System

Abstract

A system is provided.SOLUTION: A system includes a data collection means for collecting behavior data of a user, a model generation means for generating and training an AI model optimized for a user's interest based on the collected behavior data, a bug incorporation means for intentionally incorporating a statistical bug into the generated AI model and providing information unexpected for the user, an information provision means for providing the user terminal with the information obtained by using the AI model into which the bug is incorporated, and a feedback means for re-collecting a user's reaction or behavior data and using it for improving the model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional AI systems provide information optimized based on the behavioral data of individual users, which means that users only get information that is biased toward their hobbies and interests, resulting in a loss of behavioral diversity. As a result, opportunities to discover new interests and perspectives decrease, user behavior becomes fixed, and diversity in society as a whole is lost. [Means for solving the problem]

[0005] The present invention provides a system that collects user behavioral data and intentionally embeds statistical bugs into an AI model trained based on this data to provide unexpected information to the user. Specifically, the system includes a data collection means, a model generation means, a bug embedding means, an information provision means, and a feedback means. The data collection means collects user behavioral data such as search history, ad click history, and social media activity, and the model generation means generates and trains an AI model optimized for the user's interests. The bug embedding means then randomly generates noise or irrelevant information and inserts it into the AI ​​model to provide the user with unexpected information. The user's reactions to this information and behavioral data are then collected again by the feedback means and used to improve the model. This allows users to acquire new interests and behaviors and promotes behavioral diversity.

[0006] A "data collection method" is a device, software, or process for collecting behavioral data such as user search history, ad click history, and social media activity.

[0007] "Model generation means" means a device, software, or process for generating and training an AI model optimized to the user's interests based on the behavioral data collected by the data collection means.

[0008] A "bug injection method" is a device, software, or process that intentionally inserts random noise or irrelevant information into a generated AI model.

[0009] "Information provision means" refers to a device, software, or process for transmitting and displaying information or advertisements generated using an AI model incorporating a bug to a user's device.

[0010] A "feedback mechanism" is a device, software, or process that recollects user reaction and behavior data and uses it to improve the model.

[0011] "Bug data" is data that contains information unrelated to the user's interests or random noise, and is intentionally inserted into an AI model. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0013] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0014] First, the terms used in the following description will be explained.

[0015] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0016] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0017] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0018] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0019] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0020] [First embodiment]

[0021] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0022] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0023] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0024] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0025] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0026] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0027] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0028] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0029] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0030] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0031] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0032] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0033] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user. Hereinafter, specific embodiments of the invention will be described.

[0034] System Overview

[0035] The system includes the following elements:

[0036] Data collection methods

[0037] Model Generation Method

[0038] How to Introduce Bugs

[0039] Information provision means

[0040] Feedback Methods

[0041] Data collection methods

[0042] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[0043] Model Generation Method

[0044] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[0045] How to Introduce Bugs

[0046] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[0047] Information provision means

[0048] The server then sends the information generated by the bugged AI model to the user's device, which receives it and displays it to the user. This information includes both relevant information and intentionally inserted unexpected information.

[0049] Feedback Methods

[0050] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[0051] Specific examples

[0052] Example 1: User A (Interest: Fashion)

[0053] 1. Data Collection

[0054] User A frequently searches for fashion-related sites and clicks data is collected.

[0055] This information is sent from the terminal to the server.

[0056] 2. Model Generation

[0057] The server trains an AI model optimized for User A based on fashion-related data.

[0058] 3. Introducing bugs

[0059] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0060] 4. Information provision

[0061] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0062] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0063] 5. Feedback

[0064] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[0065] The server uses this data to improve the model.

[0066] Example 2: User B (Interest: Technology)

[0067] 1. Data Collection

[0068] User B browses technology-related news and the browsing history is sent to the server.

[0069] 2. Model Generation

[0070] The server trains an optimized technology-related AI model based on User B's interests.

[0071] 3. Introducing bugs

[0072] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0073] 4. Information provision

[0074] The server sends technology-related ads, as well as art-related ads, to the device.

[0075] 5. Feedback

[0076] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0077] In this way, by providing both information that is in line with the user's interests and unexpected information, this system promotes diversity in user behavior and provides opportunities for users to discover new interests and behaviors.

[0078] The processing flow will be explained below.

[0079] Step 1:

[0080] Users conduct searches, click on ads, and post activities on social media.

[0081] Step 2:

[0082] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[0083] Step 3:

[0084] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[0085] Step 4:

[0086] The server collects user behavior data from the database and preprocesses it (noise removal, data standardization, etc.).

[0087] Step 5:

[0088] The server generates and trains an AI model optimized for each user based on the preprocessed data.

[0089] Step 6:

[0090] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[0091] Step 7:

[0092] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[0093] Step 8:

[0094] The terminal displays to the user the information sent from the server, including both information based on the user's interests and unexpected information based on bug data.

[0095] Step 9:

[0096] The user browses the displayed information, clicks on information that interests them, or searches further.

[0097] Step 10:

[0098] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[0099] Step 11:

[0100] The server collects new behavioral data sent from the terminal and stores it in a database.

[0101] Step 12:

[0102] The server analyzes the new behavioral data and analyzes which bugs affected user behavior.

[0103] Step 13:

[0104] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates in future iterations.

[0105] In this way, the system collects and analyzes data at each step and provides users with a variety of information, thereby promoting diversity in user behavior.

[0106] Example 1

[0107] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0108] Conventional information provision systems focused on providing information based on users' interests and behavioral patterns, but lacked mechanisms for eliciting new interests and behaviors. This resulted in users being confined to their existing interests and having few opportunities to explore new fields of interest. Furthermore, the system did not effectively utilize a feedback loop to improve the accuracy of models based on user behavioral data.

[0109] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0110] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, a data storage and preprocessing means for storing the data in a database and preprocessing it, and a random bug generation means for generating bug data. This makes it possible to build an effective feedback loop that not only provides information based on the user's existing interests but also promotes the exploration of new interests and behaviors and contributes to improving the accuracy of the model.

[0111] The "data collection means" is a means for collecting user behavior data.

[0112] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[0113] "Bug injection" is a method of intentionally incorporating statistical bugs into a generated AI model to provide unexpected information to the user.

[0114] "Information provision means" refers to a means of providing information obtained using an AI model with a bug incorporated into it to a user terminal.

[0115] "Feedback means" refers to a means of recollecting user reactions and behavioral data and using it to improve the model.

[0116] "Data storage and preprocessing means" refers to the means for storing data in a database and preprocessing it.

[0117] The "random bug generation means" is a means for generating bug data.

[0118] "Behavioral data" refers to data such as a user's search history, ad click history, and social media activity.

[0119] An "AI model" is a model generated and trained by machine learning algorithms that analyzes and predicts user interests and behavioral patterns.

[0120] A "statistical bug" is noise or irrelevant information that is intentionally inserted to cause errors in the predictions of an AI model.

[0121] A "user terminal" is a terminal device such as a computer or mobile device used by a user.

[0122] MODE FOR CARRYING OUT THE INVENTION

[0123] This invention relates to a system that collects user behavioral data, intentionally incorporates statistical bugs into AI models trained based on this data, and provides users with unexpected information.

[0124] Data collection methods

[0125] The server collects user behavior data. Specifically, when a user uses a web or mobile application, the server sends data such as search history, ad click history, and social media activity from the device in real time. This allows the server to collect detailed information about the user's behavioral patterns.

[0126] Model Generation Method

[0127] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This process uses machine learning algorithms and performs data preprocessing such as filling in missing values, removing noise, and normalizing the data. For example, for User A, who is interested in fashion, an AI model is generated that prioritizes providing fashion-related information.

[0128] How to Introduce Bugs

[0129] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's hobbies and preferences, as well as random noise. A specific algorithm is applied to generate bug data, randomly generating unrelated data and inserting it into the AI ​​model. For example, "gardening" information is incorporated into the model of User A, who is interested in fashion.

