System and method for providing content based on knowledge graph

Abstract

Provided is an artificial intelligence (AI) system using a machine learning algorithm and an application of the AI ​​system. A device for providing content based on a knowledge graph includes: a processor configured to execute instructions to: obtain context information related to the device; obtain a first device knowledge graph of the user of the device by inputting the obtained context information to a first AI model for determining relationships between entities related to the user of the device; request a server knowledge graph generated by the server from a server; receive the server knowledge graph; obtain a second device knowledge graph of the user by inputting the obtained first device knowledge graph and the received server knowledge graph to a second AI model for extending the first device knowledge graph; and provide content based on the obtained second device knowledge graph.

Description

Technical Field

[0001] The present disclosure relates to a system and method for providing content based on a knowledge graph, and more particularly, to a system and method for providing content based on a knowledge graph taking user privacy into consideration. Background Art

[0002] Unlike existing rule-based intelligent systems, an artificial intelligence (AI) system is a computer system that is configured to become smarter by training itself and making decisions autonomously. As AI systems are increasingly used, their recognition rates improve and they understand user preferences more accurately, rule-based intelligent systems are gradually being replaced by deep learning-based AI systems.

[0003] AI technologies include machine learning (e.g., deep learning) and element technology using machine learning. Machine learning is an algorithm technology that self-classifies / learns input data features, while element technology is a technology that uses machine learning algorithms such as deep learning and includes technical fields such as language understanding, visual understanding, inference / prediction, knowledge representation, and motion control.

[0004] The various fields where AI technology is applied are as follows. Language understanding is a technology for recognizing and applying / processing human language / characters, and includes natural language processing, machine translation, dialogue systems, question answering, and speech recognition / synthesis. Visual understanding is a technology for recognizing and processing objects like the human visual system, and includes object recognition, object tracking, image search, person recognition, scene understanding, spatial understanding, and image enhancement. Inference / prediction is a technology for judging information and logically inferring and predicting it, and includes knowledge / probability-based argumentation, optimization prediction, preference-based planning, and recommendation. Knowledge representation is an automated technology for incorporating human experience information into knowledge data, and includes knowledge establishment (e.g., data generation / classification) and knowledge management (e.g., data utilization). Motion control is a technology for controlling the automatic driving and robotic motion of an autonomous vehicle, and includes movement control (e.g., navigation, collision avoidance, or driving) and manipulation control (e.g., behavior control).

[0005] Furthermore, there is a demand for AI technology for protecting user privacy and effectively expanding knowledge graphs related to users. Summary of the invention

[0006] Systems and methods are provided for obtaining and using a knowledge graph associated with a user by using multiple artificial intelligence (AI) models.

[0007] Systems and methods for protecting individual privacy and obtaining and using knowledge graphs related to users are also provided.

[0008] Furthermore, a system and method are provided for effectively expanding a knowledge graph related to a user and providing recommended content to the user through collaboration between a device and a server.

[0009] Additional aspects will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the presented embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above and other aspects, features and advantages of certain embodiments of the present disclosure will become more apparent from the following description taken in conjunction with the accompanying drawings, in which:

[0011] Figure 1 is an illustration of a system for providing content based on a knowledge graph about a user according to an embodiment;

[0012] Figure 2 is a flowchart of a method for providing content by using a knowledge graph based on an artificial intelligence (AI) model, performed by a device according to an embodiment;

[0013] Figure 3 is a flowchart of a method for generating a first device knowledge graph performed by a device according to an embodiment;

[0014] Figure 4 is a flow chart of a method for setting a category and a privacy level to receive customized content performed by a device according to an embodiment;

[0015] Figure 5 is a flowchart of a method performed by a device according to an embodiment of the invention for extending a first device knowledge graph by using a server knowledge graph corresponding to a specific category;

[0016] Figure 6 is a flowchart of a method performed by a device for determining content to be provided to a user by using device operation information and a second device knowledge graph according to an embodiment;

[0017] Figure 7 is an illustration of a method performed by a system for providing customized content to a user according to an embodiment;

[0018] Figure 8 is a diagram showing an example in which a device according to an embodiment generates a first device knowledge graph by using text indicating a context related to the device;

[0019] Fig. 9 is a diagram showing an example of a device selecting a category to receive customized content according to an embodiment;

[0020] Fig.10 is a diagram showing an example of a device selecting a privacy level for a category according to an embodiment;

[0021] Fig.11 is a diagram illustrating a first device knowledge graph and a second device knowledge graph according to an embodiment;

[0022] Fig.12 is a diagram showing an example in which a device according to an embodiment generates a first device knowledge graph by using a first AI model;

[0023] Fig.13 is a diagram showing an example in which a device according to an embodiment generates a first device knowledge graph and a second device knowledge graph by using a first AI model and a second AI model;

[0024] Fig.14 is a diagram showing an example in which the device according to the embodiment determines recommended content by using a third AI model;

[0025] Fig.15 is a diagram showing an example in which a device according to an embodiment generates a second device knowledge graph by using a fourth AI model;

[0026] Fig.16 is a diagram showing an example in which the device according to the embodiment determines recommended content by using a fifth AI model;

[0027] Fig.17 is a block diagram of an apparatus according to an embodiment;

[0028] Fig.18 is a block diagram of an apparatus according to an embodiment; and

[0029] Fig.19 is a block diagram of a server according to an embodiment. DETAILED DESCRIPTION

[0030] According to aspects of the present disclosure, a device for providing content based on a knowledge graph includes: a device for providing content based on a knowledge graph, the device including: a communication interface; a memory storing one or more instructions; and a processor configured to execute one or more instructions to: obtain context information related to the device; obtain a first device knowledge graph of the user of the device by inputting the obtained context information into a first artificial intelligence (AI) model for determining relationships between entities related to the user of the device; control a request from a server for a server knowledge graph generated by a server; control receiving a server knowledge graph from the server; obtain a second device knowledge graph of the user by inputting the obtained first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; and provide content based on the obtained second device knowledge graph.

[0031] Each of the first AI model and the second AI model may be an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

[0032] The server knowledge graph may be generated by the server based on big data provided to the server by the device and at least one other device.

[0033] The processor may also be configured to execute one or more instructions to process the context information into text indicating sequential operations, and input the text into the first AI model.

[0034] The processor may also be configured to execute one or more instructions to determine a privacy level of the first device knowledge graph, and input the determined privacy level into the first AI model; and part of the data in the first device knowledge graph output from the first AI model may include data extracted according to the privacy level.

[0035] The processor may also be configured to execute one or more instructions to input a category into the first AI model; and a first device knowledge graph corresponding to the input category may be output from the first AI model.

[0036] The first AI model may output a first device knowledge graph through at least one of the following: entity extraction, entity analysis and extraction, and relationship extraction.

[0037] The processor may also be configured to execute one or more instructions to: control sending information about a user profile of a user to a server; and control receiving a server knowledge graph related to the user profile from the server.

[0038] The processor may also be configured to execute one or more instructions to control sending a specific category to a server, and receiving a server knowledge graph corresponding to the specific category from the server.

[0039] The processor can also be configured to execute one or more instructions to: determine the content to be recommended to the user by inputting the operation information of the device and the knowledge graph of the second device to the third AI model; and control the request to the content providing server for the determined content to be recommended.

