Electronic device for providing query / response service using persona and control method therefor

Abstract

Disclosed is an electronic device. The electronic device according to the present disclosure comprises a memory for storing instructions and at least one processor including processing circuitry, wherein the instructions, when executed individually or collectively by the at least one processor, cause the electronic device to: when a user input corresponding to a query related to content is received, acquire a persona corresponding to the query from among a plurality of personas corresponding to the content on the basis of a database; and provide a response corresponding to the query on the basis of the acquired persona, wherein the plurality of personas may be based on a knowledge graph corresponding to the content and the query / response database.

Description

Electronic device providing a question-and-answer service using personas and a method for controlling the same

[0001] The present disclosure relates to an electronic device that provides a question and answer service and a method for controlling the same.

[0002] With the expansion of OTT services, users can now watch video content such as movies and dramas in their preferred locations without having to visit movie theaters. Furthermore, thanks to the development of generative AI like LLM, users can obtain answers to their questions using natural language, without having to use computer languages ​​such as coding.

[0003] To obtain answers to questions arising while watching content, users had no choice but to use programs employing generative AI models separately from the OTT service or perform web searches. Consequently, there has been a growing need for a method to perform content-related Q&A within the OTT service by combining it with generative AI models to enhance user convenience.

[0004] An electronic device according to the present disclosure comprises at least one processor including a memory for storing instructions and processing circuitry, wherein when the instructions are executed individually or collectively by the at least one processor, the electronic device receives user input corresponding to a query related to content, obtains a persona corresponding to the query among a plurality of personas corresponding to the content based on a database, and provides a response corresponding to the query based on the obtained persona, wherein the plurality of personas may be based on a knowledge graph corresponding to the content and a query-response database.

[0005] When the above instructions are executed individually or collectively by the at least one processor, the electronic device can acquire the plurality of personas corresponding to the content through an artificial intelligence model based on a knowledge graph corresponding to the content, the database, and a first prompt.

[0006] The first prompt above may include a prompt that corresponds to a request for attribute prediction related to a user corresponding to the content for a plurality of queries of the database based on a knowledge graph corresponding to the content.

[0007] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may acquire evaluation information corresponding to an expected query corresponding to a persona for each of the plurality of personas, an expected response to the expected query, and feedback on the expected response, and update the plurality of personas through an artificial intelligence model updated based on the evaluation information.

[0008] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain the expected query through an artificial intelligence model based on the persona, a knowledge graph corresponding to the content, and a second prompt; obtain the expected response through the artificial intelligence model based on the persona, a knowledge graph corresponding to the content, the expected query, and a third prompt; and obtain the feedback through the second artificial intelligence model based on the persona, a knowledge graph corresponding to the content, the expected query, the expected response, and a fourth prompt. The second prompt may include a prompt corresponding to a user's expected query request corresponding to the persona based on a knowledge graph corresponding to the content; the third prompt may include a prompt corresponding to a user's expected response request regarding the expected query corresponding to the persona based on a knowledge graph corresponding to the content; and the fourth prompt may include a prompt corresponding to a user's feedback request regarding the expected response corresponding to the persona based on a knowledge graph corresponding to the content.

[0009] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain a first score through an evaluation model based on a prompt corresponding to a score request related to the quality of the expected query and expected response, and update the artificial intelligence model based on the first score.

[0010] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain a second score through an evaluation model based on a prompt corresponding to a score request related to the quality of the set including the expected query, expected response, and feedback, and the quality of the set, and update the artificial intelligence model based on the second score.

[0011] A knowledge graph corresponding to the above content may include, for each of the plurality of frames constituting the content, at least one node corresponding to the frame, at least one edge, and a text description of the frame. In the present disclosure, the node may correspond to an object included in the frame, and the edge may correspond to a relationship between the objects included in the frame.

[0012] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain the utilization rate of at least one of the nodes and edges corresponding to obtaining the response to the query, and update the knowledge graph based on the obtained utilization rate.

[0013] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain an average value of similarity for each of the plurality of personas based on similarities between the query and the plurality of queries corresponding to the plurality of personas stored in the database, and identify the persona having the largest value among the obtained average values ​​as the persona corresponding to the query.

[0014] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may identify the similarity between the query and a plurality of queries in the query-response database, and if there is a query among the plurality of queries with a similarity greater than or equal to a threshold, obtain a response corresponding to the query with a similarity greater than or equal to the threshold from the database and provide the response, and if there is no query among the plurality of queries with a similarity greater than or equal to the threshold, obtain a response to the query based on the obtained persona.

[0015] A control method for an electronic device according to the present disclosure may include the step of, upon receiving user input corresponding to a query related to content, obtaining a persona corresponding to the query among a plurality of personas corresponding to the content based on a database, and providing a response corresponding to the query based on the obtained persona.

[0016] Multiple personas may be based on a knowledge graph corresponding to the above content and the above question-and-answer database.

[0017] In addition, in a non-transient computer-readable recording medium that stores one or more instructions executed by a processor of an electronic device to perform an operation, the operation may include the steps of receiving user input corresponding to a query related to content, obtaining a persona corresponding to the query among a plurality of personas corresponding to the content based on a database, obtaining a response corresponding to the query based on the obtained persona, and providing the obtained response.

[0018] FIG. 1 is a drawing for explaining the operation of an electronic device according to at least one embodiment of the present disclosure.

[0019] FIG. 2 is a block diagram for explaining the configuration of an electronic device according to at least one embodiment of the present disclosure.

[0020] FIG. 3 is a block diagram illustrating the detailed configuration of an electronic device according to at least one embodiment of the present disclosure.

[0021] FIG. 4 is a flowchart illustrating a persona generation method according to at least one embodiment of the present disclosure.

[0022] FIG. 5 is a drawing for illustrating a knowledge graph according to at least one embodiment of the present disclosure.

[0023] FIGS. 6a to 6d are drawings for explaining the process of an electronic device according to at least one embodiment of the present disclosure generating a persona and updating a persona.

[0024] FIG. 7 is a flowchart illustrating a method for an electronic device according to at least one embodiment of the present disclosure to provide a response to a user.

[0025] FIGS. 8a, 8b, 9a and 9b are drawings for illustrating a user interface of a question-and-answer program according to at least one embodiment of the present disclosure.

[0026] FIG. 10 is a flowchart illustrating a method for identifying a persona according to at least one embodiment of the present disclosure.

[0027] FIG. 11 is a drawing for explaining a method of identifying a persona based on a database according to at least one embodiment of the present disclosure.

[0028] FIGS. 12 and FIGS. 13 are flowcharts for illustrating a method for optimizing a knowledge graph according to at least one embodiment of the present disclosure.

[0029] The various embodiments of the present disclosure and the terms used therein are not intended to limit the technical features described in the present disclosure to specific embodiments, and should be understood to include various modifications, equivalents, or substitutions of said embodiments.

[0030] In relation to the description of the drawings, similar reference numerals may be used for similar or related components.

[0031] The singular form of the noun corresponding to an item may include one or plural items, unless the relevant context clearly indicates otherwise.

[0032] In the present disclosure, each of the phrases such as “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” may include any one of the items listed together in the corresponding phrase, or all possible combinations thereof. For example, “A or B,” “at least one of A and B,” or “at least one of A or B” may refer to cases including (1) at least one A, (2) at least one B, or (3) both at least one A and at least one B.

[0033] Terms such as "first," "second," or "first" or "second" may be used simply to distinguish a component from another component and do not limit the components in other aspects (e.g., importance or order).

[0034] Where any (e.g., 1st) component is referred to as "coupled" or "connected" to another (e.g., 2nd) component, with or without the terms "functionally" or "communicationly," it means that the component may be connected to the other component directly (e.g., via a wire), wirelessly, or through a third component.

[0035] Terms such as "include" or "have" are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in this document, and do not preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0036] When it is said that a component is "connected," "combined," "supported," or "in contact" with another component, this includes not only cases where the components are directly connected, combined, supported, or in contact, but also cases where they are indirectly connected, combined, supported, or in contact through a third component.

[0037] When it is said that a component is located "on" another component, this includes not only cases where one component is in contact with the other, but also cases where another component exists between the two components.

[0038] The term "and / or" includes a combination of multiple related described components or any of the multiple related described components.

[0039] In some situations, the expression “device configured to do something” may mean that the device is “capable of doing something” in conjunction with other devices or components. For example, the phrase “processor configured (or set) to perform A, B, and C” may mean a dedicated processor for performing the said operations (e.g., an embedded processor), or a generic-purpose processor (e.g., a CPU or application processor) capable of performing said operations by executing one or more software programs stored in a memory device.

[0040] In the embodiments, a "module" or "part" performs at least one function or operation and may be implemented in hardware or software, or a combination of hardware and software. Additionally, a plurality of "modules" or a plurality of "parts" may be integrated into at least one module and implemented by at least one processor (not shown), except for a "module" or "part" that needs to be implemented in specific hardware.

[0041] Meanwhile, various elements and areas in the drawings are depicted schematically. Accordingly, the technical concept of the present invention is not limited by the relative sizes or spacing depicted in the attached drawings.

[0042] An embodiment of the present disclosure will be described in more detail below with reference to the attached drawings.

[0043] FIG. 1 is a drawing for explaining the operation of an electronic device according to at least one embodiment of the present disclosure.