[0130] Information provision means

[0131] The server then sends the information generated using the bugged AI model to the user's device, which receives it and displays it to the user. The user's device displays both information relevant to their interests and intentionally inserted unexpected information. For example, a gardening ad might appear alongside a fashion ad.

[0132] Feedback Methods

[0133] The device records behavioral data on the user's reactions to the displayed information (e.g., clicks, searches, etc.). This data is then sent back to the server and used to improve the AI ​​model. This feedback makes future information provision more suitable for the user. For example, if User A clicks on a gardening-related advertisement, the data is sent to the server and the model is adjusted.

[0134] Specific examples

[0135] Example 1: User A (Interest: Fashion)

[0136] 1. Data Collection

[0137] User A frequently searches for fashion-related sites and clicks data is collected.

[0138] This information is sent from the terminal to the server.

[0139] 2. Model Generation

[0140] The server trains an AI model optimized for User A based on fashion-related data.

[0141] 3. Introducing bugs

[0142] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0143] 4. Information provision

[0144] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0145] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0146] 5. Feedback

[0147] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the terminal and sent to the server.

[0148] The server uses this data to improve the model.

[0149] Example 2: User B (Interest: Technology)

[0150] 1. Data Collection

[0151] User B browses technology-related news and the browsing history is sent to the server.

[0152] 2. Model Generation

[0153] The server trains an optimized technology-related AI model based on User B's interests.

[0154] 3. Introducing bugs

[0155] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0156] 4. Information provision

[0157] The server sends technology-related ads, as well as art-related ads, to the device.

[0158] 5. Feedback

[0159] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0160] Prompt Sentence Examples

[0161] Example prompt for user A

[0162] "User A is interested in fashion, but the next information we show them should also include gardening information."

[0163] Example prompt for User B

[0164] "User B prefers technology-related information, but please also include some information about art."

[0165] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0166] Specific explanation of processing steps

[0167] Step 1: Data collection

[0168] The server collects user behavior data from the device. Input includes search terms performed by the user, ads clicked, social media posts and comments, etc. Using data collection tools, the server receives this data in real time and stores it in a database. Specifically, if a user searches for "latest fashion," the search term data is sent to the server. The output is the stored behavioral data.

[0169] Step 2: Data storage and preprocessing

[0170] The server stores the collected data in a database and performs preprocessing. The behavioral data collected in step 1 is used as input. The data storage and preprocessing means performs missing value completion, noise removal, and data normalization. For example, error values ​​are removed from the collected data to create up-to-date normalized fashion-related data. The output is the preprocessed data.

[0171] Step 3: Training the AI ​​model

[0172] The server trains an AI model based on the preprocessed data. The input is the preprocessed data generated in step 2. Using the model generation means, a machine learning algorithm generates a model that predicts the user's interests and behavioral patterns. Specifically, the server trains the AI ​​model based on the fashion data and creates a predictive model optimized for user A. The output is the trained AI model.

[0173] Step 4: Generate random bugs

[0174] The server generates random bugs for the trained AI model. The input is the AI ​​model generated in step 3. Using the random bug generation means, irrelevant data and noise are generated and incorporated into the model. For example, for a model of user A who is interested in fashion, irrelevant data related to gardening is randomly generated. The output is an AI model containing bugs.

[0175] Step 5: Incorporating bug data

[0176] The server incorporates the generated random bug data into the AI ​​model. The input is the random bug data generated in step 4. The bug injection means is used to insert irrelevant data into the model. Specifically, the server inserts "gardening" information into the AI ​​model. The output is an AI model with a bug injected.

[0177] Step 6: Generate customization information

[0178] The server generates customized information using the AI ​​model with the bug. The input is the AI ​​model with the bug created in step 5. The information to be displayed to the user is generated by predictions made through the model. Specifically, fashion advertisements and some gardening advertisements are generated. The output is customized information.

[0179] Step 7: Submit your information

[0180] The server sends the generated customization information to the user's terminal. The input is the customization information generated in step 6. The information is sent to the user's terminal using the information providing means. The terminal displays the received information on the screen. As a specific operation, a fashion advertisement and a gardening advertisement are displayed on User A's terminal. The output is the information displayed on the user's terminal.

[0181] Step 8: Recording behavioral data

[0182] The terminal records the user's reactions to the displayed information (clicks, searches, etc.). The input is the user's actions. This data is sent back to the server using feedback means. For example, if user A clicks on a gardening advertisement, the click information is recorded and sent to the server. The output is the recorded behavioral data.

[0183] Step 9: Improve the model

[0184] The server retrains the AI ​​model based on the transmitted behavioral data and improves the model. The input is the behavioral data transmitted in step 8. The model is adjusted and retrained using the model generation means. Specifically, the server improves the AI ​​model based on User A's click data and reflects this in the next information provision. The output is the improved AI model.

[0185] (Application example 1)

[0186] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0187] Conventional advertising delivery systems have focused on displaying advertisements based on users' interests, but this has the problem of narrowing the diversity of users' interests and behaviors. It also limits opportunities to discover unexpected information or new interests. Therefore, there is a need for systems that allow users to discover new interests and promote the diversity of their behaviors.

[0188] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0189] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected advertisements, an information provision means for providing the user's smartphone with advertisement information obtained using the AI ​​model with the bug injected, and a feedback means for re-collecting the user's response to advertisements and behavioral data and using it to improve the model, thereby enabling users to discover new interests and information and promoting behavioral diversity.

[0190] The "data collection means" is a means for collecting user behavior data.

[0191] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[0192] "Bug injection" refers to the intentional incorporation of statistical bugs into the generated AI model to provide users with unexpected advertisements.

[0193] The "information provision means" is a means of providing advertising information obtained using an AI model with a built-in bug to the user's smartphone.

[0194] "Feedback means" refers to a means of recollecting user responses to advertisements and behavioral data and using it to improve the model.

[0195] The "advertising delivery system" is a system that generates an AI model based on user behavior data and delivers advertisements that intentionally incorporate bugs to attract new user interest.

[0196] An "AI model" is an artificial intelligence model that is generated and trained based on user behavior data.

[0197] A "statistical bug" is unexpected information or random noise that is intentionally incorporated into an AI model.

[0198] "User behavioral data" refers to information such as a user's search history, ad click history, and social media activity.

[0199] A "smartphone" is a portable information terminal that is primarily used by users.

[0200] This invention is an advertising distribution system that collects user behavior data and intentionally incorporates statistical bugs into a trained AI model to provide users with unexpected advertisements. Specific embodiments of this system are described below.

[0201] System Overview

[0202] The system includes the following elements:

[0203] Data collection methods

[0204] Model Generation Method

[0205] How to Introduce Bugs

[0206] Information provision means

[0207] Feedback Methods

[0208] Data collection methods

[0209] The server collects user behavioral data, including search history, ad click history, social network activity, etc. This data is sent from the device to the server in real time when the user uses the smartphone.

[0210] Model Generation Method

[0211] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This AI model uses machine learning algorithms to analyze the user's interests and behavioral patterns and predict which advertisements should be displayed next.

[0212] How to Introduce Bugs

[0213] The server intentionally embeds statistical bugs into the generated AI model. This bug data includes advertising information unrelated to the user's interests and random noise. For this purpose, the server generates the bug data using a statistical algorithm.

[0214] Information provision means

[0215] The server then uses the bugged AI model to generate advertising information, which is then sent to the user's smartphone, which then receives the information and displays the ads to the user. This advertising information includes both ads tailored to the user's interests and intentionally inserted unexpected ads.

[0216] Feedback Methods

[0217] It records how users respond to the ads they see (for example, clicks, viewing time, etc.). This behavioral data is sent back to the server and used to improve the model. This feedback makes future ad serving more effective.

[0218] Specific examples

[0219] Example 1: User C (Interest: Technology)

[0220] 1. Data Collection

[0221] User C frequently browses technology-related content and their behavioral data is collected.

[0222] The data is sent from the smartphone to the server in real time.

[0223] 2. Model Generation

[0224] The server trains an AI model optimized for User C based on technology-related data.

[0225] 3. Introducing bugs

[0226] The server inserts advertising information for "sports," which differs from the technology, into the AI ​​model as a bug.