[0040] According to another aspect of the present disclosure, a method for providing content based on a knowledge graph performed by a device includes: obtaining context information related to the device; obtaining a first device knowledge graph of the user of the device by inputting the obtained context information into a first artificial intelligence (AI) model for determining relationships between entities related to the user of the device; requesting a server knowledge graph generated by a server from a server; receiving the server knowledge graph from the server; obtaining a second device knowledge graph of the user by inputting the obtained first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; and providing content based on the obtained second device knowledge graph.

[0041] Each of the first AI model and the second AI model may be an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

[0042] The server knowledge graph may be generated by the server based on big data provided to the server by the device and at least one other device.

[0043] The method may also include processing the context information into text indicating sequential operations, wherein obtaining the first device knowledge graph may include inputting the text into the first AI model.

[0044] The method may also include determining a privacy level of the first device knowledge graph, wherein obtaining the first device knowledge graph may include inputting the privacy level into the first AI model, and wherein portion of data in the first device knowledge graph output from the first AI model may include data extracted according to the privacy level.

[0045] Obtaining the first device knowledge graph may include inputting a category into the first AI model; and outputting the first device knowledge graph corresponding to the input category from the first AI model.

[0046] The first AI model may output a first device knowledge graph through at least one of the following: entity extraction, entity analysis and extraction, and relationship extraction.

[0047] Requesting the server knowledge graph from the server may include sending information about a user profile of the user to the server; and receiving the server knowledge graph may include receiving the server knowledge graph related to the user profile from the server.

[0048] Requesting the server knowledge graph from the server may include sending a specific category to the server; and receiving the server knowledge graph may include receiving the server knowledge graph corresponding to the specific category from the server.

[0049] According to another aspect of the present disclosure, a non-transitory computer-readable recording medium has recorded thereon a program executable by at least one processor to perform: obtaining context information related to a device; obtaining a first device knowledge graph of a user of the device by inputting the obtained context information into a first artificial intelligence (AI) model for determining relationships between entities related to a user of the device; controlling a request from a server to a server knowledge graph generated by a server; controlling reception of a server knowledge graph from a server; obtaining a second device knowledge graph of the user by inputting the obtained first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; and providing content based on the obtained second device knowledge graph.

[0050] Each of the first AI model and the second AI model may be an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

[0051] The server knowledge graph may be generated by the server based on big data provided to the server by the device and at least one other device.

[0052] The program can be executed by at least one processor to further perform processing the context information into text indicating sequential operations; and obtaining the first device knowledge graph can include inputting the text into the first AI model.

[0053] A program can be executed by at least one processor to further determine the privacy level of the first device knowledge graph; obtaining the first device knowledge graph can include inputting the privacy level into the first AI model; and part of the data in the first device knowledge graph output from the first AI model can include data extracted according to the privacy level.

[0054] Obtaining the first device knowledge graph may include inputting a category into the first AI model; and outputting the first device knowledge graph corresponding to the input category from the first AI model.

[0055] According to another aspect of the present disclosure, a device for providing content based on a knowledge graph includes: a memory storing one or more instructions; and a processor configured to execute one or more instructions to: obtain a first device knowledge graph of a user of the device by inputting context information related to the device into a first artificial intelligence (AI) model for determining relationships between entities related to a user of the device; control receiving a server knowledge graph from a server; obtain a second device knowledge graph of the user by inputting the obtained first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; and obtain content based on the obtained second device knowledge graph.

[0056] Each of the first AI model and the second AI model may be an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

[0057] The server knowledge graph may be generated by the server based on big data provided to the server by the device and at least one other device.

[0058] The processor may also be configured to execute one or more instructions to process the context information into text indicating sequential operations, and input the text into the first AI model.

[0059] The processor may also be configured to execute one or more instructions to determine a privacy level of the first device knowledge graph, and input the determined privacy level into the first AI model; and part of the data in the first device knowledge graph output from the first AI model may include data extracted according to the privacy level.

[0060] Invention Mode

[0061] Hereinafter, the embodiments of the present disclosure will be described in detail to fully convey the scope of the present disclosure and enable those of ordinary skill in the art to easily implement and practice the present disclosure. The embodiments of the present disclosure can be implemented in various forms, and the present disclosure is not limited to the specific embodiments described in detail below. In addition, the same reference numerals refer to the same elements.

[0062] Throughout the disclosure, expressions such as "at least one of a, b, or c" and "at least one of a, b, and c" mean only a, only b, only c, both a and b, both a and c, both b and c, all of a, b, and c, or variations thereof.

[0063] Throughout the specification, it is understood that when it is mentioned that an element is "connected" to another element, the element may be "directly connected" to the other element or "electrically connected" to the other element using an intermediate element therebetween. It is also understood that when a part "includes" or "comprises" an element, unless otherwise defined, the part may further include other elements rather than excluding other elements.

[0064] In the specification, the device knowledge graph may be a knowledge graph generated in a user device. The device knowledge graph may be generated by reflecting user privacy information collected in the user device.

[0065] In addition, the server knowledge graph may be a knowledge graph generated in the server. The server knowledge graph may be generated by the server based on big data collected from various devices.

[0066] In addition, the artificial intelligence (AI) model may be an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

[0067] In addition, content can be digital information provided through a wired / wireless communication network. Examples of content may include, but are not limited to: advertising content, dynamic image content (e.g., TV programs, video on demand (VOD), user-created content (UCC), music videos, or YOUTUBE videos), still image content (e.g., photos or pictures), text content (e.g., electronic books including poems and novels, letters, or work documents), music content (e.g., music, performances, or radio broadcasts), web pages, and application execution information.

[0068] Embodiments will now be described more fully with reference to the accompanying drawings.

[0069] Figure 1 is an illustration of a system for providing content based on a knowledge graph about a user, according to an embodiment.

[0070] refer to Figure 1 , a system for providing content based on a knowledge graph about a user may include a device 1000 , a server 2000 , and a content providing server 3000 .

[0071] The device 1000 may collect context information about the device 1000, and may generate a device knowledge graph by using the collected context information and at least one AI model. In addition, the device 1000 may receive a server knowledge graph generated by the server 2000 from the server 2000, and may expand or enlarge the device knowledge graph by using the received server knowledge graph and at least one AI model. The device knowledge graph generated and expanded by the device 1000 may include information related to user privacy, and the device knowledge graph including privacy information may be used and managed in the device 1000. In addition, the device 1000 may determine content to be recommended to the user by using the expanded device knowledge graph, and may receive the determined recommended content from the content providing server 3000.

[0072] The server 2000 may generate a server knowledge graph based on the big data received from the various devices 1000. The big data used to generate the server knowledge graph may be data that does not include information related to user privacy. For example, the big data may include information extracted from user privacy information.

[0073] The content providing server 3000 may be a server that manages content and provides the content to the device 1000. The content providing server 3000 may be a server that provides, for example, movies, games, advertisements, and the like.

[0074] Examples of device 1000 may include, but are not limited to, smart phones, tablet personal computers (PCs), PCs, smart televisions (TVs), mobile phones, personal digital assistants (PDAs), laptop computers, media players, micro servers, global positioning system (GPS) devices, electronic book terminals, digital broadcast terminals, navigation systems, kiosks, MP3 players, digital cameras, home appliances, portable media devices, and any other mobile or non-mobile computing devices. In addition, examples of device 1000 may include wearable devices with communication functions and data processing functions, such as watches, glasses, hair bands, or rings. However, it is understood that the present disclosure is not limited thereto, and device 1000 may include any type of device that sends data to / receives data from server 2000 and content providing server 3000 through a network.