[0044] Referring to FIG. 1, an electronic device (100) can generate a response corresponding to a query. The electronic device (100) may include, for example, a server device. The server device may include a device that provides services to multiple users through a network. The query may include a query corresponding to content. The content may include video content such as movies, dramas, and short-form content. For example, a user may watch content being played on a user terminal device (200) and input a query related to the content into the user terminal device (200). The user terminal device (200) may transmit the query input by the user to the electronic device (100).

[0045] For example, the electronic device (100) may receive a query (201) from a user, such as “How did the director’s directing style affect the storytelling?” The electronic device (100) may analyze the content being played based on the received query (201). Based on the analysis of the content, the electronic device (100) may transmit a control signal to the user terminal device (200) to display a response (202) corresponding to the query on the display (210).

[0046] When generating a response corresponding to a query, the electronic device (100) can predict the attributes of the user who wrote the query corresponding to the content and generate a response corresponding to the query based on the user's attributes.

[0047] For example, the electronic device (100) can identify that the user has the attribute of being a “fan of Director AA” based on a query received from the user. The electronic device (100) can generate and provide to the user a response that is predicted to be provided to the user who is a “fan of Director AA” in response to the received query.

[0048] By operating the aforementioned electronic device (100), the electronic device (100) can perform question-and-answer corresponding to content in the OTT service by combining an OTT service and a generative artificial intelligence model, thereby having the effect of enhancing user convenience.

[0049] FIG. 2 is a block diagram for explaining the configuration of an electronic device according to at least one embodiment of the present disclosure.

[0050] Referring to FIG. 2, the electronic device (100) may include a memory (110) and at least one processor (120) (hereinafter referred to as processor (120)).

[0051] According to various embodiments of the present disclosure, the memory (110) may store data necessary for the electronic device (100) to operate. Depending on the purpose of data storage, the memory (110) may be implemented as a memory embedded in the electronic device (100) (e.g., volatile memory (e.g., semi-permanent memory such as RAM (random access memory)), non-volatile memory (e.g., permanent memory such as ROM (read-only memory)), flash memory, hard drive or solid-state drive, etc.), or as a memory that can be attached to the electronic device (100) (e.g., memory card, external memory, etc.).

[0052] Instructions may be stored in the memory (110). The processor (120) may perform the operation of the electronic device (100) according to various embodiments of the present disclosure by executing the instructions in the memory (110) individually or collectively. Additionally, programs and data for operating the electronic device (100) may be stored in the memory (110). For example, the memory (110) may store one or more software applications, such as operating system (or system) software applications, firmware software applications, driver software applications, plugin (e.g., add-in, add-on, and / or applet) software applications, and / or any other suitable software applications.

[0053] Meanwhile, in the present disclosure, the term memory (110) may be used to include memory (110), ROM, RAM within the processor (120), or a memory card (e.g., micro SD card, memory stick) mounted in the electronic device (100). In addition, various information necessary within the scope of achieving the purpose of the present disclosure may be stored in the memory (110), and the information stored in the memory (110) may be updated as it is received from an external device or input by a user.

[0054] The processor (120) can control the overall operations of the electronic device (100). For example, the processor (120) can cause other components of the electronic device (100) to perform various operations by executing instructions stored in memory (110). For example, the processor (120) can control the operation of the electronic device (100) by being operatively connected to memory (110). Additionally, the processor (120) can control the operation of the electronic device (100) according to the present disclosure by executing one or more instructions stored in memory (110). The processor (120) may be composed of one or more processors.

[0055] The processor (120) may be implemented as one or more integrated circuit (or circuitry) chips and may perform various data processing operations. The processor (120) may include at least one electrical circuit and may process instructions (or programs, data, etc.) stored in memory (110) individually or collectively. The processor (120) may include a processor assembly comprising one or more processing circuits. The processor (120) may include any processing circuit that is operative to control the performance and operations of one or more components (e.g., memory (110)) of the electronic device (100). For example, the processor (120) (e.g., application processor (AP)) may be implemented as a system on chip (SoC) (e.g., a single chip or chipset). For example, the processor (120) may be implemented with a plurality of cores (or at least one core circuit), a plurality of chips, or a plurality of chipsets. For example, the processor (120) may include one or more processing circuits. For example, the processor (120) may include one or more processing circuits configured to perform the various functions of the present disclosure individually and / or collectively.

[0056] The processor (120) may include one or more of a CPU (Central Processing Unit), GPU (Graphics Processing Unit), APU (Accelerated Processing Unit), MIC (Many Integrated Core), DSP (Digital Signal Processor), NPU (Neural Processing Unit), hardware accelerator, or machine learning accelerator. The processor (120) may control one or any combination of other components of the electronic device (100) and may perform operations or data processing related to communication. The processor (120) may execute one or more programs or instructions stored in the memory (110) of the electronic device (100). For example, the processor (120) may perform a method according to one embodiment of the present disclosure by executing one or more instructions stored in the memory (110).

[0057] When a method according to one embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by a single processor or by a plurality of processors. For example, when a first operation, a second operation, and a third operation are performed by a method according to one embodiment, the first operation, the second operation, and the third operation may all be performed by a first processor, or the first operation and the second operation may be performed by a first processor (e.g., a general-purpose processor) and the third operation may be performed by a second processor (e.g., an artificial intelligence dedicated processor).

[0058] The processor (120) may be implemented as a single-core processor including one core, or as one or more multi-core processors including multiple cores (e.g., homogeneous multi-core or heterogeneous multi-core). When the processor (120) is implemented as a multi-core processor, each of the multiple cores included in the multi-core processor may include internal processor memory such as cache memory or on-chip memory, and a common cache shared by the multiple cores may be included in the multi-core processor. Additionally, each of the multiple cores included in the multi-core processor (or some of the multiple cores) may independently read and execute program instructions for implementing a method according to one embodiment of the present disclosure, or all (or some) of the multiple cores may be linked together to read and execute program instructions for implementing a method according to one embodiment of the present disclosure.

[0059] When a method according to one embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by one of the plurality of cores included in a multi-core processor, or may be performed by a plurality of cores. For example, when a first operation, a second operation, and a third operation are performed by a method according to one embodiment, the first operation, the second operation, and the third operation may all be performed by a first core included in a multi-core processor, or the first operation and the second operation may be performed by a first core included in a multi-core processor and the third operation may be performed by a second core included in a multi-core processor.

[0060] In the embodiments of the present disclosure, a processor may mean a system-on-chip (SoC) in which one or more processors and other electronic components are integrated, a single-core processor, a multi-core processor, or a core included in a single-core processor or a multi-core processor, wherein the core may be implemented as a CPU, GPU, APU, MIC, DSP, NPU, hardware accelerator, or machine learning accelerator, but the embodiments of the present disclosure are not limited thereto.

[0061] FIG. 3 is a block diagram illustrating the detailed configuration of an electronic device according to at least one embodiment of the present disclosure.

[0062] Referring to FIG. 3, the electronic device (100) may include a memory (110), a processor (120), and a communication circuit (130). However, such a configuration is exemplary, and it is understood that in carrying out the present disclosure, new configurations may be added or some configurations may be omitted in addition to such configurations. Meanwhile, detailed descriptions of configurations shown in FIG. 3 that overlap with configurations shown in FIG. 2 will be omitted.

[0063] The communication circuit (130) can perform data communication with an external electronic device under the control of the processor (120). The external electronic device may include a server, home appliance, mobile device (e.g., smartphone, tablet PC, wearable device, etc.). For example, the communication circuit (130) can communicate with the external electronic device through a network. For example, the processor (120) can receive data from the external electronic device through the communication circuit (130) and transmit data to the external electronic device through the communication circuit (130).

[0064] The communication circuit (130) may include hardware components to support the transmission and / or reception of electrical signals between the electronic device (100) and an external electronic device. For example, the communication circuit (130) may perform data communication between the electronic device (100) and the electronic device using at least one of a data communication method including wired LAN, wireless LAN, Wi-Fi, Wi-Fi Direct, Bluetooth, ZigBee, WFD (Wi-Fi Direct), infrared communication (IrDA, infrared Data Association), BLE (Bluetooth Low Energy), NFC (Near Field Communication), Wibro (Wireless Broadband Internet), WiMAX (World Interoperability for Microwave Access), SWAP (Shared Wireless Access Protocol), WiGig (Wireless Gigabit Alliances), and RF communication. The communication circuit (130) may also be expressed as a communication interface.

[0065] FIG. 4 is a flowchart illustrating an example of an operation in which an electronic device according to at least one embodiment of the present disclosure generates a persona. A processor (120) may perform at least one of the operations of FIG. 4. When instructions stored in memory (110) are executed individually or collectively by the processor (120), the electronic device (100) may be made to perform the operations of FIG. 4.

[0066] In operation 410, the electronic device (100) can generate a knowledge graph corresponding to the content. The content may include video content such as movies, dramas, and short-form videos.

[0067] A knowledge graph corresponding to content may be data that structures and represents the objects included in the content and the relationships (or interactions) between the objects in text form. A knowledge graph corresponding to content may be generated for each frame of the content.

