on an electronic device, thereby allowing the user to receive related information more intuitively and enjoyably than receiving related information through text or images.

[0235] 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.

[0236] 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.

[0237] 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.

[0238] Meanwhile, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention shall 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

Claim 1 The method comprises the steps of: receiving a search term for information retrieval from an electronic device; inferring the user's search term input intent based on a user intent inference model trained by an artificial intelligence algorithm; identifying associated information corresponding to the search term input intent among the search result information of the search term using the inference result; and providing a digital human on the electronic device that performs a conversation based on the associated information, wherein the associated information is different information based on the fact that the search term input intent is inferred differently even if the search term is the same, and the step of inferring the search term input intent includes inferring a plurality of search term input intents based on the user intent inference model, calculating an accuracy score for each of the plurality of search term input intents based on situational information related to the search term, and assigning a rank to each of the plurality of search term input intents based on the accuracy score, and the step of identifying the associated information includes the step of sequentially outputting a plurality of keywords corresponding to each of the plurality of search term input intents on the electronic device according to the assigned rank. A step of extracting emotion information of the user from a face image of the user sensed by a camera equipped in the electronic device while each of the above plurality of keywords is sequentially output on the electronic device;A method for providing a digital human that provides a user-customized conversation, characterized in that, when a specific keyword is output on the electronic device, the extracted emotional information corresponds to a first emotional information corresponding to a positive one, the search term input intention corresponding to the specific keyword is determined as the user's actual search term input intention, and the information corresponding to the determined actual search term input intention is determined as the associated information, wherein the associated information includes a first associated information and a second associated information corresponding to the determined actual search term input intention, and the step of providing the digital human further includes: a step of controlling the digital human to utter guidance information to confirm the user's intention when, while the digital human is performing a conversation with the user based on the first associated information, the emotional information extracted from the user's face image corresponds to a second emotional information corresponding to a negative one; and a step of controlling the digital human to terminate the conversation based on the first associated information and perform a conversation based on the second associated information based on receiving a request to provide the second associated information as feedback to the digital human's utterance of the guidance information. Claim 2 A method for providing a digital human that provides a user-customized conversation, characterized in that, in claim 1, the user intent inference model performs machine learning on the search term and context information related to the search term to infer the search term input intent of the user who entered the specific search term in a specific situation, and the context information includes at least one of date information, place information, weather information, season information, user schedule information, conversation information, and user history information. Claim 3 A method for providing a digital human that provides a user-customized conversation, wherein, in paragraph 2, the conversation information includes conversation content centered on the search term between the digital human and the user, and the intention to input the search term is inferred based on the conversation topic extracted from the conversation information. Claim 4 A method for providing a digital human that provides a user-customized conversation, characterized in that, in paragraph 2, the user history information includes information about the content selected by the user among the search result information for the specific search term, and the search term input intention is inferred based on the content of the content selected by the user. Claim 5 A method for providing a digital human that provides a user-customized conversation, characterized in that, in paragraph 4, the associated information includes a first associated information and a second associated information corresponding to the search term input intention, the first associated information is provided in a conversational format by the digital human, and the second associated information is provided in a keyword format on a screen area where the digital human is output. Claim 6 A method for providing a digital human that provides a user-customized conversation, characterized in that, in claim 4, it further includes the step of specifying at least one function based on the search term input intent, and the digital human provides guide information corresponding to the specified function in a virtual space where the specified function is processed. Claim 7 A method for providing a digital human that provides a user-customized conversation, characterized in that, in claim 6, the user intent inference model performs machine learning on the user's behavior information that takes place in the virtual space, and the behavior information includes the user's feedback information regarding the guide information. Claim 8 A method for providing a digital human that offers a user-customized conversation, characterized in that, in paragraph 4, the appearance of the digital human is determined based on at least one of a person search term and a person image included in the associated information. Claim 9 A communication unit that receives a search term for information retrieval from an electronic device; The invention includes a control unit that infers a user's search term input intent based on a user intent inference model trained by an artificial intelligence algorithm, wherein the control unit identifies associated information corresponding to the search term input intent among the search result information of the search term using the inference result, and provides a digital human that performs a conversation based on the associated information on the electronic device, wherein the associated information is different information based on the fact that the search term input intent is inferred differently even if the search term is the same, and the control unit infers a plurality of search term input intents based on the user intent inference model, calculates an accuracy score for each of the plurality of search term input intents based on situational information related to the search term, and assigns a rank to each of the plurality of search term input intents based on the accuracy score, and the control unit sequentially outputs a plurality of keywords corresponding to each of the plurality of search term input intents on the electronic device according to the assigned rank, and while each of the plurality of keywords is sequentially output on the electronic device, extracts the user's emotion information from the user's face image sensed by a camera equipped on the electronic device, and a specific keyword is on the electronic device When the extracted emotional information is displayed on the screen, if it corresponds to the first emotional information corresponding to a positive, the search term input intention corresponding to the specific keyword is confirmed as the user's actual search term input intention, and the information corresponding to the confirmed actual search term input intention is specified as the associated information, wherein the associated information includes the first associated information and the second associated information corresponding to the confirmed actual search term input intention, and the control unit, while the digital human is performing a conversation with the user based on the first associated information, if the emotional information extracted from the user's face image corresponds to the second emotional information corresponding to a negative,A digital human providing a user-customized conversation system characterized by controlling the digital human to utter guidance information to confirm the user's intent, and, based on receiving a request to provide the second associated information as feedback to the digital human's utterance of the guidance information, controlling the digital human to terminate the conversation based on the first associated information and perform the conversation based on the second associated information. Claim 10 A program that is executed by one or more processes in an electronic device and is storeable on a computer-readable medium, wherein the program comprises: a step of receiving a search term for information retrieval from an electronic device; a step of inferring a user's search term input intent based on a user intent inference model learned by an artificial intelligence algorithm; and a step of identifying associated information corresponding to the search term input intent among the search result information of the search term using the inference result. The method includes instructions for performing the step of providing a digital human that performs a conversation based on the associated information on the electronic device, wherein the associated information is different information based on the fact that the search term input intention is inferred differently even if the search term is the same, and the step of inferring the search term input intention includes inferring a plurality of search term input intentions based on the user intention inference model, calculating an accuracy score for each of the plurality of search term input intentions based on situational information related to the search term, and assigning a rank to each of the plurality of search term input intentions based on the accuracy score, and the step of specifying the associated information includes the step of sequentially outputting a plurality of keywords corresponding to each of the plurality of search term input intentions on the electronic device according to the assigned rank; and the step of extracting the user's emotion information from the user's face image sensed by a camera equipped in the electronic device while each of the plurality of keywords is sequentially output on the electronic device.A program stored on a computer-readable recording medium, characterized in that, when a specific keyword is output on the electronic device, the extracted emotional information corresponds to a first emotional information corresponding to a positive one, the search term input intention corresponding to the specific keyword is determined as the user's actual search term input intention, and the information corresponding to the determined actual search term input intention is determined as the associated information, wherein the associated information includes a first associated information and a second associated information corresponding to the determined actual search term input intention, and the step of providing the digital human further includes: a step of controlling the digital human to utter guidance information to confirm the user's intention when, while the digital human is performing a conversation with the user based on the first associated information, the emotional information extracted from the user's face image corresponds to a second emotional information corresponding to a negative one; and a step of controlling the digital human to terminate the conversation based on the first associated information and perform a conversation based on the second associated information based on receiving a request to provide the second associated information as feedback to the digital human's utterance of the guidance information.

Images