[0227] 4. Information provision

[0228] The server generates technology-related ads, along with some sports-related ads, and sends them to the smartphone.

[0229] When the smartphone displays this information, User C sees sports ads along with technology ads.

[0230] 5. Feedback

[0231] If User C is interested in a sports-related advertisement and clicks on it, his / her behavior is recorded and sent to the server.

[0232] The server uses this data to improve the model.

[0233] Prompt Sentence Examples

[0234] "Generate an AI model based on the categories of web pages frequently visited by users, and add ads for irrelevant categories with a set probability."

[0235] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0236] Step 1:

[0237] The server collects user behavior data. Specifically, it collects data in real time when users use their smartphones to browse websites or click on ads (search history, ad click history, social network activity, etc.). The input is user behavior data, and the output is the user behavior data stored on the server.

[0238] Step 2:

[0239] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. Specifically, it analyzes this behavioral data using a machine learning algorithm (e.g., RandomForestClassifier) ​​to build a model that predicts the next advertisement the user is likely to be interested in. The input is the collected behavioral data, and the output is the trained AI model.

[0240] Step 3:

[0241] The server intentionally incorporates statistical bugs into the generated AI model. Specifically, it applies an algorithm that adds random noise or advertising information unrelated to the user's interests (e.g., sports-related information) to the AI ​​model. The input is the trained AI model and the added bug data, and the output is an AI model with the bug incorporated.

[0242] Step 4:

[0243] The server uses the AI ​​model with the bug to generate advertising information and send it to the user's smartphone. Specifically, the advertising information generated by the AI ​​model is sent as a data packet to the smartphone, which receives and displays it. The input is the AI ​​model with the bug, and the output is the advertising information sent to the user's smartphone.

[0244] Step 5:

[0245] The device records how users respond to displayed ads (clicks, viewing time, etc.). Specifically, the smartphone application captures user responses as events and sends the data to a server. The input is the user's response to the ad, and the output is the feedback data sent to the server.

[0246] Step 6:

[0247] The server retrains and improves the AI ​​model based on the collected feedback data. Specifically, it adds the user's new behavioral data to the training dataset and updates the AI ​​model. The input is the new behavioral data including the feedback data, and the output is the improved AI model.

[0248] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0249] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with an emotion engine that recognizes the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[0250] System Overview

[0251] The system includes the following elements:

[0252] Data collection methods

[0253] Model Generation Method

[0254] How to Introduce Bugs

[0255] Information provision means

[0256] Feedback Methods

[0257] Emotion Engine

[0258] Data collection methods

[0259] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[0260] Model Generation Method

[0261] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[0262] How to Introduce Bugs

[0263] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[0264] Information provision means

[0265] The server then provides the user with information generated using the AI ​​model with the bugs. The device receives the information and displays it to the user. This information includes both information relevant to the user's interests and unexpected information based on the bug data.

[0266] Feedback Methods

[0267] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[0268] Emotion Engine

[0269] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The emotion engine acquires emotion data from the user's webcam, microphone, and text input. The server analyzes this data to determine the user's current emotional state.

[0270] The emotional data obtained by the emotion engine influences how information is presented to users. For example, if a user is excited, ads with content that matches that emotion will be dynamically served. Emotional data is also used as a feedback tool to help train AI models.

[0271] Specific examples

[0272] Example 1: User A (Interest: Fashion)

[0273] 1. Data Collection

[0274] User A frequently searches for fashion-related sites and clicks data is collected.

[0275] This information is sent from the terminal to the server.

[0276] 2. Model Generation

[0277] The server trains an AI model optimized for User A based on fashion-related data.

[0278] 3. Introducing bugs

[0279] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0280] 4. Information provision

[0281] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0282] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0283] 5. Feedback

[0284] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[0285] The server uses this data to improve the model.

[0286] 6. Emotion recognition

[0287] While User A is browsing the information, the emotion engine analyzes User A's emotions from their facial expressions and voice and sends the results to the server.

[0288] The server provides appropriate information to User A based on the emotion data.

[0289] Example 2: User B (Interest: Technology)

[0290] 1. Data Collection

[0291] User B browses technology-related news and the browsing history is sent to the server.

[0292] 2. Model Generation

[0293] The server trains an optimized technology-related AI model based on User B's interests.

[0294] 3. Introducing bugs

[0295] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0296] 4. Information provision

[0297] The server sends technology-related ads, as well as art-related ads, to the device.

[0298] 5. Feedback

[0299] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0300] 6. Emotion recognition

[0301] While User B is browsing the information, the emotion engine analyzes User B's emotions from his / her tone of voice and text input and sends the results to the server.

[0302] The server dynamically changes the advertisement content it provides based on the analyzed emotion data.

[0303] In this way, this system not only provides users with information that is in line with their interests and unexpected information, but also uses emotional data to provide more personalized information, thereby simultaneously improving the diversity of user behavior and satisfaction.

[0304] The processing flow will be explained below.

[0305] Step 1:

[0306] Users conduct searches, click on ads, and post activities on social media.

[0307] Step 2:

[0308] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[0309] Step 3:

[0310] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[0311] Step 4:

[0312] The server collects user behavior data from the database and preprocesses the data (noise removal, data standardization, etc.).

[0313] Step 5:

[0314] The server uses the preprocessed data to generate and train an AI model optimized for each user.

[0315] Step 6:

[0316] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[0317] Step 7:

[0318] Users express their emotions through a webcam and microphone.

[0319] Step 8:

[0320] The device collects the user's voice, facial expressions, and text in real time and sends them to the emotion engine.

[0321] Step 9:

[0322] The emotion engine analyzes the received data, identifies the user's emotional state, and sends the results to the server.

[0323] Step 10:

[0324] The server analyzes the emotion data sent from the emotion engine and dynamically adjusts the information and advertisements provided based on the user's emotional state.

[0325] Step 11:

[0326] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[0327] Step 12:

[0328] The terminal displays the information sent from the server to the user, including both information of interest to the user and unexpected information from bug data.

[0329] Step 13:

[0330] The user browses the displayed information, clicks on the information that interests them, or searches further.

[0331] Step 14:

[0332] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[0333] Step 15:

[0334] The server collects new behavioral data sent from the terminal and stores it in a database.

[0335] Step 16:

[0336] The server analyzes the new behavioral data and determines which bugs affected user behavior.

[0337] Step 17:

[0338] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates from the next time onwards.

[0339] In this way, the system collects and analyzes data at each step and provides users with a variety of information, promoting diverse user behavior. Furthermore, it uses an emotion engine to understand the user's emotional state and provide more personalized information.

[0340] Example 2

[0341] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0342] Modern information delivery systems are required to accurately predict user interests. However, typical systems lack the flexibility to respond to unexpected user behaviors and emotions, limiting the improvement of user experience. Furthermore, current systems have difficulty acquiring user emotional data in real time and providing information based on that data. Therefore, there is a need for a system that can simultaneously analyze a user's diverse behaviors and the emotions behind them, and dynamically provide appropriate information.

[0343] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0344] In this invention, the server includes means for collecting user behavioral data, means for generating and training a generative AI model optimized for the user's interests based on the collected behavioral data, means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected information, means for providing information obtained using the AI ​​model with the bugs incorporated to the user terminal, means for re-collecting user reaction and behavioral data and using it to improve the model, and emotion recognition means for analyzing emotions from the user's voice, facial expressions, and text to obtain emotion data. This makes it possible to efficiently analyze the user's diverse behavioral data and emotion data and provide the user with unexpected and useful information.

[0345] "User behavioral data" refers to data generated when a user uses the web or mobile applications, such as search history, ad click history, and social network service activity.

[0346] A "generative AI model" is a machine learning model built to analyze a user's interests and behavioral patterns based on collected user behavioral data and predict what information should be displayed next.

[0347] A "statistical bug" is data that contains information unrelated to the user's preferences or random noise that is intentionally incorporated into the generated AI model.