[0068] In one embodiment, the electronic device (100) can obtain a knowledge graph corresponding to content using an artificial intelligence model. For example, the artificial intelligence model may include a first artificial intelligence model and a second artificial intelligence model. The electronic device (100) can input a frame of content and audio for the frame into the first artificial intelligence model to obtain data regarding objects and relationships between objects in the frame from the first artificial intelligence model. The first artificial intelligence model may include, for example, a model trained to identify objects and relationships between objects in the frame by processing data of different modalities. For example, the first artificial intelligence model may be a computer vision model. The electronic device (100) can input the data (e.g., output data of the first artificial intelligence model) into the second artificial intelligence model to obtain a knowledge graph from the second artificial intelligence model. The second artificial intelligence model may be a large language model (LLM). An LLM is a type of generative model that can be trained based on large-scale training data. For example, the LLM can understand the context, structure, and meaning of language and perform various language tasks such as text generation, translation, summarization, and responses to questions. However, the present disclosure is not limited thereto, and the electronic device (100) can acquire a knowledge graph using various methods.

[0069] FIG. 5 shows an example of a knowledge graph according to at least one embodiment of the present disclosure.

[0070] For example, Fig. 5 may be an example of a knowledge graph for a frame in which a dog is chasing a cat in the living room and a person in the living room is watching TV.

[0071] Referring to 510 in FIG. 5, the electronic device (100) can input output data of the first artificial intelligence model into the second artificial intelligence model to obtain a knowledge graph (510) from the second artificial intelligence model. The knowledge graph (510) may include data in which information about the content, atmosphere, objects, interactions between objects, etc. included in each frame is expressed in text form.

[0072] The knowledge graph (510) may contain information about objects included in the frame. For example, the knowledge graph (510) may include text indicating that the frame contains a dog, a cat, a person, and a TV (e.g., Objects: Dog, Cat, Person). In the knowledge graph, objects may correspond to nodes. Nodes are the basic components of the knowledge graph and may represent specific objects.

[0073] The knowledge graph (510) may include text representing the coordinates of an object (e.g., dog: 200,150,50,100, cat: 100,100,0,50, person: 123,456,789,12, tv: 255,255,255,255). For example, the coordinates of an object may be coordinates representing a bounding box for the object.

[0074] The knowledge graph (510) may include text representing interactions between objects (e.g., interaction: {dog, cat, chase}). Interactions between objects may correspond to edges. Edges are the basic components of the knowledge graph and may represent relationships between nodes.

[0075] The knowledge graph (510) may include text representing a background contained in a frame (e.g., Background: Living Room).

[0076] Referring to 520 in FIG. 5, the knowledge graph (510) can be represented in a schematic form. For example, nodes can be represented in the form of text inside a circle, and edges can be represented in the form of lines connecting nodes.

[0077] In operation 420, the electronic device (100) can generate multiple personas corresponding to content based on a knowledge graph.

[0078] For example, the electronic device (100) can input a knowledge graph, a question and answer, and a first prompt into a second artificial intelligence model to obtain multiple personas from the second artificial intelligence model. For example, the second artificial intelligence model may be an LLM. The electronic device (100) can store the obtained personas in a question and answer database.

[0079] A question-and-answer database may be a database that stores questions and answers from users using the question-and-answer function. The question-and-answer database may store not only questions and answers corresponding to content corresponding to the knowledge graph, but also questions and answers corresponding to various content. For example, questions and answers may include a query such as "Summarize the content of the movie OO" and an answer corresponding to the query such as "The movie OO is about a lonely protagonist meeting friends and becoming happy." The term "question-and-answer database" may be replaced with a term such as "database."

[0080] Additionally, in the present disclosure, a persona may include information about the attributes of a user expected to view the content.

[0081] The first prompt may include a prompt for requesting attributes of a user expected to view content based on multiple questions stored in a knowledge graph and a question-and-answer database.

[0082] For example, referring to FIG. 6a, an electronic device (100) can input a knowledge graph (601) corresponding to movie A, a question and answer (602) stored in a question and answer database, and a first prompt (603) into a second artificial intelligence model (604). The electronic device (100) can obtain multiple personas (605) from the second artificial intelligence model (604) as a result of the input data. For example, the first prompt (603) may include text such as, "Predict the attributes of the viewer in each query based on the knowledge graph and question and answer. The attributes are intended to be used to improve the performance of the response model and to generate mock questions. The attributes are general categories that can include various aspects of the viewer." The multiple personas (605) may include "50s, male, office worker, fan of actor B," "30s, female, fan of director C," "30s, female, viewer with an emotional disposition," etc.

[0083] In operation 430, the electronic device (100) can generate expected question responses for multiple personas.

[0084] The anticipated Q&A may include anticipated questions and anticipated responses to the anticipated questions corresponding to each of the multiple personas.

[0085] For example, anticipated queries may include what is predicted to be asked about the content by users of each persona. Additionally, anticipated responses to anticipated queries may include what is predicted to be answered by users of each persona in response to the anticipated queries.

[0086] For example, the electronic device (100) can input a persona, a knowledge graph, and a second prompt into a second artificial intelligence model to obtain an expected query corresponding to the persona from the second artificial intelligence model. The second artificial intelligence model may be an LLM.

[0087] The second prompt may include a prompt to request the creation of content that is predicted to be queried by the user of each persona.

[0088] For example, referring to FIG. 6a, the electronic device (100) inputs a persona (605), a knowledge graph (601) corresponding to Movie A, and a second prompt (606) into a second artificial intelligence model (604), and can obtain an expected query (607) corresponding to the persona from the second artificial intelligence model (604). For example, the second prompt (606) may include text such as, "What kind of questions would a viewer of a persona like {Predicted age group: 30s, Predicted interest: Movie directing, Predicted knowledge level: High, Loyalty to content series: Die-hard fan, Other attributes: Emphasize director} (e.g., Persona (605)) ask?" The expected query (607) may include text such as, "How did the director's directing style affect the storytelling?" The electronic device (100) can obtain an expected query (607) for each of a plurality of personas through the method described above.

[0089] The electronic device (100) can obtain an expected response (609) corresponding to an expected query.

[0090] For example, the electronic device (100) can input a persona, an expected query about the persona, and a third prompt into a second artificial intelligence model to obtain an expected response from the second artificial intelligence model. The second artificial intelligence model may be an LLM.

[0091] A third prompt may include a prompt to request the generation of content that is expected to be responded to by the user of each persona in response to an expected query.

[0092] For example, referring to FIG. 6a, the electronic device (100) inputs a persona (605), a predicted query (607) corresponding to the persona, and a third prompt (608) into a second artificial intelligence model (604), and obtains a predicted response (609) corresponding to the predicted query from the second artificial intelligence model (604). For example, the third prompt (608) may include text such as, "If a viewer of a persona such as {Predicted age group: 30s, Predicted interest: Movie directing, Predicted knowledge level: High, Series loyalty: Die-hard fan, Other attribute: Emphasizes director} (e.g., Persona (605)) asked a question such as, 'How did the director's directing style affect the storytelling?', what would be the appropriate response to this?" The predicted response (609) corresponding to the predicted query may include text such as, "Director C added depth to the story through non-linear storytelling and character-centered development." The electronic device (100) can obtain an expected response (609) corresponding to an expected query (607) for each of the multiple personas through the method described above.

[0093] The electronic device (100) can predict feedback corresponding to each of the multiple personas for the expected questions and expected responses corresponding to each of the multiple personas.

[0094] Feedback may include a numerical score indicating the degree to which a response to a question is appropriate. For example, feedback may indicate that the higher the number, the more appropriate the expected response to the expected question. For instance, feedback may be expressed as a score from 1 to 5. In this case, 1 point may correspond to very inappropriate, 2 points to somewhat inappropriate, 3 points to average, 4 points to somewhat appropriate, and 5 points to very appropriate.

[0095] The electronic device (100) can input a persona, an expected query corresponding to the persona, an expected response corresponding to the expected query, and a fourth prompt into a second artificial intelligence model to obtain feedback from the second artificial intelligence model. The second artificial intelligence model (604) may be an LLM. The electronic device (100) may store the expected query corresponding to the persona, the expected response corresponding to the expected query, and the feedback in a question-and-answer database.

[0096] The fourth prompt may include a prompt to request user feedback for each persona regarding a response corresponding to an expected query.

[0097] For example, referring to FIG. 6b, the electronic device (100) inputs a persona (605), an expected query (607) corresponding to the persona, an expected response (609) corresponding to the expected query, and a fourth prompt (610) into a second artificial intelligence model (604), and can obtain feedback (611) from the second artificial intelligence model (604). For example, the fourth prompt (610) may include text such as, "Regarding the expected query and expected response, what feedback would a viewer of a persona such as {Predicted age group: 30s, Predicted interest: Movie directing, Predicted knowledge level: High, Series loyalty: Die-hard fan, Other attributes: Emphasize director} (e.g., Persona (605)) show? Please express the score for the feedback on a scale of 5." The feedback (611) may include text such as, "4 points." The electronic device (100) can store the obtained feedback in a question-and-answer database. The electronic device (100) can obtain feedback on the question and answer of each of the multiple personas through the method described above.

[0098] In operation 440, the electronic device (100) can update multiple personas stored in a question-and-answer database.

[0099] The electronic device (100) can update multiple personas stored in a question-and-answer database based on expected queries for multiple personas, expected responses to expected queries, and feedback.

[0100] In the following description, an anticipated query, an anticipated response to the anticipated query, and a set of feedback are explained as evaluation information.