[0075] Examples of networks may include local area networks (LANs), wide area networks (WANs), value-added networks (VANs), mobile radio communication networks, satellite communication networks, and combinations thereof. In addition, a network broadly refers to a data communication network for smooth communication between network components, and examples of networks may include wired Internet, wireless Internet, and mobile wireless communication networks.

[0076] Examples of wireless communications may include, but are not limited to, Wi-Fi, Bluetooth, Bluetooth Low Energy, Zigbee, Wi-Fi Direct (WFD), Ultra-Wideband (UWB), Infrared Data Association (IrDA), and Near Field Communication (NFC).

[0077] Figure 2 is a flowchart of a method for providing content by using an AI model-based knowledge graph, performed by the device 1000 according to an embodiment.

[0078] refer to Figure 2 In operation S200, the device 1000 may obtain context information related to the device 1000. The device 1000 may monitor operations of applications in the device 1000, and may collect information related to the device 1000 from a sensor of the device 1000 and another device connected to the device 1000.

[0079] The context information may include, but is not limited to, at least one of the following: surrounding environment information of the device 1000, state information of the device 1000, state information of the user using the device 1000, device usage history information of the user, or schedule information of the user. The surrounding environment information of the device 1000 refers to the environment information within a certain radius from the device 1000, and may include, but is not limited to, at least one of the following: weather information, temperature information, humidity information, illumination information, noise information, and sound information. The state information of the device 1000 may include, but is not limited to, at least one of the following: information about the mode of the device 1000 (e.g., sound mode, vibration mode, silent mode, power saving mode, cutoff mode, multi-window mode, or automatic rotation mode), location information of the device 1000, time information, activation information of the communication module (e.g., Wi-Fi is on, Bluetooth is off, GPS is on, or NFC is off), network connection state information of the device 1000, and information about the application executed by the device 1000 (e.g., application identification information, application type, application usage time, or application usage cycle). The user's state information refers to information about the user's movement or life pattern, and may include, but is not limited to, at least one of the following: information about the user's walking state, motion state, driving state, sleeping state, and emotional state. The user's device usage history information refers to information about events of the user's use of the device 1000, and may include, but is not limited to, information about at least one of the following: execution of an application, functions executed by an application, the user's phone conversations, and the user's text messages.

[0080] In operation S210, the device 1000 may obtain a first device knowledge graph by inputting context information to a first AI model. For example, the device 1000 may process the context information and may obtain the first device knowledge graph by inputting the processed context information to the first AI model.

[0081] The first device knowledge graph may be a knowledge graph generated based on a context related to the user and / or device 1000, and may be generated by reflecting information related to user privacy. The first AI model may be an AI model that generates and updates a knowledge graph based on the context of the user and / or device 1000. The AI ​​model may be trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm. The first AI model may be used to extract entities in context information and infer relationships between the extracted entities.

[0082] The device 1000 may generate a first device knowledge graph for each of the specific categories. The device 1000 may generate the first device knowledge graph according to a privacy level to protect the user's personal information. The privacy level may indicate the degree to which the user's personal information is protected, and according to the privacy level, the degree to which data related to the user's privacy in the data in the first device knowledge graph is extracted may be determined.

[0083] In operation S220, the device 1000 may request a server knowledge graph from the server 2000. For example, the device 1000 may transmit at least one of information related to a user profile and a category identification value selected by a user to the server 2000, and may request the server knowledge graph from the server 2000.

[0084] The server knowledge graph is a knowledge graph generated by the server 2000, and the server knowledge graph may be a knowledge graph generated based on big data received from various users and various devices. The big data used to generate the server knowledge graph may include contextual information related to various environments, and may be data that does not include information related to individual privacy. In addition, the server knowledge graph may be generated by a specific AI model using big data as an input value, and may be generated, for example, for each category and user's profile features.

[0085] In operation S230, the device 1000 may receive a server knowledge graph from the server 2000. The device 1000 may receive a server knowledge graph related to the user profile. In addition, the device 1000 may receive a server knowledge graph related to a category selected by the user.

[0086] In operation S240, the device 1000 may obtain a second device knowledge graph by inputting the first device knowledge graph and the server knowledge graph into the second AI model. The second device knowledge graph may be a knowledge graph extended from the first device knowledge graph. The second AI model may be an AI model for extending and updating the first device knowledge graph. The second AI model may be trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm. By analyzing and integrating the first device knowledge graph and the server knowledge graph, the second AI model may be used to extend the first device knowledge graph.

[0087] In operation S250, the device 1000 may provide content based on the second device knowledge graph. The device 1000 may identify the current context of the device 1000, and may determine content suitable for the current context as recommended content by using the second device knowledge graph. In addition, the device 1000 may request the determined recommended content from the content providing server 3000, and may receive the requested recommended content from the content providing server 3000.

[0088] Content can be digital information provided through a wired / wireless communication network. Examples of content may include, but are not limited to: advertising content, dynamic image content (e.g., TV programs, VOD, UCC, music videos, or YOUTUBE videos), still image content (e.g., photos or pictures), text content (e.g., electronic books including poems and novels, letters, or work documents), music content (e.g., music, performances, or radio broadcasts), web pages, game content, application execution information, etc.

[0089] Figure 3 It is a flowchart of a method for generating a first device knowledge graph performed by device 1000 according to an embodiment.

[0090] refer to Figure 3 , in operation S300, the device 1000 may obtain context information related to the device 1000. The context information may include, but is not limited to, at least one of the following: surrounding environment information of the device 1000, state information of the device 1000, state information of a user using the device 1000, device usage history information of the user, or schedule information of the user. By monitoring the operation of the application installed in the device 1000, the device 1000 may collect, for example, at least one of the following: an operation performed by the application, data input to the application, and data output by the application. In addition, by obtaining sensing data sensed by a sensor in the device 1000, the device 1000 may obtain information about the surrounding environment of the device 1000 and information about the state of the device 1000.

[0091] In operation S310, the device 1000 may process the context information into text indicating a sequential operation. Based on the context information, the device 1000 may generate a sentence indicating a context related to the device 1000 and / or the user. For example, the device 1000 may generate text such as "the user downloaded a travel application," "the user searched for Okinawa," and "the user purchased a camera at nine o'clock on Saturday morning."

[0092] In operation S320, the device 1000 may set a privacy level for the device knowledge graph. In order to extract data related to user privacy from data in the device knowledge graph, the device 1000 may set a privacy level for the device knowledge graph. For example, the device 1000 may display a graphical user interface (GUI) for setting a privacy level for the device knowledge graph, and may set the privacy level for the device knowledge graph based on user input through the GUI.

[0093] In operation S330, the device 1000 may apply the text in operation S310 and the privacy level set in operation S320 to the first AI model. The device 1000 may extract data related to individual privacy by inputting the text and the privacy level to the first AI model, and first user knowledge data based on the extracted data may be output from the first AI model. In addition, according to the privacy level, the degree to which data related to user privacy is extracted from the data in the first device knowledge graph may be determined according to the privacy level.