[0348] The "emotion recognition means" is an engine or algorithm for analyzing emotions from the user's voice, facial expressions, and text, and obtaining emotion data based on this.

[0349] "Information provision means" refers to a system or process that provides information generated using an AI model with a bug to a user's device and displays it to the user.

[0350] "Feedback means" refers to a means of recollecting the user's reactions to the displayed information (clicks, searches, etc.) and using this information to improve the AI ​​model.

[0351] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into a generative AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with emotion recognition means for recognizing the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[0352] System Overview

[0353] The system includes the following elements:

[0354] Behavioral data collection methods

[0355] Model Generation Method

[0356] How to Introduce Bugs

[0357] Information provision means

[0358] Feedback Methods

[0359] emotion recognition means

[0360] Behavioral data collection methods

[0361] The server collects behavioral data generated when users use web and mobile applications, such as search history, ad click history, social network service activity, etc. This behavioral data is collected in real time and transmitted to the server.

[0362] Model Generation Method

[0363] The server generates and trains a generative AI model optimized for the user's interests based on the collected behavioral data. This generative AI model uses machine learning algorithms (e.g., TensorFlow, PyTorch) to analyze the user's behavioral data and predict what information should be displayed next.

[0364] How to Introduce Bugs

[0365] The server intentionally incorporates statistical bugs into the trained generative AI model. This bug data includes information unrelated to the user's preferences and random noise, and uses a specific algorithm (e.g., a noise generation algorithm) to generate it.

[0366] Information provision means

[0367] The server provides information generated using the generative AI model with the bug to the user's device. The device receives this information and displays it to the user. The displayed information includes both information relevant to the user's interests and unexpected information based on the bug data.

[0368] Feedback Methods

[0369] The device records the user's reactions to the displayed information (clicks, searches, etc.) and sends the behavioral data back to the server. This feedback data is used to improve the AI ​​model, making future information provision even more effective.

[0370] emotion recognition means

[0371] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. Emotional data is acquired through the user's webcam, microphone, and text input, and the server analyzes this data to identify the user's current emotional state. The analyzed emotional data is also used for information provision and feedback means, making it possible to dynamically adjust information.

[0372] Examples of concrete examples and prompts

[0373] Example 1: User A (Interest: Fashion)

[0374] 1. Data Collection

[0375] User A frequently searches for fashion-related sites and clicks data is collected.

[0376] This information is sent from the terminal to the server.

[0377] 2. Model Generation

[0378] The server trains a generative AI model optimized for User A based on fashion-related data.

[0379] 3. Introducing bugs

[0380] The server inserts "gardening" information, which is different from fashion, into the generative AI model as a bug.

[0381] 4. Information provision

[0382] The server generates some gardening-related advertisements along with fashion-related advertisements and sends them to the terminal. When the terminal displays this information, User A sees the gardening advertisements along with the fashion advertisements.

[0383] 5. Feedback

[0384] If User A is interested and clicks on a gardening-related ad, that behavior is recorded by the device and sent to the server, which uses this data to improve the model.

[0385] 6. Emotion recognition

[0386] While User A is browsing information, the emotion recognition means analyzes User A's emotions from their facial expressions and voice and sends the results to the server. The server then provides appropriate information to User A based on the emotion data.

[0387] Example prompt sentence:

[0388] "User A is interested in fashion, but how will he react if we mix in random gardening information? Analyze the sentiment data as well."

[0389] Example 2: User B (Interest: Technology)

[0390] 1. Data Collection

[0391] User B browses technology-related news and the browsing history is sent to the server.

[0392] 2. Model Generation

[0393] The server trains an optimized generative AI model related to technology based on User B's interests.

[0394] 3. Introducing bugs

[0395] The server inserts "art" information, which is different from technology, into the generative AI model as bug data.

[0396] 4. Information provision

[0397] The server sends technology-related ads, as well as art-related ads, to the device.

[0398] 5. Feedback

[0399] If User B shows interest in an art-related ad and clicks on it, that data is recorded on the device and sent to the server, which uses this data to retrain the generative AI model.

[0400] 6. Emotion recognition

[0401] The emotion recognition means analyzes emotions from the voice and text input of User B and sends the data to the server. The server changes the advertisement content to be provided based on the analyzed emotion data.

[0402] Example prompt sentence:

[0403] "User B is interested in technology, but if you add some random art information, how will they behave? We'll take their sentiment data into account and do some analysis."

[0404] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0405] System program processing flow and processing steps

[0406] Step 1: Collect behavioral data

[0407] The server collects user behavioral data. As input, it obtains search history, ad click history, social network service activity, and other data in real time when users use web and mobile applications. This data is automatically sent from the device to the server when the user visits a specific website or clicks on a product. As output, the collected behavioral data is stored in a database.

[0408] Specific operation: A user searches a fashion-related website, enters the keyword "fall fashion," and clicks on several items. The data is sent from the device to the server in real time.

[0409] Step 2: Creating a generative AI model

[0410] The server creates and trains a generative AI model based on the collected behavioral data. It uses the behavioral data stored in the database as input. It uses a machine learning framework (e.g., TensorFlow, PyTorch) to analyze the user's interests and behavioral patterns. This analysis generates an AI model optimized for the user's interests. The output is a trained generative AI model.

[0411] How it works: The server analyzes fashion-related behavioral data and determines that "this user is interested in fall fashion." It then trains a generative AI model based on this interest.

[0412] Step 3: Incorporating the bug

[0413] The server intentionally incorporates bug data into a trained generative AI model. As input, it uses the trained generative AI model and a specific algorithm (e.g., a noise generation algorithm) to generate bug data. The bug data includes information unrelated to the user's preferences and random noise. As output, it obtains a generative AI model incorporating the bug data.

[0414] Specific operation: The server adds "gardening" information as a bug to the generative AI model, which generates "gardening"-related information along with fashion-related information.

[0415] Step 4: Provide information

[0416] The server uses a generative AI model with bugs built in to provide generated information to the terminal. As input, the server uses a generative AI model with bugs built in. The server uses this model to generate information based on the user's interests and unexpected information based on bug data. As output, the generated information is sent to the terminal and displayed to the user.

[0417] Specific operation: The server sends information including fashion-related advertisements and "gardening"-related advertisements to the terminal, which receives it and displays it to the user.

[0418] Step 5: Gather feedback

[0419] The user reacts to the displayed information (clicks, searches, etc.). The device records the user's reaction and sends it back to the server. The user's reaction data is used as input. As output, this reaction data is sent to the server and used to improve the AI ​​model.

[0420] How it works: When a user clicks on a gardening-related ad, the click data is recorded on the device and sent to the server, which uses this data to retrain the AI ​​model and improve the information provided next time.

[0421] Step 6: Emotion Recognition

[0422] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. As input, emotion data obtained through a webcam, microphone, and text input is used. The emotion recognition engine analyzes this data to identify the user's current emotional state. As output, the analyzed emotion data is sent to a server and used for information provision and feedback means.

[0423] Specific operation: While the user is browsing information using a webcam, the emotion recognition means collects facial expression data and detects an "excited" state. The server dynamically adjusts the information provided based on this data.

[0424] (Application example 2)

[0425] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0426] While conventional systems can provide information tailored to a user's interests, they lack the ability to intentionally provide different information to stimulate new interests. Furthermore, the information provided did not reflect the user's emotions, which could have further improved the quality of the user experience. In addition, the feedback loop for improving the AI ​​model based on collected data was not fully functional.

[0427] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0428] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, and an emotion engine for analyzing emotions from the user's voice, facial expressions, and text and providing information based on the emotion data. This makes it possible to provide advertisements that correspond to the user's emotions and discover new areas of interest.

[0429] The "data collection means" is a means for collecting user behavior data.

[0430] "Model generation means" means a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[0431] "Bug injection methods" are methods for intentionally incorporating statistical bugs into generated AI models to provide users with unexpected information.

[0432] "Information provision means" refers to a means for providing information obtained using an AI model incorporating a bug to a user terminal.

[0433] "Feedback means" refers to a means for recollecting user reactions and behavioral data and using it to improve the model.