[0101] For example, the electronic device (100) can update a second artificial intelligence model based on evaluation information and update multiple personas stored in a question-and-answer database using the updated second artificial intelligence model. That is, the electronic device (100) can update the second artificial intelligence model based on evaluation information to improve the quality of expected questions and expected responses, and can update multiple personas by generating multiple personas using the updated second artificial intelligence model.

[0102] The electronic device (100) can update the second artificial intelligence model based on evaluation information.

[0103] For example, the electronic device (100) can obtain a first score by inputting an expected query, an expected response, and a fifth prompt into a third artificial intelligence model. Additionally, the electronic device (100) can obtain a second score by inputting an expected query, an expected response, feedback, and a fifth prompt into the third artificial intelligence model. Furthermore, the electronic device (100) can update the second artificial intelligence model by inputting the first score and the second score into the second artificial intelligence model.

[0104] The third artificial intelligence model may be a model trained to evaluate the quality of the output of the second artificial intelligence model. The third artificial intelligence model may be a model implemented based on deep learning. The third artificial intelligence model may be expressed using alternative terms such as a reward model or an evaluation model.

[0105] For example, the third artificial intelligence model receives an expected question and an expected response as input and can output a score indicating the degree of appropriateness of the expected response. In the following description, the score indicating the degree of appropriateness of the expected response is described as the first score.

[0106] Additionally, the third artificial intelligence model receives expected queries, expected responses, and feedback as input, and can output a score that predicts the user's evaluation regarding the set of expected queries, expected responses, and feedback. In the following description, the score that predicts the user's evaluation regarding the set of expected queries, expected responses, and feedback is described as the second score. The score that predicts the user's evaluation may include a numerical representation of the user's evaluation regarding whether the set of expected queries, expected responses, and feedback is a set likely to exist in reality. For example, if the expected response to an expected query is appropriate, it is desirable for the feedback to be high. Therefore, being likely to exist in reality may include cases where high-score feedback is assigned when the expected response to an expected query is desirable. Furthermore, being likely to exist in reality may include cases where low-score feedback is assigned when the expected response to an expected query is undesirable.

[0107] On the other hand, being unlikely to exist may include cases where low-score feedback is given when the expected response to an expected query is desirable. Additionally, being unlikely to exist may include cases where high-score feedback is given when the expected response to an expected query is undesirable.

[0108] The fifth prompt may include a prompt for requesting the first score and the second score.

[0109] For example, the fifth prompt (613) may include text such as “For persona-specific expected questions and expected responses to expected questions, please generate a score on how appropriate the expected response is. And for expected questions, expected responses to expected questions, and feedback, please generate a score on how appropriate and likely the expected questions, expected responses, and feedback are by referring to the question-and-answer database. Please express each score on a scale of 5.”

[0110] For example, referring to FIG. 6c, the electronic device (100) can obtain a first score (614) by inputting an expected query (607) corresponding to a persona, an expected response (609) corresponding to the expected query, and a fifth prompt (613) into a third artificial intelligence model (612). The first score (614) may include text such as "3 points".

[0111] When the electronic device (100) obtains the first score (614) from the third artificial intelligence model (612), it can update the second artificial intelligence model (604) based on the first score (614) so ​​that the second artificial intelligence model (604) outputs an expected response with improved quality.

[0112] For example, the electronic device (100) inputs a first score (614), a persona (605), an expected query (607) corresponding to the persona, and a third prompt (608) into the second artificial intelligence model (604) to obtain a new expected response (617) for the expected query from the second artificial intelligence model (604). The electronic device (100) inputs the newly obtained response (617) into the third artificial intelligence model (612) and can obtain a first score (e.g., 4 points) for the expected query input into the second artificial intelligence model (604) and the expected response generated by the second artificial intelligence model (604) according to the expected query. At this time, the second artificial intelligence model (604) can output an expected response (617) such that the first score (614) obtained from the third artificial intelligence model (612) increases. The electronic device (100) may repeatedly perform the operation of obtaining an expected response (617) from a second artificial intelligence model (604) until the first score (614) reaches a preset score. However, the present disclosure is not limited thereto. For example, the electronic device (100) may repeatedly perform the operation of obtaining an expected response according to various conditions, such as performing the operation of obtaining an expected response (617) a preset number of times.

[0113] Additionally, referring to FIG. 6d, the electronic device (100) can input evaluation information (616) and a fifth prompt (613) into a third artificial intelligence model (612) to obtain a second score (615). The second score (615) may include text such as "1 point".

[0114] The electronic device (100) obtains a second score (615) from a third artificial intelligence model (612) and can update the second artificial intelligence model (604) based on the second score (615) so that the second artificial intelligence model (604) outputs a pair (618) of expected responses and feedback with improved quality.

[0115] For example, when the electronic device (100) inputs the second score (615), persona (605), predicted query (607) corresponding to the persona, the third prompt (608), and the fourth prompt (610) into the second artificial intelligence model (604), it can newly obtain a pair of predicted responses and feedback (618) from the second artificial intelligence model (604). At this time, the second artificial intelligence model (604) can output the pair of predicted responses and feedback (618) so that the second score (615) obtained from the third artificial intelligence model (612) increases. The electronic device (100) inputs the newly obtained pair of predicted responses and feedback (618) into the third artificial intelligence model (612), and can obtain a second score (e.g., 3 points) for the predicted query (607) input into the second artificial intelligence model (604) and the predicted response and feedback generated by the second artificial intelligence model (604) according to the predicted query. The electronic device (100) may repeatedly perform the operation of obtaining expected response and feedback pairs (618) from the second artificial intelligence model (604) until the second score (615) reaches a preset score. However, the present disclosure is not limited thereto. For example, the electronic device (100) may repeatedly perform the operation of obtaining expected response and feedback pairs according to various conditions, such as performing the operation of obtaining expected response and feedback pairs (618) a preset number of times.

[0116] Through the example described above, the electronic device (100) can generate high-quality expected questions and answers corresponding to the persona (605) and the persona (605).

[0117] In an embodiment, the electronic device (100) can train the third artificial intelligence model by inputting a query entered by a user, a response generated by the second artificial intelligence model for the query, an expected query generated by the second artificial intelligence model, a response corresponding to the expected query, and feedback to the third artificial intelligence model. Through training, the third artificial intelligence model can generate a more accurate first score and second score.

[0118] The electronic device (100) can provide a response from a user's input query by generating a persona and an expected question and answer corresponding to the persona.

[0119] A specific method by which the electronic device (100) provides a response to a user is described based on FIG. 7. The processor (120) may perform at least one of the operations of FIG. 7. When instructions stored in memory (110) are executed individually or collectively by the processor (120), the electronic device (100) may be made to perform the operations of FIG. 7.

[0120] In operation 710, the electronic device (100) can receive a query corresponding to the content based on user input.

[0121] As mentioned above, the content may include video content such as movies, dramas, and short-form videos.

[0122] The electronic device (100) can receive a query corresponding to the content based on user input through a communication circuit (130).

[0123] For example, when a touch input is received for an icon displayed on the display (210), the user terminal device (200) may execute an application installed on the user terminal device (200). In the present disclosure, the application may be an application that provides content.

[0124] The user terminal device (200) can display a user interface capable of playing content on a display. The user terminal device (200) can receive a touch input corresponding to the content to be played among a list of multiple contents displayed on the user interface. When the user terminal device (200) receives a touch input, it can transmit a signal to the electronic device (100) requesting the transmission of data corresponding to the content for which the touch input was received. Based on the received signal, the electronic device (100) can transmit data corresponding to the content to the user terminal device (200).

[0125] Referring to FIGS. 8a and FIGS. 9a, the electronic device (100) receives data corresponding to content from a server and can display a content playback screen (810) on a display (210).

[0126] An application that plays content may include an application that provides a content query-and-answer service to the user. The content query-and-answer service may include a service that allows the user to receive a response corresponding to the query when the user inputs a query about the content.

[0127] For example, a content query response service may include a service that, when an electronic device (100) receives a query corresponding to content from a user, the electronic device (100) can provide an appropriate response to the query to the user based on the analysis of the content.

[0128] In the present disclosure, the query corresponding to the content may include questions related to, for example, the content of the content, characters, filming techniques, the atmosphere of a specific scene, etc. As an example, the question corresponding to the content may include questions such as "Summarize the content of the movie OO" or "What is the personality of the protagonist in the movie OO like?"

[0129] The user terminal device (200) can display a button (811) that allows the user to make a query corresponding to the content in one area of ​​the playback screen (810).

[0130] When a touch input for a button (811) is received, the user terminal device (200) can display an input field (911) for inputting a query corresponding to the content on the display (210).

[0131] When the user terminal device (200) receives a touch input for an input field (911), it may display a soft keyboard on the display that allows the user to input text. The soft keyboard may include a virtual keyboard displayed on the display. When the user terminal device (200) receives a touch input for the soft keyboard, it may display text corresponding to the location where the touch input was received in the input field (911).

[0132] When a user terminal device (200) receives a touch input for a button (912) requesting that the input text be transmitted to the electronic device (100), the user terminal device (200) can transmit the text entered by the user to the electronic device (100). The electronic device (100) can identify the received text as a query entered by the user. The electronic device (100) can display the received text on the display (210) (913). Additionally, the electronic device (100) can display text containing a response corresponding to the query on the display (210) (914).

[0133] In the example described above, it was explained that the electronic device (100) receives a query based on the user's touch input, but it is not limited thereto and the electronic device (100) may also receive a query based on the user's voice input.