[0094] For example, when the text "the user purchased a Big Mac from McDonald's" and a specific privacy level value "1" are input to the first AI model, the first device knowledge graph based on the extracted data "the user purchased a hamburger from McDonald's" can be output. In addition, for example, when the text "the user purchased a Big Mac from McDonald's" and a specific privacy level value "2" are input to the first AI model, the first device knowledge graph based on the extracted data "the user purchased a hamburger from the hamburger shop" can be output.

[0095] Figure 4 is a flowchart of a method of setting a category and a privacy level to receive customized content, performed by the device 1000 according to an embodiment.

[0096] refer to Figure 4 In operation S400, the device 1000 may select at least one category in the category list for receiving user customized content. The device 1000 may display a specific category list on the screen of the device 1000, and may receive a user input selecting a specific category in the specific category list. The categories may be divided according to at least one of the type of content, the field of content, or the size of content. For example, the categories may be divided into, but not limited to, movies, sports, games, and advertisements.

[0097] In operation S410, the device 1000 may set a privacy level for each selected category. The device 1000 may display a GUI for inputting a privacy level on the screen of the device 1000, and may set a privacy level for each category based on a user input through the GUI. Accordingly, the privacy level may be set differently for each category, and the device 1000 may generate a device knowledge graph for each category according to the differently set privacy levels.

[0098] Figure 5 is a flowchart of a method performed by the device 1000 to expand a first device knowledge graph by using a server knowledge graph corresponding to a specific category according to an embodiment.

[0099] refer to Figure 5 , in operation S500, the device 1000 may select a category of the knowledge graph, and then in operation S510, the device 1000 may request a server knowledge graph corresponding to the selected category from the server 2000. By using the big data received from various devices, the server 2000 may generate and store a server knowledge graph for each specific category. In this case, by inputting the big data and the category value into a specific AI model, the server 2000 may generate a server knowledge graph for each category.

[0100] In operation S520, the device 1000 may receive a server knowledge graph from the server 2000. For example, the device 1000 may receive a server knowledge graph corresponding to the selected category from the server 2000.

[0101] In operation S530, the device 1000 may apply the first device knowledge graph, the server knowledge graph, and the privacy level value to the second AI model. The device 1000 may extract the first device knowledge graph corresponding to the specific category from the memory of the device 1000, and may input the extracted first device knowledge graph and the server knowledge graph corresponding to the specific category into the second AI model. In addition, the device 1000 may input the privacy level value set for the first device knowledge graph corresponding to the specific category into the second AI model. Accordingly, the device 1000 may obtain the second device knowledge graph extended from the first device knowledge graph from the second AI model.

[0102] Figure 6 is a flowchart of a method performed by the device 1000 to determine content to be provided to a user by using operation information of the device 1000 and a second device knowledge graph according to an embodiment.

[0103] refer to Figure 6In operation S600, the device 1000 may obtain a second device knowledge graph. The device 1000 may obtain the second device knowledge graph extended from the first device knowledge graph by using the server knowledge graph.

[0104] In operation S610, the device 1000 may obtain in real time the operation information of the device 1000. The device 1000 may obtain a context related to the device 1000 as the real-time operation information of the device 1000, such as at least one of the following: an application being executed in the device 1000, a function of the application being executed, a surrounding environment of the device 1000, and schedule information stored in the device 1000.

[0105] In operation S620, the device 1000 may determine the recommended content to be provided to the user by applying the second device knowledge graph and the operation information of the device 1000 to the third AI model. The third AI model may be an AI model for determining the recommended content to be recommended to the user based on the operation information of the device 1000 and / or the user and the second device knowledge graph. The third AI model may be trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm. According to the context of the device 1000 and the second device knowledge graph based on the real-time operation information of the device 1000, the third AI model may be used to recommend content in areas of interest to the user.

[0106] In operation S630, the device 1000 may provide the determined recommended content to the user. For example, the device 1000 may identify the content providing server 3000 that provides the determined recommended content, and may request the recommended content from the content providing server 3000. In addition, the device 1000 may receive the recommended content from the content providing server 3000. Accordingly, without providing information related to user privacy to the content providing server 3000, the device 1000 may provide the user with the recommended content suitable for the user. In addition, examples of recommended content may include, but are not limited to: advertising content, dynamic image content (e.g., TV programs, VOD, UCC, music videos, or YOUTUBE videos), still image content (e.g., photos or pictures), text content (e.g., electronic books, letters, or work documents including poems and novels), music content (e.g., music, performances, or radio broadcasts), web pages, game content, and application execution information.

[0107] Figure 7 is an illustration of a method performed by a system for providing customized content to a user according to an embodiment.

[0108] refer to Figure 7In operation S700, the device 1000 may obtain context information. The device 1000 may collect context information related to the device 1000 and / or a user, and may process the collected context information into a specific text.

[0109] In operation S705, the device 1000 may obtain a first device knowledge graph by inputting context information into the first AI model. For example, the device 1000 may input text processed from the context information into the first AI model. In addition, the device 1000 may input a privacy level and a category value into the first AI model. The device 1000 may obtain a first device knowledge graph from the first AI model.

[0110] In operation S710, the server 2000 may generate a server knowledge graph. The server 2000 may collect various big data from various devices, and may generate a server knowledge graph by using the collected big data. The big data collected by the server 2000 may be data extracted from individual privacy information.

[0111] In operation S715, the device 1000 may request a server knowledge graph from the server 2000. For example, the device 1000 may send information about a user profile and / or a category value to the server 2000, and may request a server knowledge graph from the server 2000. In this case, information about user privacy among the information about the user profile may be the extracted information.

[0112] In operation S720, the device 1000 may receive a server knowledge graph from the server 2000. Considering the user profile and / or the category, the server 2000 may extract a server knowledge graph to be provided to the device 1000 from a database (DB), and may provide the extracted server knowledge graph to the device 1000. In this case, the server knowledge graph may be generated according to the user profile and / or the category and may be stored in the DB of the server 2000.

[0113] In operation S725, the device 1000 may obtain a second device knowledge graph by inputting the first device knowledge graph and the server knowledge graph into the second AI model. The second AI model may output the second device knowledge graph by extending the first device knowledge graph using the server knowledge graph.

[0114] In operation S730, the device 1000 may remove the privacy information in the second device knowledge graph. The device 1000 may extract data related to user privacy from the data in the second device knowledge graph.

[0115] In operation S735, the device 1000 may provide the server 2000 with the device knowledge graph from which the privacy information is removed.

[0116] In operation S740, the server 2000 may update the server knowledge graph by using the device knowledge graph from which the privacy information has been removed. The server 2000 may update the server knowledge graph by inputting the server knowledge graph generated in operation S710 and the device knowledge graph received in operation S735 into a specific AI model. The updated server knowledge graph may then be used by the device 1000 to generate a second device knowledge graph.

[0117] In operation S745, the device 1000 may determine the recommended content to be provided to the user by applying the second device knowledge graph to the third AI model. By inputting the operation information of the device 1000 and the second device knowledge graph to the third AI model, the device 1000 may receive information about the recommended content from the third AI model. The information about the recommended content may include, but is not limited to, at least one of the following: an identification value of the recommended content, a type of the recommended content, or a field of the recommended content.