[0434] The "emotion engine" is an engine that analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotion data.

[0435] This invention is a system that collects user behavioral data, intentionally incorporates statistical bugs into an AI model trained based on this data, and provides users with unexpected information, and further combines this with an emotion engine that recognizes the user's emotions.

[0436] The server implements the system using the following hardware and software.

[0437] Data collection method: Collect user behavior data (search history, ad click history, social media activity) using tools such as Google Analytics API and Mixpanel.

[0438] Model generation: Generate and train an AI model optimized for user interests based on collected behavioral data, using machine learning algorithms such as TensorFlow and Scikit-learn.

[0439] Bug injection: Intentionally injecting statistical bugs into the generated AI model to provide unexpected information. This involves using a specific algorithm to randomly generate noise or irrelevant information.

[0440] Information provision method: Information obtained using an AI model with a bug is provided to the user's device.

[0441] Feedback methods: Recollect user responses and behavioral data and use it to improve the AI ​​model. Specifically, use collection tools such as Firebase Analytics.

[0442] Emotion engine: Analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotional data. Specifically, it uses emotion recognition tools such as OpenCV and IBM Watson.

[0443] Example of User A (interested in fashion)

[0444] 1. Data collection: Collect data on User A's searches and clicks on fashion-related sites. The information is collected through the Google Analytics API.

[0445] 2. Model generation: Based on the collected data, the server trains an optimized fashion-related AI model using TensorFlow.

[0446] 3. Bug insertion: The server inserts information about "gardening," which is unrelated to fashion, into the AI ​​model as a bug.

[0447] 4. Information provision: The server generates fashion and some gardening-related advertisements and sends them to the user's terminal.

[0448] 5. Feedback: User A's response to the ad (click or ignore) is recorded and sent to the server via Firebase Analytics.

[0449] 6. Emotion recognition: While User A is browsing information, an emotion engine using OpenCV analyzes User A's facial expressions and voice and provides appropriate information based on that data.

[0450] Prompt Sentence Examples

[0451] "The user is interested in 'smartwatches,' so use an AI model to show them ads related to smartwatches. Also, intentionally show ads related to 'gardening,' and prioritize uplifting information because the user is emotionally excited."

[0452] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0453] Step 1:

[0454] The server uses data collection tools to collect user behavior data. Inputs include the user's search history, ad click history, and social media activity on web and mobile applications. This data is collected in real time through tools such as Google Analytics API and Mixpanel and stored in a database as behavioral data. The output is a compiled user behavior data.

[0455] Step 2:

[0456] The server generates and trains an AI model optimized for the user's interests based on the behavioral data collected using the model generation means. The behavioral data collected in step 1 is used as input. A machine learning algorithm (e.g., TensorFlow or Scikit-learn) is used to analyze the user's interests and behavioral patterns and generate a predictive model. The output is an optimized AI model.

[0457] Step 3:

[0458] The server intentionally embeds statistical bugs into the generated AI model using a bug injection method. The AI ​​model generated in step 2 is used as input. Random noise and irrelevant information (bug data) are generated using a specific algorithm and inserted into the model. The output is an AI model with embedded statistical bugs.

[0459] Step 4:

[0460] The server uses an information provision means to provide information generated based on the AI ​​model with the bug to the user's device. The AI ​​model obtained in step 3 is used as input. Based on the AI ​​model, advertisements and content containing unexpected information are generated and delivered to the device, taking into account the user's interests. The information displayed on the user's device is obtained as output.

[0461] Step 5:

[0462] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The inputs are provided by the user's webcam, microphone, and text input. For example, OpenCV or IBM Watson are used to collect emotion data and analyze it in real time. The output is the user's current emotional state, which is sent to the server.

[0463] Step 6:

[0464] The server reflects the emotional data obtained from the emotion engine and dynamically influences the information provided by the information providing means. For example, if the user is excited, it provides relaxing content, and if the user is uncomfortable, it provides information to excite the user. The emotional data obtained in step 5 is used as input. Information that matches the emotion is provided as output.

[0465] Step 7:

[0466] The server again collects user reactions and behavioral data through feedback channels and uses it to improve the model. The input is behavioral data, such as user click history and search history, and emotional data collected from the device. The data is accumulated and analyzed using collection tools such as Firebase Analytics. The output is an improved model that can be reflected in the next information provision.

[0467] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0468] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0469] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0470] [Second embodiment]

[0471] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0472] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0473] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0474] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0475] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0476] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0477] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0478] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0479] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0480] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0481] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0482] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0483] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user. Hereinafter, a specific embodiment of this invention will be described.

[0484] System Overview

[0485] The system includes the following elements:

[0486] Data collection methods

[0487] Model Generation Method

[0488] How to Introduce Bugs

[0489] Information provision means

[0490] Feedback Methods

[0491] Data collection methods

[0492] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[0493] Model Generation Method

[0494] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[0495] How to Introduce Bugs

[0496] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[0497] Information provision means

[0498] The server then sends the information generated by the bugged AI model to the user's device, which receives it and displays it to the user. This information includes both relevant information and intentionally inserted unexpected information.

[0499] Feedback Methods

[0500] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[0501] Specific examples

[0502] Example 1: User A (Interest: Fashion)

[0503] 1. Data Collection

[0504] User A frequently searches for fashion-related sites and clicks data is collected.

[0505] This information is sent from the terminal to the server.

[0506] 2. Model Generation

[0507] The server trains an AI model optimized for User A based on fashion-related data.

[0508] 3. Introducing bugs

[0509] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0510] 4. Information provision

[0511] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0512] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0513] 5. Feedback

[0514] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[0515] The server uses this data to improve the model.

[0516] Example 2: User B (Interest: Technology)

[0517] 1. Data Collection

[0518] User B browses technology-related news and the browsing history is sent to the server.

[0519] 2. Model Generation

[0520] The server trains an optimized technology-related AI model based on User B's interests.

[0521] 3. Introducing bugs

[0522] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0523] 4. Information provision

[0524] The server sends technology-related ads, as well as art-related ads, to the device.

[0525] 5. Feedback

[0526] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0527] In this way, by providing both information that is in line with the user's interests and unexpected information, this system promotes diversity in user behavior and provides opportunities for users to discover new interests and behaviors.

[0528] The processing flow will be explained below.

[0529] Step 1:

[0530] Users conduct searches, click on ads, and post activities on social media.

[0531] Step 2:

[0532] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[0533] Step 3:

[0534] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[0535] Step 4:

[0536] The server collects user behavior data from the database and preprocesses it (noise removal, data standardization, etc.).

[0537] Step 5:

[0538] The server generates and trains an AI model optimized for each user based on the preprocessed data.

[0539] Step 6:

[0540] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[0541] Step 7:

[0542] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[0543] Step 8:

[0544] The terminal displays to the user the information sent from the server, including both information based on the user's interests and unexpected information based on bug data.

[0545] Step 9:

[0546] The user browses the displayed information, clicks on information that interests them, or searches further.

[0547] Step 10:

[0548] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[0549] Step 11:

[0550] The server collects new behavioral data sent from the terminal and stores it in a database.

[0551] Step 12:

[0552] The server analyzes the new behavioral data and analyzes which bugs affected user behavior.

[0553] Step 13:

[0554] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates in future iterations.

[0555] In this way, the system collects and analyzes data at each step and provides users with a variety of information, thereby promoting diversity in user behavior.

[0556] Example 1

[0557] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0558] Conventional information provision systems focused on providing information based on users' interests and behavioral patterns, but lacked mechanisms for eliciting new interests and behaviors. This resulted in users being confined to their existing interests and having few opportunities to explore new fields of interest. Furthermore, the system did not effectively utilize a feedback loop to improve the accuracy of models based on user behavioral data.

[0559] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0560] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, a data storage and preprocessing means for storing the data in a database and preprocessing it, and a random bug generation means for generating bug data. This makes it possible to build an effective feedback loop that not only provides information based on the user's existing interests but also promotes the exploration of new interests and behaviors and contributes to improving the accuracy of the model.

[0561] The "data collection means" is a means for collecting user behavior data.