[0134] For example, the user terminal device (200) may include a microphone. When the user terminal device (200) receives user voice input using the microphone, it can convert the received voice data into text form. The user terminal device (200) can transmit the user question converted into text form to the electronic device (100).

[0135] In addition, in one embodiment, the electronic device (100) receives voice data and the electronic device (100) may convert the voice signal into text. Additionally, the user terminal device (200) may transmit the data to a separate server capable of converting the voice data into text. The separate server capable of converting the voice data into text may convert the voice data into text and transmit the voice data to the electronic device (100).

[0136] In addition, based on FIGS. 8a and FIGS. 9a, the above description describes that the user terminal device (200) provides a content question and answer service to the user, but it is not limited thereto and the content question and answer service can be provided by various devices.

[0137] For example, referring to FIGS. 8b and FIGS. 9b, the display device (300) may provide a content query-response service to the user.

[0138] In one embodiment, referring to FIG. 8b, a display device (300) can display content (820) on a display screen (310). The display device (300) can display a button (821) in a part of the display screen (310) that allows a user to make a query corresponding to the content.

[0139] When the display device (300) receives input for the button (821), it can display a screen providing a content query response service on the display screen (310).

[0140] For example, the display device (300) can divide the display screen (310) into multiple areas to simultaneously display content and a screen providing content question and answer services.

[0141] Referring to FIG. 9b, the display device (300) can divide the display screen (310) into two areas (920, 930). The display device (300) can display content in the first area (920) and a screen providing content query response services in the second area (930).

[0142] As described above, in the second area (930) providing a question and answer service, an input field (931), a button (932) for sending text entered in the input field (931) to a server, and a button (933) for entering a question by voice may be displayed.

[0143] Additionally, as described above, when a user's query is transmitted to the server, the display device (300) may also display a UI element (934) that displays the user's query and a UI element (935) that displays a response corresponding to the query on the display screen (310).

[0144] The content regarding the UI elements displayed in the second area (930) is the same as described above, so the duplicate description is omitted.

[0145] In operation 720, the electronic device (100) can identify a persona corresponding to a query among a plurality of personas corresponding to content based on a question-and-answer database.

[0146] When the electronic device (100) generates a persona and an expected question and answer corresponding to the persona, it can identify the persona corresponding to the question received from the user.

[0147] The electronic device (100) can identify the similarity between the query received from the user and the query stored in the query-response database before identifying the persona corresponding to the query received from the user. If the similarity is less than or equal to a preset value, the electronic device (100) can identify the persona corresponding to the query received from the user. This is explained based on FIG. 10.

[0148] In operation 1010, the electronic device (100) can identify whether a query corresponding to the content has been received based on user input.

[0149] The electronic device (100) can identify whether a query has been received using a communication circuit (130) based on user input. The details regarding this are the same as those described above.

[0150] In operation 1020, the electronic device (100) can identify whether the received query is similar to a query stored in the query-response database.

[0151] The electronic device (100) can identify the similarity between a query received from a user and a plurality of queries stored in a query-response database.

[0152] For example, an electronic device (100) can identify the similarity between each query. A method for the electronic device (100) to identify the similarity between queries may include converting each sentence into a vector to identify the cosine similarity. The electronic device (100) can identify that the closer the cosine similarity is to 1, the higher the similarity between each sentence.

[0153] Additionally, the electronic device (100) may include a Euclidean distance that converts sentences into vectors and calculates the straight-line distance between vectors. The electronic device (100) may identify that the smaller the identified Euclidean distance, the higher the similarity. The method by which the electronic device (100) identifies similarity between sentences is not limited to the examples described above, and similarity between sentences may be identified through various methods. Additionally, the electronic device (100) may identify similarity between sentences based on identifying the context included in the sentences using the methods described above and a second artificial intelligence model.

[0154] The electronic device (100) can identify that a query received from a user and a query included in the query-response database are similar if the similarity of each query exceeds a threshold. On the other hand, the electronic device (100) can identify that a query received from a user and a query included in the query-response database are dissimilar if the similarity of each query is less than a threshold.

[0155] In operation 1020-Y, 1030, if the electronic device (100) is similar to a query received from a user and a query included in a query-response database, it can provide a response to the user based on a response stored in the query-response database.

[0156] The electronic device (100) can provide a user with an answer corresponding to a response included in a question-and-answer database. When the electronic device (100) identifies that among a plurality of queries stored in the question-and-answer database, there exists a query with a similarity level greater than or equal to a threshold, it can obtain a response corresponding to the identified query from the question-and-answer database.

[0157] For example, if the electronic device (100) identifies that a query received from a user is similar to a query included in a query-response database, it can identify a response among the queries stored in the query-response database that corresponds to the query received from the user.

[0158] The electronic device (100) can identify the identified response as a response corresponding to a query received from the user and transmit a control signal to the user terminal device (200) to command the response to be displayed on the display (210).

[0159] In operation 1120-N, 1140, the electronic device (100) can identify a persona corresponding to the user among the personas stored in the question-and-answer database.

[0160] If the electronic device (100) is dissimilar to the query received from the user and the query included in the query-response database, it can identify the persona corresponding to the user among the multiple personas stored in the query-response database.

[0161] For example, referring to FIG. 11, the electronic device (100) can identify the similarity between a query (1110) included in a query-response database and a query (1120) received from a user. The method by which the electronic device (100) identifies the similarity between queries is as described above.

[0162] The electronic device (100) can identify the similarity between a query (1110) corresponding to each of a plurality of personas and a query (1120) received from a user. The electronic device (100) can obtain an average value of the similarity of the query corresponding to each persona. The electronic device (100) can obtain (1130) the persona with the highest average value of the similarity of the query corresponding to each persona as the persona corresponding to the user's query.

[0163] For example, the electronic device (100) can identify that the average similarity value of the query corresponding to the first persona is 60, the average similarity value of the query corresponding to the second persona is 40, and the average similarity value of the query corresponding to the third persona is 10. At this time, the electronic device (100) can identify that the persona corresponding to the user is the first persona because the persona with the highest average similarity value of the query is the first persona.

[0164] In operation 1050, the electronic device (100) may provide a response to the user in response to a query based on an identified persona. Specific details regarding this are explained based on operation 730.

[0165] In operation 730, the electronic device (100) can obtain a response corresponding to a query based on an identified persona.

[0166] When the electronic device (100) identifies that there is no query among multiple queries that has a similarity level greater than or equal to a threshold value, it can obtain a response corresponding to the query based on the identified persona.

[0167] When the electronic device (100) identifies that a question received from a user and a question included in the question-answer database are dissimilar, it can identify a persona corresponding to the user from the personas stored in the question-answer database. When the electronic device (100) identifies a persona, it can input the question into a second artificial intelligence model to generate an answer based on the user's attributes.

[0168] In operation 740, the electronic device (100) can provide the user with the obtained response.

[0169] The electronic device (100) can transmit a control signal to the user terminal device using a communication circuit (130) to command the generated response to be displayed on the display (210) of the user terminal device.

[0170] The electronic device (100) can transmit a control signal to the user terminal device (200) that commands the generated response to be displayed on the display (210) of the user terminal device (200).

[0171] Based on receiving a control signal, the user terminal device (200) can display the response (1040) generated by the electronic device (100) on the display (210) as shown in FIG. 10.

[0172] Based on FIG. 7, through the operation described above, the electronic device (100) can generate an appropriate response based on user attributes.

[0173] The electronic device (100) can perform the task of optimizing the knowledge graph to generate a more appropriate response and to perform memory management more efficiently.

[0174] For example, the electronic device (100) can perform the task of optimizing the knowledge graph by generating a persona, an expected query for each persona, and an expected response to the expected query.

[0175] The operation of optimizing the knowledge graph may include deleting or adding nodes and edges included in the knowledge graph. The operation of the electronic device (100) optimizing the knowledge graph is described with reference to FIGS. 12 and FIGS. 13.

[0176] FIG. 12 is a flowchart illustrating the operation of deleting nodes and edges included in a knowledge graph according to at least one embodiment of the present disclosure.

[0177] The electronic device (100) can obtain the usage rate of at least one of the nodes and edges used to obtain a response corresponding to a question, and can update the knowledge graph based on the obtained usage rate.

[0178] In operation 1210, the electronic device (100) can evaluate the utilization rate of nodes and edges included in the knowledge graph. The utilization rate of nodes and edges included in the knowledge graph may include the ratio of the number of times the second artificial intelligence model used the nodes and edges included in the knowledge graph to generate a response, respectively, to the number of queries stored in the question-answer database.

[0179] At this time, the questions stored in the question-and-answer database may include expected questions generated by the second artificial intelligence model and questions received from actual users.

[0180] The electronic device (100) can identify the number of queries stored in the query-response database and identify the number of times each node and edge included in the knowledge graph are used to generate a response corresponding to the query. The electronic device (100) can identify the number of times at least one of the nodes and edges is used for the number of queries stored in the query-response database as the usage rate of at least one of the nodes and edges.

[0181] In operation 1220, the electronic device (100) can assign weight scores to nodes and edges included in the knowledge graph.

[0182] The electronic device (100) can assign weight scores to each node and edge based on identifying the usage rates of the nodes and edges.

[0183] The weight score can be proportional to the utilization rate of nodes and edges.