[0118] In operation S750, the device 1000 may request recommended content from the content providing server 3000, and then in operation S755, the device 1000 may receive the recommended content from the content providing server 3000.

[0119] Figure 8 is a diagram showing an example in which the device 1000 according to an embodiment generates a first device knowledge graph by using text indicating a context related to the device.

[0120] refer to Figure 8 , the device 1000 may obtain context information about device usage records and personal information stored in or input to the device 1000. In addition, the device 1000 may generate text such as “the user downloaded a travel application,” “the user searched for Okinawa,” and “the user purchased a camera.”

[0121] Device 1000 may input the generated text into the first AI model, and the first AI model may input the first device knowledge graph that defines entities and relationships between entities. For example, the entities of the first device knowledge graph may include "Me", "Okinawa", "Camera", and "Travel App". In this case, the relationship between the entity "Me" and the entity "Okinawa" may be "Search". In addition, in this example, the relationship between the entity "Me" and the entity "Camera" may be "Purchase". Similarly, the relationship between the entity "Me" and the entity "Travel App" may be "Download".

[0122] Fig. 9 is a diagram illustrating an example in which the device 1000 selects a category to receive customized content according to an embodiment.

[0123] refer to Fig. 9 In order for the user to receive only user customized content of a specific category, the device 1000 may display a category list for selecting a specific category on the display screen of the device 1000. For example, the device 1000 may display a guide paragraph for allowing the user to select a category, and may display category identification values ​​such as "movies", "sports", "games", and "advertisements", thereby enabling the user to select a specific category.

[0124] Fig.10 is a diagram illustrating an example in which the device 1000 selects a privacy level for a category according to an embodiment.

[0125] refer to Fig.10 , the device 1000 may display a GUI for allowing the user to set a privacy level for each category, and may set a privacy level for each category based on user input through the GUI. For example, based on user input through the GUI, the device 1000 may set the privacy level of the category "movies" to "2", may set the privacy level of the category "sports" to "1", may set the privacy level of the category "games" to "3", and may set the privacy level of the category "advertisements" to "4". Accordingly, a first device knowledge graph for the category "movies", a first device knowledge graph for the category "sports", a first device knowledge graph for the category "games", and a first device knowledge graph for the category "advertisements" may be generated at different privacy levels.

[0126] Fig.11 is a diagram illustrating a first device knowledge graph and a second device knowledge graph according to an embodiment.

[0127] refer to Fig.11 , the entities of the first device knowledge graph 110 may include "I", "Okinawa", "Camera", and "Travel Application". In addition, for example, the relationship between the entity "I" and the entity "Okinawa" may be "Search", the relationship between the entity "I" and the entity "Camera" may be "Purchase", and the relationship between the entity "I" and the entity "Travel Application" may be "Download".

[0128] By inputting the first device knowledge graph 110 and the server knowledge graph received from the server 2000 into the first AI model, the device 1000 can generate a second device knowledge graph 120. The second device knowledge graph 120 can be a knowledge graph extended from the first device knowledge graph 110. The entities in the first device knowledge graph 110 and the entities in the server knowledge graph can be mapped by specific criteria, and according to the specific criteria, the entities in the server knowledge graph and the entities in the first device knowledge graph 110 can be integrated together. For example, the second device knowledge graph 120 may include entities "restaurant" and "aquarium" extended from the entity "Okinawa". In addition, for example, the relationship between the entity "Okinawa" and the entity "restaurant" can be determined as "food", and the relationship between the entity "Okinawa" and the entity "aquarium" can be determined as "travel".

[0129] Fig.12 is a diagram showing an example in which the device 1000 according to an embodiment generates a first device knowledge graph by using a first AI model.

[0130] refer to Fig.12 , the device 1000 can generate structured data by collecting and preprocessing context information, and then generate a first device knowledge graph by using the structured data. The structured data can be, for example, text indicating sequential operations or sentences indicating context related to the device 1000 and / or the user.

[0131] The device 1000 may input structured data to the first AI model, and then by using the structured data as an input value, the first AI model may generate a first device graph by entity extraction, entity resolution and disambiguation (entity analysis and extraction), and relationship extraction. The first AI model may be implemented by using, for example, an ontology learning module, and may be implemented by using a network including a plurality of layers.

[0132] Fig.13 is a diagram illustrating an example in which the device 1000 generates a first device knowledge graph and a second device knowledge graph by using the first AI model 130 and the second AI model 132 according to an embodiment.

[0133] refer to Fig.13 , the first AI model 130 may receive context information, a privacy level, and a category, and may output a first device knowledge graph. The context information may be processed into text indicating a sequential operation and may be input to the first AI model 130. In addition, the first AI model 130 may output a first device knowledge graph corresponding to the input category. In addition, data in the first device knowledge graph may be extracted based on the input privacy level to protect user privacy.

[0134] The second AI model 132 may receive the first device knowledge graph and the server knowledge graph, and may output the second device knowledge graph. The server knowledge graph may be generated by the server based on big data from various devices. In addition, the second AI model 132 may output the second device knowledge graph by extending the first device knowledge graph using the server knowledge graph.

[0135] Fig.14 is a diagram illustrating an example in which the device 1000 determines recommended content by using the third AI model 140 according to an embodiment.

[0136] refer to Fig.14 , the third AI model 140 can receive the operation information of the device 1000, the second device knowledge graph and the privacy setting information, and can determine the recommended content. The operation information of the device 1000 can be collected by the device 1000 in real time and can be input into the third AI model 140. In this case, the operation information of the device 1000 can be preprocessed into a preset format. Based on the privacy setting information, the third AI model 140 can determine the recommended content to be recommended to the user. The third AI model 140 can output but is not limited to the identification value of the recommended content. The third AI model 140 can output at least one of the following: recommended content type, category, or link information. In addition, the third AI model 140 can receive information about the category set by the user, and can output recommended content related to the category set by the user.

[0137] Fig.15 is a diagram illustrating an example in which the device 1000 generates a second device knowledge graph by using the fourth AI model 150 according to an embodiment.

[0138] refer to Fig.15 , the fourth AI model 150 can receive the first device knowledge graph, context information, privacy level, category and server knowledge graph, and can output the second device knowledge graph. The fourth AI model 150 can be an AI model for expanding and updating the first device knowledge graph. The fourth AI model 150 can be trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm or a classification algorithm. By analyzing and integrating the first device knowledge graph and the server knowledge graph, the fourth AI model 150 can be used to expand the first device knowledge graph.

[0139] Fig.16 is a diagram illustrating an example in which the device 1000 determines recommended content by using the fifth AI model 160 according to an embodiment.

[0140] refer to Fig.16, the fifth AI model 160 may receive the first device knowledge graph, context information, privacy level, category, and server knowledge graph, and may output recommended content. In this case, the fifth AI model 160 may be an AI model for expanding the first device knowledge graph and determining recommended content by using the server knowledge graph. The fifth AI model 160 may be trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

[0141] Fig.17 and Fig.18 is a block diagram of a device 1000 according to one or more embodiments.

[0142] like Fig.17 As shown, the device 1000 according to the embodiment may include a user input device 1100, an output device 1200, a controller 1300, and a communicator 1500. However, it is understood that the device 1000 may include more than Fig.17 More or fewer elements than shown.