[0562] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[0563] "Bug injection" is a method of intentionally incorporating statistical bugs into a generated AI model to provide unexpected information to the user.

[0564] "Information provision means" refers to a means of providing information obtained using an AI model with a bug incorporated into it to a user terminal.

[0565] "Feedback means" refers to a means of recollecting user reactions and behavioral data and using it to improve the model.

[0566] "Data storage and preprocessing means" refers to the means for storing data in a database and preprocessing it.

[0567] The "random bug generation means" is a means for generating bug data.

[0568] "Behavioral data" refers to data such as a user's search history, ad click history, and social media activity.

[0569] An "AI model" is a model generated and trained by machine learning algorithms that analyzes and predicts user interests and behavioral patterns.

[0570] A "statistical bug" is noise or irrelevant information that is intentionally inserted to cause errors in the predictions of an AI model.

[0571] A "user terminal" is a terminal device such as a computer or mobile device used by a user.

[0572] MODE FOR CARRYING OUT THE INVENTION

[0573] This invention relates to a system that collects user behavioral data, intentionally incorporates statistical bugs into AI models trained based on this data, and provides users with unexpected information.

[0574] Data collection methods

[0575] The server collects user behavior data. Specifically, when a user uses a web or mobile application, the server sends data such as search history, ad click history, and social media activity from the device in real time. This allows the server to collect detailed information about the user's behavioral patterns.

[0576] Model Generation Method

[0577] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This process uses machine learning algorithms and performs data preprocessing such as filling in missing values, removing noise, and normalizing the data. For example, for User A, who is interested in fashion, an AI model is generated that prioritizes providing fashion-related information.

[0578] How to Introduce Bugs

[0579] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's hobbies and preferences, as well as random noise. A specific algorithm is applied to generate bug data, randomly generating unrelated data and inserting it into the AI ​​model. For example, "gardening" information is incorporated into the model of User A, who is interested in fashion.

[0580] Information provision means

[0581] The server then sends the information generated using the bugged AI model to the user's device, which receives it and displays it to the user. The user's device displays both information relevant to their interests and intentionally inserted unexpected information. For example, a gardening ad might appear alongside a fashion ad.

[0582] Feedback Methods

[0583] The device records behavioral data on the user's reactions to the displayed information (e.g., clicks, searches, etc.). This data is then sent back to the server and used to improve the AI ​​model. This feedback makes future information provision more suitable for the user. For example, if User A clicks on a gardening-related advertisement, the data is sent to the server and the model is adjusted.

[0584] Specific examples

[0585] Example 1: User A (Interest: Fashion)

[0586] 1. Data Collection

[0587] User A frequently searches for fashion-related sites and clicks data is collected.

[0588] This information is sent from the terminal to the server.

[0589] 2. Model Generation

[0590] The server trains an AI model optimized for User A based on fashion-related data.

[0591] 3. Introducing bugs

[0592] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0593] 4. Information provision

[0594] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0595] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0596] 5. Feedback

[0597] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the terminal and sent to the server.

[0598] The server uses this data to improve the model.

[0599] Example 2: User B (Interest: Technology)

[0600] 1. Data Collection

[0601] User B browses technology-related news and the browsing history is sent to the server.

[0602] 2. Model Generation

[0603] The server trains an optimized technology-related AI model based on User B's interests.

[0604] 3. Introducing bugs

[0605] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0606] 4. Information provision

[0607] The server sends technology-related ads, as well as art-related ads, to the device.

[0608] 5. Feedback

[0609] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0610] Prompt Sentence Examples

[0611] Example prompt for user A

[0612] "User A is interested in fashion, but the next information we show them should also include gardening information."

[0613] Example prompt for User B

[0614] "User B prefers technology-related information, but please also include some information about art."

[0615] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0616] Specific explanation of processing steps

[0617] Step 1: Data collection

[0618] The server collects user behavior data from the device. Input includes search terms performed by the user, ads clicked, social media posts and comments, etc. Using data collection tools, the server receives this data in real time and stores it in a database. Specifically, if a user searches for "latest fashion," the search term data is sent to the server. The output is the stored behavioral data.

[0619] Step 2: Data storage and preprocessing

[0620] The server stores the collected data in a database and performs preprocessing. The behavioral data collected in step 1 is used as input. The data storage and preprocessing means performs missing value completion, noise removal, and data normalization. For example, error values ​​are removed from the collected data to create up-to-date normalized fashion-related data. The output is the preprocessed data.

[0621] Step 3: Training the AI ​​model

[0622] The server trains an AI model based on the preprocessed data. The input is the preprocessed data generated in step 2. Using the model generation means, a machine learning algorithm generates a model that predicts the user's interests and behavioral patterns. Specifically, the server trains the AI ​​model based on the fashion data and creates a predictive model optimized for user A. The output is the trained AI model.

[0623] Step 4: Generate random bugs

[0624] The server generates random bugs for the trained AI model. The input is the AI ​​model generated in step 3. Using the random bug generation means, irrelevant data and noise are generated and incorporated into the model. For example, for a model of user A who is interested in fashion, irrelevant data related to gardening is randomly generated. The output is an AI model containing bugs.

[0625] Step 5: Incorporating bug data

[0626] The server incorporates the generated random bug data into the AI ​​model. The input is the random bug data generated in step 4. The bug injection means is used to insert irrelevant data into the model. Specifically, the server inserts "gardening" information into the AI ​​model. The output is an AI model with a bug injected.

[0627] Step 6: Generate customization information

[0628] The server generates customized information using the AI ​​model with the bug. The input is the AI ​​model with the bug created in step 5. The information to be displayed to the user is generated by predictions made through the model. Specifically, fashion advertisements and some gardening advertisements are generated. The output is customized information.

[0629] Step 7: Submit your information

[0630] The server sends the generated customization information to the user's terminal. The input is the customization information generated in step 6. The information is sent to the user's terminal using the information providing means. The terminal displays the received information on the screen. As a specific operation, a fashion advertisement and a gardening advertisement are displayed on User A's terminal. The output is the information displayed on the user's terminal.

[0631] Step 8: Recording behavioral data

[0632] The terminal records the user's reactions to the displayed information (clicks, searches, etc.). The input is the user's actions. This data is sent back to the server using feedback means. For example, if user A clicks on a gardening advertisement, the click information is recorded and sent to the server. The output is the recorded behavioral data.

[0633] Step 9: Improve the model

[0634] The server retrains the AI ​​model based on the transmitted behavioral data and improves the model. The input is the behavioral data transmitted in step 8. The model is adjusted and retrained using the model generation means. Specifically, the server improves the AI ​​model based on User A's click data and reflects this in the next information provision. The output is the improved AI model.

[0635] (Application example 1)

[0636] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0637] Conventional advertising delivery systems have focused on displaying advertisements based on users' interests, but this has the problem of narrowing the diversity of users' interests and behaviors. It also limits opportunities to discover unexpected information or new interests. Therefore, there is a need for systems that allow users to discover new interests and promote the diversity of their behaviors.

[0638] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0639] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected advertisements, an information provision means for providing the user's smartphone with advertisement information obtained using the AI ​​model with the bug injected, and a feedback means for re-collecting the user's response to advertisements and behavioral data and using it to improve the model, thereby enabling users to discover new interests and information and promoting behavioral diversity.

[0640] The "data collection means" is a means for collecting user behavior data.

[0641] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[0642] "Bug injection" refers to the intentional incorporation of statistical bugs into the generated AI model to provide users with unexpected advertisements.

[0643] The "information provision means" is a means of providing advertising information obtained using an AI model with a built-in bug to the user's smartphone.

[0644] "Feedback means" refers to a means of recollecting user responses to advertisements and behavioral data and using it to improve the model.

[0645] The "advertising delivery system" is a system that generates an AI model based on user behavior data and delivers advertisements that intentionally incorporate bugs to attract new user interest.

[0646] An "AI model" is an artificial intelligence model that is generated and trained based on user behavior data.

[0647] A "statistical bug" is unexpected information or random noise that is intentionally incorporated into an AI model.