[0184] For example, it is assumed that 100 questions are stored in a question-and-answer database. The electronic device (100) can identify that the first node stored in the knowledge graph is used 90 times to generate responses corresponding to the 100 questions. Additionally, the electronic device (100) can identify that the second node is used 95 times and the third node is used 30 times.

[0185] The electronic device (100) can be identified as having a usage rate of 0.9 for the first node, 0.95 for the second node, and 0.3 for the third node.

[0186] The electronic device (100) can assign weight scores to each node based on usage. Thus, the electronic device (100) can identify that the second node has the highest weight score and the third node has the lowest weight score. The method by which the electronic device (100) assigns weight scores to each node based on usage may be determined by a rule-based method of the processor (120). A rule-based method may be a method in which the electronic device (100) uses pre-set rules to perform a specific action.

[0187] In operation 1230, the electronic device (100) can identify nodes and edges included in the knowledge graph whose weight score is less than or equal to a preset score.

[0188] The electronic device (100) can identify nodes and edges whose scores are less than or equal to a preset score by assigning weight scores to each node and edge.

[0189] If the electronic device (100) has nodes and edges included in multiple knowledge graphs, it can identify nodes and edges included in multiple knowledge graphs, rather than identifying only the nodes and edges included in a specific knowledge graph.

[0190] In operation 1230-Y, 1240, the electronic device (100) can delete at least one of the nodes and edges whose weight score is less than or equal to a preset score from the knowledge graph.

[0191] The electronic device (100) can delete identified nodes and edges from the knowledge graph.

[0192] If the electronic device (100) has nodes and edges included in multiple knowledge graphs, it can delete nodes and edges included in multiple knowledge graphs, rather than just deleting nodes and edges included in a specific knowledge graph.

[0193] On the other hand, the electronic device (100) may not delete nodes and edges having a preset score or higher from the knowledge graph.

[0194] According to the operation of Fig. 12, the electronic device (100) can efficiently manage memory by deleting nodes and edges with low usage rates from the knowledge graph.

[0195] The electronic device (100) may delete nodes and edges from the knowledge graph, but may also add them. The specific operation of the electronic device (100) adding nodes and edges to the knowledge graph is explained based on FIG. 13.

[0196] FIG. 13 is a flowchart illustrating the operation of generating nodes and edges according to at least one embodiment of the present disclosure.

[0197] In operation 1310, the electronic device (100) can identify whether the coverage level for queries on nodes and edges included in the knowledge graph is less than or equal to a preset value.

[0198] The degree of coverage for queries on nodes and edges may include a value indicating whether the electronic device (100) can generate a response based on an existing knowledge graph when generating a response corresponding to a query. The value indicating whether the electronic device (100) can generate a response based on a knowledge graph may include the ratio of the number of queries for which a response can be generated based on a knowledge graph to the number of queries stored in a query-response database.

[0199] At this time, the queries stored in the question-and-answer database may include queries received from the user and predicted queries generated by the second artificial intelligence model.

[0200] If the electronic device (100) can identify the factual relationship of the content for generating a response based on the knowledge graph for a query, it can identify that the degree of coverage for the query of the nodes and edges is high.

[0201] For example, an electronic device (100) may receive a query such as "What are the dog and the cat doing?". The electronic device (100) may refer to a knowledge graph (e.g., FIG. 5) to identify that the dog and the cat are included in the nodes contained in the knowledge graph. Additionally, the electronic device (100) may refer to the knowledge graph to identify that the dog and the cat are chasing each other based on the edges contained in the knowledge graph. Thus, the electronic device (100) may generate a response such as "The dog and the cat are chasing each other" in response to a query such as "What are the dog and the cat doing?". Since the electronic device (100) can generate a response based on the knowledge graph for the received query, it can be identified that the electronic device (100) can cover the query stored in the query-response database using the nodes and edges contained in the knowledge graph.

[0202] There may be cases where the electronic device (100) fails to generate a response based on a knowledge graph among the queries stored in the question-and-answer database. For example, if the electronic device (100) cannot identify the factual relationship of the content for generating a response based on the knowledge graph, the electronic device (100) may not be able to generate a response.

[0203] Accordingly, if the electronic device (100) identifies that the number of queries for which a response cannot be generated using a knowledge graph is greater than or equal to a preset value for the number of queries stored in the question-and-answer database, it can identify that the coverage level of the queries of the nodes and edges included in the knowledge graph is low.

[0204] In operation 1320, the electronic device (100) can identify at least one of a potential node and a potential edge.

[0205] If the electronic device (100) identifies that the coverage level for a query of a node and edge included in the knowledge graph is low, it can create a new node and edge included in the knowledge graph.

[0206] For example, if the electronic device (100) fails to generate a response based on the knowledge graph, it may identify at least one of a potential node and a potential edge. A potential node may include a node scheduled to be added to the knowledge graph. Additionally, a potential edge may include an edge scheduled to be added to the knowledge graph.

[0207] Accordingly, the electronic device (100) can identify potential nodes and potential edges for generating a response corresponding to a query if the number of queries stored in the query-response database for which a response cannot be generated using a knowledge graph is greater than or equal to a preset ratio of the total number of questions.

[0208] For example, an electronic device (100) may receive a query such as "What color are dogs and cats?". It is assumed that the knowledge graph (e.g., FIG. 5) contains information about dogs and cats, but does not contain information about the colors of dogs and cats. The electronic device (100) may identify that it cannot generate a response by referring to the knowledge graph. The electronic device (100) may identify that it cannot generate a response by referring to the knowledge graph for the query, and that it will acquire information about the colors of dogs and cats if the number of queries about the colors of cats and dogs is greater than or equal to the number of queries stored in the query-response database. At this time, the electronic device (100) may identify the information about the colors of dogs and cats as potential nodes. The electronic device (100) may store keywords of nodes to be generated (e.g., dog color and cat color) in memory (110).

[0209] The method by which the electronic device (100) identifies potential nodes and potential edges is not limited to the examples described above, and potential nodes and potential edges can be identified in various ways. Additionally, the electronic device (100) may identify only one of the potential nodes and potential edges, or it may identify both potential nodes and potential edges.

[0210] In operation 1330, the electronic device (100) can generate at least one of the identified potential nodes and potential edges.

[0211] When the electronic device (100) identifies at least one of the potential node and potential edge, it can analyze the content to identify information about the potential node and potential edge.

[0212] As described above, the electronic device (100) can analyze content using a first artificial intelligence model. Based on the results of analyzing the content, the electronic device (100) can identify information about at least one of potential nodes and potential edges.

[0213] When the electronic device (100) identifies information about potential nodes and potential edges, it can add the potential nodes and potential edges as nodes and edges to the knowledge graph.

[0214] The electronic device (100) can add nodes and edges to a knowledge graph for a specific point in time of content if the query related to potential nodes and potential edges concerns a specific point in time of content. For example, the electronic device (100) can receive a query such as, "I want to ask about the scene that is displayed when the playback time of the content reaches 30 minutes. What action did the protagonist take at that time?" The electronic device (100) can analyze the content and, based on the results of the analysis, add nodes and edges regarding the protagonist's actions to a knowledge graph corresponding to the scene that is displayed when the playback time of the content reaches 30 minutes.

[0215] Additionally, the electronic device (100) can add nodes and edges to the entire knowledge graph of the content if the query related to the potential nodes and potential edges is about the content as a whole rather than a specific point in time of the content. For example, the electronic device (100) can receive a query such as "What are the colors of dogs and cats?". The electronic device (100) can analyze the content and, based on the results of the analysis, add nodes and edges related to the colors of cats and dogs to the knowledge graph containing the nodes corresponding to cats and dogs within the knowledge graph of the content.

[0216] On the other hand, the electronic device (100) may not add nodes and edges to the knowledge graph if the degree of question coverage of nodes and edges included in the knowledge graph performed in operation 1310 exceeds a preset value.

[0217] In the example described above, a plurality of artificial intelligence models used to perform the operation of the electronic device (100) are related to a technology that automatically processes human experience information into knowledge data and can perform knowledge construction (e.g., data generation / classification), knowledge management (e.g., data utilization), etc.

[0218] In addition, in the example described above, the electronic device (100) is assumed to be a server device, but the operation according to the present disclosure can also be performed on a user terminal device (e.g., smartphone, TV, tablet PC, etc.).

[0219] If the electronic device (100) is a user terminal device, the electronic device (100) may include a display. If the electronic device (100) has a built-in display, the electronic device (100) may control the display to display content and a user interface related to the content.

[0220] Additionally, the electronic device (100) can perform operations such as generating a persona using a plurality of artificial intelligence models and processors stored in the memory of the electronic device, generating expected questions and answers based on the generated persona and updating the artificial intelligence model, and identifying a persona corresponding to a user.

[0221] Additionally, if the electronic device (100) is a user terminal device, there may be cases where it does not include a display unit or plays content on an external display device. In this case, the electronic device (100) may include video and audio output ports. The audio and video output ports may include ports capable of transmitting audio and video signals to a display device. The audio and video output ports may include, for example, HDMI, DP, Thunderbolt, etc. The electronic device (100) may simultaneously transmit video and audio signals to a display device using the audio and video output ports.

[0222] Additionally, the electronic device (100) may transmit video and audio signals individually to a display device using video and audio output ports.

[0223] The electronic device (100) can transmit video and audio signals to an external display device through video and audio output ports so that content can be played on the external display device. Accordingly, the electronic device (100) can control the display included in the external display device to display content or a user interface related to the content.