[0143] For example, Fig.18 As shown, in addition to the user input 1100, the output 1200, the controller 1300 and the communicator 1500, the device 1000 according to an embodiment may further include a sensing unit 1400 (e.g., a sensor), an audio / video (A / V) input 1600 and a memory 1700.

[0144] The user input 1100 is a unit for a user to input data for controlling the device 1000. Examples of the user input 1100 may include, but are not limited to, at least one of a keyboard, a dome switch, a touch pad (e.g., a contact capacitance method, a pressure resistance film method, an infrared sensing method, a surface ultrasonic transmission method, an integral tension measurement method, or a piezoelectric effect method), a proximity sensor, a gesture sensor, a jog wheel, and a jug switch.

[0145] The user inputter 1100 may receive user input for providing recommended content based on the device knowledge graph.

[0146] The outputter 1200 may output an audio signal, a video signal, or a vibration signal, and may include a display 1210 , a sound outputter 1220 , and a vibration motor 1230 .

[0147] The display 1210 displays and outputs information processed by the device 1000. For example, the display 1210 may display a user interface for providing recommended content based on the device knowledge graph.

[0148] When the display 1210 and the touch pad have a layered structure to form a touch screen, the display 1210 may be used as both an input device and an output device.

[0149] The sound outputter 1220 outputs audio data received from the communicator 1500 or stored in the memory 1700. In addition, the sound outputter 1220 outputs a sound signal (e.g., a call signal reception sound, a message reception sound, or a notification sound) related to a function performed by the device 1000. The sound outputter 1220 may include a speaker or a buzzer.

[0150] The vibration motor 1230 may output a vibration signal. For example, the vibration motor 1230 may output a vibration signal corresponding to the output of audio data or video data (e.g., a call signal reception sound, a message reception sound, or a notification sound). In addition, when a touch is input to the touch screen, the vibration motor 1230 may output a vibration signal.

[0151] The controller 1300 (e.g., at least one processor) generally controls the overall operation of the device 1000. For example, by executing a program or instruction stored in the memory 1700, the controller 1300 may control the user input 1100, the output 1200, the sensing unit 1400, the communicator 1500, and the A / V input 1600. By controlling the user input 1100, the output 1200, the sensing unit 1400, the communicator 1500, and the A / V input 1600, the controller 1300 may control the operation of the device 1000.

[0152] Specifically, the controller 1300 may obtain context information related to the device 1000. The controller 1300 may monitor operations of applications in the device 1000, and may collect information related to the device 1000 from a sensor of the device 1000 and another device connected to the device 1000.

[0153] By inputting context information to the first AI model, the controller 1300 can obtain a first device knowledge graph. By processing the context information and inputting the processed context information to the first AI model, the controller 1300 can obtain a first device knowledge graph. The controller 1300 can generate a first device knowledge graph for each specific category. The controller 1300 can generate a first device knowledge graph according to a privacy level to protect the user's personal information.

[0154] The controller 1300 may request the server knowledge graph from the server 2000. The controller 1300 may request the server knowledge graph from the server 2000 by transmitting information related to the user profile and an identification value of a category selected by the user to the server 2000.

[0155] The controller 1300 may receive a server knowledge graph from the server 2000. The controller 1300 may receive a server knowledge graph related to a user profile. In addition, the controller 1300 may receive a server knowledge graph related to a category selected by a user.

[0156] By inputting the first device knowledge graph and the server knowledge graph into the second AI model, the controller 1300 can obtain the second device knowledge graph. The second device knowledge graph can be a knowledge graph extended from the first device knowledge graph.

[0157] The controller 1300 may provide content based on the second device knowledge graph. The controller 1300 may identify the current context of the device 1000, and may determine content suitable for the current context as recommended content by using the second device knowledge graph. In addition, the controller 1300 may request the determined recommended content from the content providing server 3000, and may receive the requested recommended content from the content providing server 3000.

[0158] The controller 1300 may process the context information into a text indicating a sequential operation. For example, the controller 1300 may generate a sentence indicating a context related to the device 1000 and / or the user based on the context information.

[0159] The controller 1300 can set the privacy level of the device knowledge graph. The controller 1300 can set the privacy level of the device knowledge graph so as to extract data related to user privacy from the data of the device knowledge graph.

[0160] The controller 1300 may apply the text and the privacy level to the first AI model. By inputting the text and the privacy level to the first AI model, the controller 1300 may extract data related to individual privacy, and may output first user knowledge data based on the extracted data from the first AI model.

[0161] The controller 1300 may select at least one category in the category list for receiving user customized content. The controller 1300 may display a specific category list on the screen of the device 1000, and may receive a user input selecting a specific category in the specific category list.

[0162] The controller 1300 may set a privacy level for each selected category. The controller 1300 may display a GUI for inputting a privacy level on the screen of the device 1000, and may set a privacy level for each category based on a user input through the GUI. Accordingly, the privacy level may be set differently for each category, and the controller 1300 may generate a device knowledge graph for each category according to the differently set privacy levels.

[0163] The controller 1300 may obtain in real time the operation information of the device 1000. The controller 1300 may obtain the context related to the device 1000 as the operation information of the device 1000 in real time, such as the application being executed in the device 1000, the function of the application being executed, the surrounding environment of the device 1000, and the schedule information stored in the device 1000.

[0164] By applying the second device knowledge graph and operation information of the device 1000 to the third AI model, the controller 1300 may determine recommended content to be provided to the user.

[0165] The controller 1300 may provide the determined recommended content to the user. The controller 1300 may identify the content providing server 3000 that provides the determined recommended content, and may request the recommended content from the identified content providing server 3000. In addition, the controller 1300 may receive the recommended content from the content providing server 3000. Accordingly, without providing information related to user privacy to the content providing server 3000, the controller 1300 may provide the user with recommended content suitable for the user.

[0166] The sensing unit 1400 may detect a state of the device 1000 or a state around the device 1000 and may transmit the detected information to the controller 1300 .

[0167] The sensing unit 1400 may include, but is not limited to, at least one of the following: a geomagnetic sensor 1410, an acceleration sensor 1420, a temperature / humidity sensor 1430, an infrared sensor 1440, a gyroscope sensor 1450, a positioning sensor (e.g., a global positioning system (GPS)) 1460, an air pressure sensor 1470, a proximity sensor 1480, or an RGB sensor (e.g., an illumination sensor) 1490.

[0168] The communicator 1500 may include one or more elements (e.g., interfaces, circuits, ports, etc.) through which the device 1000 communicates with the server 2000 and the content providing server 3000. For example, the communicator 1500 may include a short-range communicator 1510, a mobile communicator 1520, and a broadcast receiver 1530.

[0169] Examples of the short-range communicator 1510 may include, but are not limited to, a Bluetooth communicator, a Bluetooth low energy (BLE) communicator, a near field communicator, a WLAN (Wi-Fi) communicator, a ZigBee communicator, an infrared data association (IrDA) communicator, a Wi-Fi Direct (WFD) communicator, an ultra-wideband (UWB) communicator, and an Ant+ communicator.

[0170] The mobile communicator 1520 transmits / receives wireless signals to / from at least one of a base station, an external terminal, or a server through a mobile communication network. Examples of wireless signals may include voice call signals, video call signals, and any of various data segments transmitted / received according to text / multimedia messages.