[0648] "User behavioral data" refers to information such as a user's search history, ad click history, and social media activity.

[0649] A "smartphone" is a portable information terminal that is primarily used by users.

[0650] This invention is an advertising distribution system that collects user behavior data and intentionally incorporates statistical bugs into a trained AI model to provide users with unexpected advertisements. Specific embodiments of this system are described below.

[0651] System Overview

[0652] The system includes the following elements:

[0653] Data collection methods

[0654] Model Generation Method

[0655] How to Introduce Bugs

[0656] Information provision means

[0657] Feedback Methods

[0658] Data collection methods

[0659] The server collects user behavioral data, including search history, ad click history, social network activity, etc. This data is sent from the device to the server in real time when the user uses the smartphone.

[0660] Model Generation Method

[0661] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This AI model uses machine learning algorithms to analyze the user's interests and behavioral patterns and predict which advertisements should be displayed next.

[0662] How to Introduce Bugs

[0663] The server intentionally embeds statistical bugs into the generated AI model. This bug data includes advertising information unrelated to the user's interests and random noise. For this purpose, the server generates the bug data using a statistical algorithm.

[0664] Information provision means

[0665] The server then uses the bugged AI model to generate advertising information, which is then sent to the user's smartphone, which then receives the information and displays the ads to the user. This advertising information includes both ads tailored to the user's interests and intentionally inserted unexpected ads.

[0666] Feedback Methods

[0667] It records how users respond to the ads they see (for example, clicks, viewing time, etc.). This behavioral data is sent back to the server and used to improve the model. This feedback makes future ad serving more effective.

[0668] Specific examples

[0669] Example 1: User C (Interest: Technology)

[0670] 1. Data Collection

[0671] User C frequently browses technology-related content and their behavioral data is collected.

[0672] The data is sent from the smartphone to the server in real time.

[0673] 2. Model Generation

[0674] The server trains an AI model optimized for User C based on technology-related data.

[0675] 3. Introducing bugs

[0676] The server inserts advertising information for "sports," which differs from the technology, into the AI ​​model as a bug.

[0677] 4. Information provision

[0678] The server generates technology-related ads, along with some sports-related ads, and sends them to the smartphone.

[0679] When the smartphone displays this information, User C sees sports ads along with technology ads.

[0680] 5. Feedback

[0681] If User C is interested in a sports-related advertisement and clicks on it, his / her behavior is recorded and sent to the server.

[0682] The server uses this data to improve the model.

[0683] Prompt Sentence Examples

[0684] "Generate an AI model based on the categories of web pages frequently visited by users, and add ads for irrelevant categories with a set probability."

[0685] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0686] Step 1:

[0687] The server collects user behavior data. Specifically, it collects data in real time when users use their smartphones to browse websites or click on ads (search history, ad click history, social network activity, etc.). The input is user behavior data, and the output is the user behavior data stored on the server.

[0688] Step 2:

[0689] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. Specifically, it analyzes this behavioral data using a machine learning algorithm (e.g., RandomForestClassifier) ​​to build a model that predicts the next advertisement the user is likely to be interested in. The input is the collected behavioral data, and the output is the trained AI model.

[0690] Step 3:

[0691] The server intentionally incorporates statistical bugs into the generated AI model. Specifically, it applies an algorithm that adds random noise or advertising information unrelated to the user's interests (e.g., sports-related information) to the AI ​​model. The input is the trained AI model and the added bug data, and the output is an AI model with the bug incorporated.

[0692] Step 4:

[0693] The server uses the AI ​​model with the bug to generate advertising information and send it to the user's smartphone. Specifically, the advertising information generated by the AI ​​model is sent as a data packet to the smartphone, which receives and displays it. The input is the AI ​​model with the bug, and the output is the advertising information sent to the user's smartphone.

[0694] Step 5:

[0695] The device records how users respond to displayed ads (clicks, viewing time, etc.). Specifically, the smartphone application captures user responses as events and sends the data to a server. The input is the user's response to the ad, and the output is the feedback data sent to the server.

[0696] Step 6:

[0697] The server retrains and improves the AI ​​model based on the collected feedback data. Specifically, it adds the user's new behavioral data to the training dataset and updates the AI ​​model. The input is the new behavioral data including the feedback data, and the output is the improved AI model.

[0698] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0699] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with an emotion engine that recognizes the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[0700] System Overview

[0701] The system includes the following elements:

[0702] Data collection methods

[0703] Model Generation Method

[0704] How to Introduce Bugs

[0705] Information provision means

[0706] Feedback Methods

[0707] Emotion Engine

[0708] Data collection methods

[0709] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[0710] Model Generation Method

[0711] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[0712] How to Introduce Bugs

[0713] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[0714] Information provision means

[0715] The server then provides the user with information generated using the AI ​​model with the bugs. The device receives the information and displays it to the user. This information includes both information relevant to the user's interests and unexpected information based on the bug data.

[0716] Feedback Methods

[0717] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[0718] Emotion Engine

[0719] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The emotion engine acquires emotion data from the user's webcam, microphone, and text input. The server analyzes this data to determine the user's current emotional state.

[0720] The emotional data obtained by the emotion engine influences how information is presented to users. For example, if a user is excited, ads with content that matches that emotion will be dynamically served. Emotional data is also used as a feedback tool to help train AI models.

[0721] Specific examples

[0722] Example 1: User A (Interest: Fashion)

[0723] 1. Data Collection

[0724] User A frequently searches for fashion-related sites and clicks data is collected.

[0725] This information is sent from the terminal to the server.

[0726] 2. Model Generation

[0727] The server trains an AI model optimized for User A based on fashion-related data.

[0728] 3. Introducing bugs

[0729] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0730] 4. Information provision

[0731] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0732] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0733] 5. Feedback

[0734] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[0735] The server uses this data to improve the model.

[0736] 6. Emotion recognition

[0737] While User A is browsing the information, the emotion engine analyzes User A's emotions from their facial expressions and voice and sends the results to the server.

[0738] The server provides appropriate information to user A based on the emotion data.

[0739] Example 2: User B (Interest: Technology)

[0740] 1. Data Collection

[0741] User B browses technology-related news and the browsing history is sent to the server.

[0742] 2. Model Generation

[0743] The server trains an optimized technology-related AI model based on User B's interests.

[0744] 3. Introducing bugs

[0745] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0746] 4. Information provision

[0747] The server sends technology-related ads, as well as art-related ads, to the device.

[0748] 5. Feedback

[0749] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0750] 6. Emotion recognition

[0751] While User B is browsing the information, the emotion engine analyzes User B's emotions from his / her tone of voice and text input and sends the results to the server.

[0752] The server dynamically changes the advertisement content it provides based on the analyzed emotion data.

[0753] In this way, this system not only provides users with information that is in line with their interests and unexpected information, but also uses emotional data to provide more personalized information, thereby simultaneously improving the diversity of user behavior and satisfaction.

[0754] The processing flow will be explained below.

[0755] Step 1:

[0756] Users conduct searches, click on ads, and post activities on social media.

[0757] Step 2:

[0758] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[0759] Step 3:

[0760] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[0761] Step 4:

[0762] The server collects user behavior data from the database and preprocesses the data (noise removal, data standardization, etc.).

[0763] Step 5:

[0764] The server uses the preprocessed data to generate and train an AI model optimized for each user.

[0765] Step 6:

[0766] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[0767] Step 7:

[0768] Users express their emotions through a webcam and microphone.

[0769] Step 8:

[0770] The device collects the user's voice, facial expressions, and text in real time and sends them to the emotion engine.

[0771] Step 9:

[0772] The emotion engine analyzes the received data, identifies the user's emotional state, and sends the results to the server.

[0773] Step 10:

[0774] The server analyzes the emotion data sent from the emotion engine and dynamically adjusts the information and advertisements provided based on the user's emotional state.

[0775] Step 11:

[0776] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[0777] Step 12:

[0778] The terminal displays the information sent from the server to the user, including both information of interest to the user and unexpected information from bug data.