[0224] Additionally, a remote control application may be installed on the electronic device (100). The remote control application may include an application that enables the electronic device (100) to control an external display device like a remote control. The electronic device (100) may transmit various control signals to the external display device through the remote control application. The applications that can be installed on the electronic device (100) are not limited to remote control applications, and various applications may be installed.

[0225] An electronic device according to the present disclosure may include at least one processor comprising a memory for storing instructions and processing circuitry. When the instructions are executed individually or collectively by the at least one processor, the electronic device may, upon receiving user input corresponding to a query related to content, acquire a persona corresponding to the query among a plurality of personas corresponding to the content based on a database, and provide a response corresponding to the query based on the acquired persona. The plurality of personas may be based on a knowledge graph corresponding to the content and the database. This can improve the relevance and quality of the generated response. By tailoring the response to the predicted user persona instead of providing a general response, the system can increase user satisfaction and provide more meaningful interactions that are directly adapted to the user's potential interests and knowledge level.

[0226] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may acquire the plurality of personas corresponding to the content through an artificial intelligence model based on a knowledge graph corresponding to the content, the database, and a first prompt. The first prompt may include a prompt corresponding to a prompt requesting the prediction of user attributes corresponding to the content for multiple queries in the database based on the knowledge graph corresponding to the content. This automates the generation of user personas and enables scalability. Since the artificial intelligence model can autonomously discover complex and non-obvious user archetypes from content data instead of relying on manual work, the system can be applied to a vast amount of different content without human intervention.

[0227] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may acquire evaluation information corresponding to an expected query for each of the plurality of personas, an expected response to the expected query, and feedback on the expected response, and update the plurality of personas through an updated artificial intelligence model based on the evaluation information. In this way, a self-improving system can be implemented. By the system itself generating synthetic data and evaluating it, the accuracy of the personas and the quality of the artificial intelligence model can be continuously refined over time, and the system performance and response quality can be progressively improved without new actual user data.

[0228] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain the expected query through an artificial intelligence model based on the persona, a knowledge graph corresponding to the content, and a second prompt; obtain the expected response through the artificial intelligence model based on the persona, a knowledge graph corresponding to the content, the expected query, and a third prompt; and obtain the feedback through the second artificial intelligence model based on the persona, a knowledge graph corresponding to the content, the expected query, the expected response, and a fourth prompt. The second prompt may include a prompt corresponding to a user's expected query request corresponding to the persona based on a knowledge graph corresponding to the content; the third prompt may include a prompt corresponding to a user's expected response request regarding the expected query corresponding to the persona based on a knowledge graph corresponding to the content; and the fourth prompt may include a prompt corresponding to a user's feedback request regarding the expected response corresponding to the persona based on a knowledge graph corresponding to the content. This can automate the generation of complete and structured datasets for synthetic learning. By using a series of prompts, the system can generate a consistent set of query, response, and feedback data, which is more effective for training artificial intelligence models than unstructured or incomplete data, thereby leading to more robust and realistic model behavior.

[0229] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain a first score through an evaluation model based on a prompt corresponding to a score request regarding the quality of the expected query and expected response, and update the artificial intelligence model based on the first score. This can improve the direct quality and relevance of the generated response. By optimizing a high score that directly measures the question-response pair, the system can be fine-tuned to generate a response that is not only relevant to the persona but also factual and contextually appropriate.

[0230] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain a second score through an evaluation model based on a prompt corresponding to a score request regarding a set including the expected query, expected response, and feedback and the quality of the set, and update the artificial intelligence model based on the second score. This can improve the overall plausibility and human similarity of the simulated user interaction. By evaluating the entire interaction loop, the system learns how to generate overall consistent data, and as a result, a more realistically learned and robust artificial intelligence model is built, which can handle more nuanced user interactions.

[0231] A knowledge graph corresponding to the above content includes, for each of the multiple frames constituting the content, at least one node, at least one edge, and a text description of the frame corresponding to the frame, wherein the node corresponds to an object included in the frame, and the edge corresponds to a relationship between the objects included in the frame. This can generate a structured and machine-readable representation of the semantic information of the content. Through this graph structure, an AI model can efficiently query, explore, and infer complex relationships between characters, objects, and events within the content, which is much more efficient than analyzing raw video data for every query.

[0232] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may acquire the utilization of at least one of the nodes and edges corresponding to the acquisition of the response to the query, and update the knowledge graph based on the acquired utilization. This can improve memory efficiency and reduce data storage requirements. By dynamically pruning irrelevant or low-usage information from the knowledge graph, the system can reduce memory usage, accelerate future graph search operations, and further improve overall system performance.

[0233] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may obtain an average value of similarity for each of the plurality of personas based on similarities between the query and the plurality of queries corresponding to the plurality of personas stored in the database, and identify the persona having the largest value among the obtained average values ​​as the persona corresponding to the query. This can provide a robust and deterministic method for identifying the most relevant user persona. Instead of relying on heuristic-based estimation, the present method can enable more accurate and reliable response customization by using quantifiable indicators to ensure that subsequent personalization is based on the optimal match possible.

[0234] When the above instructions are executed individually or collectively by the at least one processor, the electronic device may identify the similarity between the query and a plurality of queries in the database; if there is a query among the plurality of queries with a similarity greater than or equal to a threshold, it may obtain a response corresponding to the query with the threshold greater than or equal to the query from the database and provide the response; and if there is no query among the plurality of queries with a similarity greater than or equal to the threshold, it may obtain a response corresponding to the query based on the obtained persona. This can reduce system latency and reduce computational load. By implementing a two-tier response strategy, the system may not perform an AI-based personalization process with high computational costs for general or simple queries, but may perform it only when necessary, thereby providing a more efficient and responsive user experience. The electronic device may adopt an adaptive multipath response strategy based on the characteristics of the received user query. This strategy may include a step of determining a similarity value between the user query and a plurality of queries stored in the database. If it is determined that the similarity value exceeds a predetermined threshold, the electronic device may be configured to perform a first operation, such as selecting and providing a corresponding stored response, which provides an efficient and low-latency response path. Conversely, if it is determined that the similarity value is below a predetermined threshold, the electronic device may be configured to perform a second operation distinct from the first. The second operation may include the step of identifying a user persona corresponding to a query and generating a new personalized response based on the identified persona, thereby enabling new or complex queries to be processed with maximum relevance and quality.

[0235] According to the present disclosure, a control method for an electronic device may include the steps of receiving user input corresponding to a query related to content, obtaining a persona corresponding to the query among a plurality of personas corresponding to said content based on a database, and providing a response corresponding to said query based on said persona. The plurality of personas may be based on a knowledge graph corresponding to said content and said database. This can improve the relevance of the provided response and user satisfaction. By adapting the response to a specific user persona, the present method can transform a general question-and-answer interaction into a personalized conversation, making the information provided to the user more interesting and useful.

[0236] The above method may further include the step of acquiring the plurality of personas corresponding to the content through an artificial intelligence model based on a knowledge graph corresponding to the content, the database, and a first prompt. The first prompt may include a prompt corresponding to a prompt requesting the prediction of user attributes corresponding to the content for a plurality of queries in the database based on the knowledge graph corresponding to the content. This can automate the generation of rich and diverse user personas. Through this, the present method can be applied to large and diverse content libraries without manual work, and can implement a scalable personalization system that autonomously discovers complex user profiles.

[0237] The above method may further include the step of obtaining evaluation information corresponding to a predicted query corresponding to a persona, a predicted response to the predicted query, and feedback on the predicted response for each of the plurality of personas; and the step of updating the plurality of personas through an artificial intelligence model updated based on the evaluation information. This can form a self-improving feedback loop within the system. Through this, the present method can continuously learn and improve the quality of the personas and the underlying artificial intelligence model over time, and as a result, the system performance and the accuracy of personalized responses can be gradually improved.

[0238] The step of updating the plurality of personas may include: a step of obtaining the predicted query through an artificial intelligence model based on the persona, a knowledge graph corresponding to the content, and a second prompt; a step of obtaining the predicted response through the artificial intelligence model based on the persona, a knowledge graph corresponding to the content, the predicted query, and a third prompt; and a step of obtaining the feedback through the second artificial intelligence model based on the persona, a knowledge graph corresponding to the content, the predicted query, the predicted response, and a fourth prompt. The second prompt may include a prompt corresponding to a user's predicted query request corresponding to the persona based on a knowledge graph corresponding to the content, the third prompt may include a prompt corresponding to a user's predicted response request regarding the predicted query corresponding to the persona based on a knowledge graph corresponding to the content, and the fourth prompt may include a prompt corresponding to a user's feedback request regarding the predicted response corresponding to the persona based on a knowledge graph corresponding to the content. This can automatically generate a complete and structured training dataset for the artificial intelligence model. By using a sequence of dedicated prompts, this method can generate a consistent set of query, response, and feedback data, which can be utilized more effectively for learning to contribute to the implementation of a more robust and realistic artificial intelligence model.

[0239] The step of updating the artificial intelligence model may include: a step of obtaining a first score through an evaluation model based on a prompt corresponding to a score request related to the quality of the expected query and expected response; and a step of updating the artificial intelligence model based on the first score. This may enable the output quality of the artificial intelligence model to be optimized in a goal-oriented and quantitative manner. By focusing the update process on a specific score that measures the appropriateness of the response, the present method may enable the artificial intelligence model to be specifically fine-tuned to generate responses of higher quality and contextual relevance.