[0171] The broadcast receiver 1530 receives a broadcast signal and / or broadcast-related information from the outside through a broadcast channel. Examples of the broadcast channel may include a satellite channel and a terrestrial channel. According to an embodiment, the device 1000 may not include the broadcast receiver 1530.

[0172] In addition, the communicator 1500 can transmit / receive information for providing recommended content based on the device knowledge graph to / from the server 2000 and the content providing server 3000 .

[0173] The A / V input device 1600 for inputting an audio signal or a video signal may include a camera 1610 and a microphone 1620. The camera 1610 may obtain an image frame such as a still image or a dynamic image by using an image sensor in a video call mode or an imaging mode. The image captured by the image sensor may be processed by the controller 1300 or an additional image processor.

[0174] The image frames processed by the camera 1610 may be stored in the memory 1700 or may be transmitted to the outside through the communicator 1500. Two or more cameras 1610 may be provided according to the configuration of the terminal.

[0175] Microphone 1620 receives external sound signals and processes the external sound signals into electronic voice data. For example, microphone 1620 may receive sound signals from an external device or a speaker. Microphone 1620 may use any of various noise removal algorithms to remove noise that occurs when (e.g., based on) receiving external sound signals.

[0176] The memory 1700 may store a program or instruction for processing and controlling the controller 1300 , and may store data input to or output from the device 1000 .

[0177] Memory 1700 may include at least one of the following types of storage media: flash memory type, hard disk type, multimedia card micro, card-type memory (for example, secure digital (SD) or XD memory), random access memory (RAM), static RAM (SRAM), read-only memory (ROM), electrically erasable programmable ROM (EEPROM), programmable FROM (PROM), magnetic storage, disk or optical disk.

[0178] The memory 1700 may store the first device knowledge graph and the second device knowledge graph. The memory 1700 may establish a database of the first device knowledge graph for each category and may store the first device knowledge graph. The memory 1700 may establish a database of the first device knowledge graph for each privacy level and may store the first device knowledge graph.

[0179] In addition, the memory 1700 may establish a database of the second device knowledge graph for each category and may store the second device knowledge graph. The memory 1700 may establish a database of the second device knowledge graph for each privacy level and may store the second device knowledge graph.

[0180] The programs stored in the memory 1700 may be classified into a plurality of modules according to the functions of the memory 1700, for example, a user interface (UI) module 1710, a touch screen module 1720, and a notification module 1730.

[0181] The UI module 1710 may provide a dedicated UI for each application or a GUI for interactive operation with the device 1000. The touch screen module 1720 may detect a touch gesture of a user on the touch screen, and may send information about the touch gesture to the controller 1300. The touch screen module 1720 according to an embodiment may recognize and analyze a touch code. The touch screen module 1720 may be configured as a separate hardware including a controller.

[0182] Various sensors may be located within or near the touch screen to detect a touch or near touch of the touch screen. An example of a sensor for detecting a touch of the touch screen may be a tactile sensor. A tactile sensor refers to a sensor that detects contact with a specific object to a degree that humans can feel or higher. The tactile sensor may detect any of a variety of information such as the roughness of the contact surface, the hardness of the contact object, or the temperature of the contact point.

[0183] In addition, an example of the sensor for detecting a touch of the touch screen may be a proximity sensor.

[0184] A proximity sensor refers to a sensor that detects an object approaching a specific detection surface or an adjacent object by using the intensity of an electromagnetic field or infrared ray without mechanical contact. Examples of a user's touch gesture may include tap, touch and hold, double tab, drag, pan, flick, drop, and slide.

[0185] The notification module 1730 may generate a signal for notifying an event occurring in the device 1000. The notification module 1730 may output the notification signal as a video signal through the display 1210; may output the notification signal as an audio signal through the sound outputter 1220; or may output the notification signal as a vibration signal through the vibration motor 1230.

[0186] Fig.19 is a block diagram of a server 2000 according to an embodiment.

[0187] refer to Fig.19 , the server 2000 according to the embodiment may include a communicator 2100 , a storage 2200 , and a processor 2300 .

[0188] The communicator 2100 may include one or more elements for communicating with the device 1000 and the content providing server 3000. For example, the communicator 2100 may include a short-range communicator, a mobile communicator, and a broadcast receiver. In addition, the communicator 2100 may send information for providing recommended content based on the device knowledge graph to / from the device 1000 and the content providing server 3000.

[0189] The storage 2200 may store programs or instructions for processing and controlling the processor 2300, and may store data input to or output from the server 2000. In addition, the storage 2200 may store a server knowledge graph. The storage 2200 may establish a database of a server knowledge graph for each category, and may store the server knowledge graph.

[0190] The processor 2300 (e.g., at least one processor) may generally control the overall operation of the server 2000. For example, the processor 2300 may control the communicator 2100 and the storage 2200 by executing a program or instruction stored in the storage 2200. The processor 2300 may control the operation of the server 2000 by controlling the communicator 2100 and the storage 2200.

[0191] Specifically, the processor 2300 may generate a server knowledge graph. The processor 2300 may collect various big data from various devices, and may generate a server knowledge graph by using the collected big data. The big data collected by the processor 2300 may be data extracted from individual privacy information.

[0192] The processor 2300 may receive a request for a server knowledge graph from the device 1000. The processor 2300 may receive information about a user profile and / or a category value from the device 1000. In this case, the information about user privacy in the information about the user profile may be the extracted information.

[0193] The processor 2300 may provide the server knowledge graph to the device 1000. Considering the user profile and / or the category, the processor 2300 may extract the server knowledge graph to be provided to the device 1000 from the storage 2200, and may provide the extracted server knowledge graph to the device 1000. In this case, the server knowledge graph may be generated according to the user profile and the category, and may be stored in the storage 2200.

[0194] The processor 2300 may receive a device knowledge graph from which privacy information has been removed from the device 1000. The processor 2300 may update the server knowledge graph by using the device knowledge graph from which privacy information has been removed. The processor 2300 may update the server knowledge graph by inputting the server knowledge graph and the device knowledge graph into a specific AI model. The updated server knowledge graph may then be used by the device 1000 to generate a second device knowledge graph.

[0195] The AI-related functions according to the embodiment are performed by a processor and a memory. The processor may include at least one processor. In this case, at least one processor may include a general-purpose processor such as a central processing unit (CPU), an access point (AP), or an electronic signal processor (DSP); a graphics processor such as a graphics processing unit (GPU) or a visual processing unit (VPU); or an AI processor such as a neural processing unit (NPU). According to predefined operating rules or an AI model stored in a memory, at least one processor controls the input data to be processed. Alternatively, when at least one processor is an AI processor, the AI ​​processor may be designed to have a hardware structure dedicated to processing a specific AI model.

[0196] The predefined operating rules or AI models are created by learning. When the predefined operating rules or AI models are created by learning, it means that when the basic AI model is trained using multiple training data by using a learning algorithm, the predefined operating rules or AI models set to achieve the desired characteristics (or purposes) are created. Learning can be performed by the device itself using the AI ​​according to the embodiment, or it can be performed by a separate server and / or system. Examples of learning algorithms may include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0197] The AI ​​model may include multiple neural network layers. Multiple neural network layers have multiple weight values, and each neural network layer performs a neural network operation by calculating between multiple weight values ​​and the calculation results of the previous layer. Multiple weight values ​​of multiple neural network layers can be optimized by the training results of the AI ​​model. For example, during the learning process, multiple weight values ​​can be updated to reduce or minimize the loss value or cost value obtained by the AI ​​model. The artificial neural network may include a deep neural network (DNN), and examples of artificial neural networks may include, but are not limited to, convolutional neural networks (CNNs), DNNs, recurrent neural networks (RNNs), restricted Boltzmann machines (RBMs), deep belief networks (DBNs), bidirectional recursive deep neural networks (BRDNNs), and deep Q networks.

[0198] In addition, the first AI model can be generated by learning a criterion for generating the first device knowledge graph. The first AI model can be generated by learning a criterion for determining which data to generate or use to generate the first device knowledge graph, and how to generate the first device knowledge graph by using the data.

[0199] In addition, the second AI model can be generated by learning a criterion for generating the second device knowledge graph. The second AI model can be generated by learning a criterion for determining which data to use to generate the second device knowledge graph and how to generate the second device knowledge graph by using the data.

[0200] In addition, the third AI model may be generated by learning a criterion for determining recommended content. The third AI model may be generated by learning a criterion for determining which data to use to determine recommended content and how to determine recommended content by using the data.

[0201] By using specific data as input values ​​according to preset criteria, the first AI model can output a first device knowledge graph. The result value output by the first AI model using the obtained data as input values ​​can be used to update the first AI model.

[0202] In addition, by using specific data as input values ​​according to preset criteria, the second AI model can output a second device knowledge graph. The result value output by the second AI model using the obtained data as input values ​​can be used to update the second AI model.

[0203] In addition, by using specific data as an input value according to a preset criterion, the third AI model can determine the recommended content. The result value output by the third AI model using the obtained data as an input value can be used to update the third AI model.

[0204] One or more embodiments may be implemented as a recording medium including computer executable instructions, such as a program module executed by a computer. Computer readable media may be any available media accessible to a computer, and examples thereof include all volatile and non-volatile media, and detachable and non-detachable media. In addition, examples of computer readable media may include computer storage media and communication media. Examples of computer storage media include all volatile and non-volatile media, and detachable and non-detachable media implemented by any method or technology for storing information such as computer readable instructions, data structures, program modules, or other data. Communication media typically include computer readable instructions, data structures, program modules, or other data of a modulated data signal.

[0205] In addition, the term "unit" used in this specification may be a hardware component such as a processor or a circuit, and / or a software component executed in a hardware component such as a processor.

[0206] Although the present disclosure has been particularly shown and described with reference to the embodiments of the present disclosure, it will be appreciated by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the present disclosure as defined at least in the appended claims. Therefore, it will be appreciated that the embodiments described above do not limit the scope of the present disclosure. For example, each component described in a single type may be executed in a distributed manner, while a component described in a distributed manner may be executed in an integrated form.

[0207] The scope of the present disclosure is at least represented by the claims rather than the detailed description, and it should be understood that the claims and all modifications or modified forms derived from the concepts of the claims are included in the scope of the present disclosure.

Claims

1. A device for providing content based on a knowledge graph, the device comprising: Communication interface; a memory storing one or more instructions; as well as A processor configured to execute the one or more instructions to: obtaining context information associated with the device; Generating a first device knowledge graph of the user of the device using the device by inputting the obtained context information into a first AI model for determining relationships between entities related to the user of the device; Controlling requesting from a server a server knowledge graph generated by the server; Control receiving the server knowledge graph from the server; Generating a second device knowledge graph of the user using the device by inputting the generated first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; as well as providing content based on the generated second device knowledge graph, The context information includes at least one piece of information related to the privacy of the user. wherein the server knowledge graph is generated without using any information related to the privacy of the user, and Wherein, at least one information related to the privacy of the user is not transmitted to the outside of the device.

2. The device according to claim 1, wherein: Each of the first AI model and the second AI model is an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

3. The device according to claim 1, wherein: The server knowledge graph is generated by the server based on big data provided to the server by the device and at least one other device.

4. The device according to claim 1, wherein: The processor is also configured to execute the one or more instructions to process the context information into text indicating sequential operations, and input the text into the first AI model.

5. The apparatus of claim 1, wherein: The processor is further configured to execute the one or more instructions to determine a privacy level of the first device knowledge graph, and input the determined privacy level to the first AI model; and Part of the data in the first device knowledge graph output from the first AI model includes data extracted according to the privacy level.

6. The device according to claim 1, wherein The processor is further configured to execute the one or more instructions to input a category to the first AI model; and Output the first device knowledge graph corresponding to the input category from the first AI model.

7. The device according to claim 1, wherein: The first AI model outputs the first device knowledge graph through at least one of the following: entity extraction, entity analysis and extraction, and relationship extraction.

8. The device according to claim 1, wherein: The processor is further configured to execute the one or more instructions to: controlling sending information about a user profile of the user to the server; and Controlling receiving, from the server, the server knowledge graph associated with the user profile.

9. The device according to claim 1, wherein: The processor is also configured to execute the one or more instructions to control sending a specific category to the server, and receiving the server knowledge graph corresponding to the specific category from the server.

10. The device according to claim 1, wherein: The processor is further configured to execute the one or more instructions to: Determine content to be recommended to the user by inputting the operation information of the device and the second device knowledge graph into a third AI model; as well as Control requests the content providing server for the determined content to be recommended.

11. A method for providing content based on a knowledge graph, performed by a device, the method comprising: obtaining context information associated with the device; Generating a first device knowledge graph of the user of the device using the device by inputting the obtained context information into a first AI model for determining relationships between entities related to the user of the device; Requesting from a server a server knowledge graph generated by the server; Receiving the server knowledge graph from the server; Generating a second device knowledge graph of the user using the device by inputting the generated first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; as well as providing content based on the generated second device knowledge graph, The context information includes at least one piece of information related to the privacy of the user. wherein the server knowledge graph is generated without using any information related to the privacy of the user, and Wherein, at least one information related to the privacy of the user is not transmitted to the outside of the device.

12. The method according to claim 11, wherein: Each of the first AI model and the second AI model is an AI model trained by using at least one of the following as an AI algorithm: a machine learning algorithm, a neural network algorithm, a genetic algorithm, a deep learning algorithm, or a classification algorithm.

13. The method according to claim 11, wherein: The server knowledge graph is generated by the server based on big data provided to the server by the device and at least one other device.

14. The method according to claim 11, further comprising: processing the context information into text indicating sequential operations, Among them, obtaining the first device knowledge graph includes inputting the text into the first AI model.

15. A non-transitory computer-readable recording medium having recorded thereon a program executable by at least one processor to perform: Obtaining contextual information related to the device; Generating a first device knowledge graph of the user of the device using the device by inputting the obtained context information into a first AI model for determining relationships between entities related to the user of the device; Controlling requesting from a server a server knowledge graph generated by the server; Control receiving the server knowledge graph from the server; Generating a second device knowledge graph of the user using the device by inputting the generated first device knowledge graph and the received server knowledge graph into a second AI model for expanding the first device knowledge graph; as well as providing content based on the generated second device knowledge graph, The context information includes at least one piece of information related to the privacy of the user. wherein the server knowledge graph is generated without using any information related to the privacy of the user, and Wherein, at least one information related to the privacy of the user is not transmitted to the outside of the device.

Images