[0240] Although various embodiments have been described above, each embodiment is not necessarily implemented individually, and may be combined with at least one other embodiment, either wholly or partially, to be implemented together in a single product.

[0241] Meanwhile, embodiments of the present disclosure may also be implemented in the form of a recording medium containing computer-executable instructions, such as program modules executed by a computer. A computer-readable medium may be any available medium accessible by a computer and includes both volatile and non-volatile media, and both removable and non-removable media. Additionally, a computer-readable medium may include computer storage media and communication media. Computer storage media include both volatile and non-volatile, removable and non-removable media implemented by any method or technique for storing information, such as computer-readable instructions, data structures, program modules, or other data. Communication media may typically include other data of modulated data signals, such as computer-readable instructions, data structures, or program modules.

[0242] Additionally, computer-readable storage media may be provided in the form of non-transitory storage media. Here, 'non-transitory storage media' simply means that it is a tangible device and does not contain a signal (e.g., electromagnetic waves), and this term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily. For example, 'non-transitory storage media' may include a buffer in which data is stored temporarily.

[0243] According to one embodiment, the method according to the various embodiments disclosed herein may be provided by being included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)), or distributed online (e.g., download or upload) through an application store or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product (e.g., downloadable app) may be temporarily stored or temporarily created on a device-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0244] The foregoing description of the present disclosure is for illustrative purposes only, and those skilled in the art will understand that other specific forms can be easily modified without altering the technical spirit or essential features of the present disclosure. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. For example, each component described as a single unit may be implemented in a distributed manner, and components described as distributed may likewise be implemented in a combined form.

[0245] The scope of the present disclosure is defined by the claims set forth below rather than by the detailed description above, and all modifications or variations derived from the meaning and scope of the claims and equivalent concepts thereof should be interpreted as being included within the scope of the present disclosure.

Claims

1. In an electronic device (100), Memory (110) for storing instructions; and It includes at least one processor (120) including a processing circuitry; and When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), When user input corresponding to a query (201, 1110, 1120) related to content is received, a persona (605) corresponding to the query (201, 1110, 1120) is obtained among a plurality of personas (605) corresponding to the content based on a database, and Based on the above-mentioned acquired persona (605), a response (202, 617, 1040) corresponding to the above-mentioned query (201, 1110, 1120) is provided, and The above plurality of personas (605) are electronic devices based on a knowledge graph (510, 601) corresponding to the content and the database.

2. In Paragraph 1, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), Based on the knowledge graph (510, 601) corresponding to the above content, the database, and the first prompt (603), the plurality of personas (605) corresponding to the above content are obtained through an artificial intelligence model, and The above first prompt (603) is, An electronic device comprising a prompt that corresponds to a request to predict attributes related to a user corresponding to the content for a plurality of queries of the database based on a knowledge graph (510, 601) corresponding to the content.

3. In Paragraph 2, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), For each of the above multiple personas (605), evaluation information corresponding to an expected query (607) corresponding to the persona (605), an expected response (609) to the expected query (607), and feedback (611) to the expected response (609) is obtained. An electronic device that updates the plurality of personas (605) through an artificial intelligence model updated based on the above evaluation information.

4. In Paragraph 3, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), The predicted query (607) is obtained through an artificial intelligence model based on the above persona (605), the knowledge graph (510, 601) corresponding to the above content, and the second prompt (606), and Based on the above persona (605), the knowledge graph (510, 601) corresponding to the above content, the above expected query (607), and the third prompt (608), the above expected response (609) is obtained through the above artificial intelligence model, and The feedback (611) is obtained through the second artificial intelligence model (604) based on the persona (605), the knowledge graph (510, 601) corresponding to the content, the expected query (607), the expected response (609), and the fourth prompt (610). The second prompt (606) includes a prompt that corresponds to an expected query (607) request from a user corresponding to the persona (605) based on a knowledge graph (510, 601) corresponding to the content, and The third prompt (608) includes a prompt that corresponds to a request for an expected response (609) to an expected query (607) of a user corresponding to the persona (605) based on a knowledge graph (510, 601) corresponding to the content, and The above fourth prompt (610) is an electronic device comprising a prompt that responds to a request for feedback (611) on the expected response (609) of a user corresponding to the persona (605) based on a knowledge graph (510, 601) corresponding to the content.

5. In Paragraph 3 or 4, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), A first score (614) is obtained through an evaluation model based on a prompt corresponding to a score request related to the quality of the above-mentioned expected query (607) and expected response (609), and An electronic device that updates the artificial intelligence model based on the first score (614) above.

6. In any one of paragraphs 3 through 5, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), A second score (615) is obtained through an evaluation model based on a set including the above-mentioned expected query (607), expected response (609), and feedback (611), and a prompt corresponding to a score request related to the quality of the set, and An electronic device that updates the artificial intelligence model based on the second score (615) above.

7. In any one of paragraphs 1 through 6, The knowledge graph (510, 601) corresponding to the above content is, For each of the plurality of frames constituting the above content, it includes at least one node corresponding to the frame, at least one edge, and a description of the text (914) of the frame, and The above node corresponds to an object included in the above frame, and The above edge is an electronic device corresponding to the relationship between objects included in the frame.

8. In Paragraph 7, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), Acquire the utilization rate of at least one of the nodes and edges corresponding to the acquisition of the response (202, 617, 1040) to the above query (201, 1110, 1120), and An electronic device that updates the knowledge graph (510, 601) based on the above-mentioned usage rate.

9. In any one of paragraphs 1 through 8, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), Based on the similarities between the above queries (201, 1110, 1120) and the multiple queries corresponding to the multiple personas (605) stored in the database, an average value of similarity for each of the multiple personas (605) is obtained, and An electronic device that identifies the persona (605) having the largest value among the above-mentioned average values ​​as the persona (605) corresponding to the above-mentioned query (201, 1110, 1120).

10. In any one of paragraphs 1 through 9, When the above instructions are executed individually or collectively by the at least one processor (120), the electronic device (100), Identify the similarity between the above query (201, 1110, 1120) and multiple queries in the database, and If there is a query (201, 1110, 1120) among the above multiple queries that has a similarity level greater than or equal to a threshold, a response (202, 617, 1040) corresponding to the query (201, 1110, 1120) that has a similarity level greater than or equal to the threshold is obtained from the database and the response (202, 617, 1040) is provided. An electronic device that, if there is no query (201, 1110, 1120) among the above multiple queries whose similarity is greater than or equal to the threshold value, obtains a response (202, 617, 1040) corresponding to the query (201, 1110, 1120) based on the obtained persona (605).

11. A method for controlling an electronic device (100), When receiving user input corresponding to a query (201, 1110, 1120) related to content, a step of obtaining a persona (605) corresponding to the query (201, 1110, 1120) among a plurality of personas (605) corresponding to the content based on a database; and The method includes the step of providing a response (202, 617, 1040) corresponding to the query (201, 1110, 1120) based on the acquired persona (605); The above plurality of personas (605) are a control method based on a knowledge graph (510, 601) corresponding to the content and the database.

12. In Paragraph 11, The method further includes the step of obtaining the plurality of personas (605) corresponding to the content through an artificial intelligence model based on the knowledge graph (510, 601) corresponding to the content, the database, and the first prompt (603). The above first prompt (603) is, A control method comprising a prompt corresponding to a request to predict attributes related to a user corresponding to the content for a plurality of queries of the database based on a knowledge graph (510, 601) corresponding to the content.

13. In Paragraph 12, A step of obtaining evaluation information corresponding to an expected query (607) corresponding to a persona for each of the plurality of personas (605), an expected response (609) to the expected query (607), and feedback (611) to the expected response (609); and A control method further comprising the step of updating the plurality of personas through an artificial intelligence model updated based on the above evaluation information.

14. In Paragraph 13, The step of updating the above multiple personas (605) is, A step of obtaining the predicted query (607) through an artificial intelligence model based on the persona (605), the knowledge graph (510, 601) corresponding to the content, and the second prompt (606); A step of obtaining the predicted response (609) through the artificial intelligence model based on the persona (605), the knowledge graph (510, 601) corresponding to the content, the predicted query (607), and the third prompt (608); and The method includes the step of obtaining the feedback (611) through the second artificial intelligence model 604() based on the persona (605), the knowledge graph (510, 601) corresponding to the content, the expected query (607), the expected response (609), and the fourth prompt (610); The second prompt (606) includes a prompt that corresponds to an expected query (607) request from a user corresponding to the persona (605) based on a knowledge graph (510, 601) corresponding to the content, and The third prompt (608) includes a prompt that corresponds to a request for an expected response (609) to an expected query (607) of a user corresponding to the persona (605) based on a knowledge graph (510, 601) corresponding to the content, and A control method comprising the above-mentioned fourth prompt (610), which includes a prompt that responds to a request for feedback (611) on the expected response (609) of a user corresponding to the persona (605) based on a knowledge graph (510, 610) corresponding to the content.

15. In Paragraph 13 or 14, The step of updating the above artificial intelligence model is, A step of obtaining a first score (614) through an evaluation model based on a prompt corresponding to a score request related to the quality of the above-mentioned expected query (607) and expected response (609); and A control method comprising the step of updating the artificial intelligence model based on the first score (614).

Images