[0779] Step 13:

[0780] The user browses the displayed information, clicks on the information that interests them, or searches further.

[0781] Step 14:

[0782] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[0783] Step 15:

[0784] The server collects new behavioral data sent from the terminal and stores it in a database.

[0785] Step 16:

[0786] The server analyzes the new behavioral data and determines which bugs affected user behavior.

[0787] Step 17:

[0788] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates from the next time onwards.

[0789] In this way, the system collects and analyzes data at each step and provides users with a variety of information, promoting diverse user behavior. Furthermore, it uses an emotion engine to understand the user's emotional state and provide more personalized information.

[0790] Example 2

[0791] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0792] Modern information delivery systems are required to accurately predict user interests. However, typical systems lack the flexibility to respond to unexpected user behaviors and emotions, limiting the improvement of user experience. Furthermore, current systems have difficulty acquiring user emotional data in real time and providing information based on that data. Therefore, there is a need for a system that can simultaneously analyze a user's diverse behaviors and the emotions behind them, and dynamically provide appropriate information.

[0793] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0794] In this invention, the server includes means for collecting user behavioral data, means for generating and training a generative AI model optimized for the user's interests based on the collected behavioral data, means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected information, means for providing information obtained using the AI ​​model with the bugs incorporated to the user terminal, means for re-collecting user reaction and behavioral data and using it to improve the model, and emotion recognition means for analyzing emotions from the user's voice, facial expressions, and text to obtain emotion data. This makes it possible to efficiently analyze the user's diverse behavioral data and emotion data and provide the user with unexpected and useful information.

[0795] "User behavioral data" refers to data generated when a user uses the web or mobile applications, such as search history, ad click history, and social network service activity.

[0796] A "generative AI model" is a machine learning model built to analyze a user's interests and behavioral patterns based on collected user behavioral data and predict what information should be displayed next.

[0797] A "statistical bug" is data that contains information unrelated to the user's preferences or random noise that is intentionally incorporated into the generated AI model.

[0798] The "emotion recognition means" is an engine or algorithm for analyzing emotions from the user's voice, facial expressions, and text, and obtaining emotion data based on this.

[0799] "Information provision means" refers to a system or process that provides information generated using an AI model with a bug to a user's device and displays it to the user.

[0800] "Feedback means" refers to a means of recollecting the user's reactions to the displayed information (clicks, searches, etc.) and using this information to improve the AI ​​model.

[0801] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into a generative AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with emotion recognition means for recognizing the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[0802] System Overview

[0803] The system includes the following elements:

[0804] Behavioral data collection methods

[0805] Model Generation Method

[0806] How to Introduce Bugs

[0807] Information provision means

[0808] Feedback Methods

[0809] emotion recognition means

[0810] Behavioral data collection methods

[0811] The server collects behavioral data generated when users use web and mobile applications, such as search history, ad click history, social network service activity, etc. This behavioral data is collected in real time and transmitted to the server.

[0812] Model Generation Method

[0813] The server generates and trains a generative AI model optimized for the user's interests based on the collected behavioral data. This generative AI model uses machine learning algorithms (e.g., TensorFlow, PyTorch) to analyze the user's behavioral data and predict what information should be displayed next.

[0814] How to Introduce Bugs

[0815] The server intentionally incorporates statistical bugs into the trained generative AI model. This bug data includes information unrelated to the user's preferences and random noise, and uses a specific algorithm (e.g., a noise generation algorithm) to generate it.

[0816] Information provision means

[0817] The server provides information generated using the generative AI model with the bug to the user's device. The device receives this information and displays it to the user. The displayed information includes both information relevant to the user's interests and unexpected information based on the bug data.

[0818] Feedback Methods

[0819] The device records the user's reactions to the displayed information (clicks, searches, etc.) and sends the behavioral data back to the server. This feedback data is used to improve the AI ​​model, making future information provision even more effective.

[0820] emotion recognition means

[0821] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. Emotional data is acquired through the user's webcam, microphone, and text input, and the server analyzes this data to identify the user's current emotional state. The analyzed emotional data is also used for information provision and feedback means, making it possible to dynamically adjust information.

[0822] Examples of concrete examples and prompts

[0823] Example 1: User A (Interest: Fashion)

[0824] 1. Data Collection

[0825] User A frequently searches for fashion-related sites and clicks data is collected.

[0826] This information is sent from the terminal to the server.

[0827] 2. Model Generation

[0828] The server trains a generative AI model optimized for User A based on fashion-related data.

[0829] 3. Introducing bugs

[0830] The server inserts "gardening" information, which is different from fashion, into the generative AI model as a bug.

[0831] 4. Information provision

[0832] The server generates some gardening-related advertisements along with fashion-related advertisements and sends them to the terminal. When the terminal displays this information, User A sees the gardening advertisements along with the fashion advertisements.

[0833] 5. Feedback

[0834] If User A is interested and clicks on a gardening-related ad, that behavior is recorded by the device and sent to the server, which uses this data to improve the model.

[0835] 6. Emotion recognition

[0836] While User A is browsing information, the emotion recognition means analyzes User A's emotions from their facial expressions and voice and sends the results to the server. The server then provides appropriate information to User A based on the emotion data.

[0837] Example prompt sentence:

[0838] "User A is interested in fashion, but how will he react if we mix in random gardening information? Analyze the sentiment data as well."

[0839] Example 2: User B (Interest: Technology)

[0840] 1. Data Collection

[0841] User B browses technology-related news and the browsing history is sent to the server.

[0842] 2. Model Generation

[0843] The server trains an optimized generative AI model related to technology based on User B's interests.

[0844] 3. Introducing bugs

[0845] The server inserts "art" information, which is different from technology, into the generative AI model as bug data.

[0846] 4. Information provision

[0847] The server sends technology-related ads, as well as art-related ads, to the device.

[0848] 5. Feedback

[0849] If User B shows interest in an art-related ad and clicks on it, that data is recorded on the device and sent to the server, which uses this data to retrain the generative AI model.

[0850] 6. Emotion recognition

[0851] The emotion recognition means analyzes emotions from the voice and text input of User B and sends the data to the server. The server changes the advertisement content to be provided based on the analyzed emotion data.

[0852] Example prompt sentence:

[0853] "User B is interested in technology, but if you add some random art information, how will they behave? We'll take their sentiment data into account and do some analysis."

[0854] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0855] System program processing flow and processing steps

[0856] Step 1: Collect behavioral data

[0857] The server collects user behavioral data. As input, it obtains search history, ad click history, social network service activity, and other data in real time when users use web and mobile applications. This data is automatically sent from the device to the server when the user visits a specific website or clicks on a product. As output, the collected behavioral data is stored in a database.

[0858] Specific operation: A user searches a fashion-related website, enters the keyword "fall fashion," and clicks on several items. The data is sent from the device to the server in real time.

[0859] Step 2: Creating a generative AI model

[0860] The server creates and trains a generative AI model based on the collected behavioral data. It uses the behavioral data stored in the database as input. It uses a machine learning framework (e.g., TensorFlow, PyTorch) to analyze the user's interests and behavioral patterns. This analysis generates an AI model optimized for the user's interests. The output is a trained generative AI model.

[0861] How it works: The server analyzes fashion-related behavioral data and determines that "this user is interested in fall fashion." It then trains a generative AI model based on this interest.

[0862] Step 3: Incorporating the bug

[0863] The server intentionally incorporates bug data into a trained generative AI model. As input, it uses the trained generative AI model and a specific algorithm (e.g., a noise generation algorithm) to generate bug data. The bug data includes information unrelated to the user's preferences and random noise. As output, it obtains a generative AI model incorporating the bug data.

[0864] Specific operation: The server adds "gardening" information as a bug to the generative AI model, which generates "gardening"-related information along with fashion-related information.

[0865] Step 4: Provide information

[0866] The server uses a generative AI model with bugs built in to provide generated information to the terminal. As input, the server uses a generative AI model with bugs built in. The server uses this model to generate information based on the user's interests and unexpected information based on bug data. As output, the generated information is sent to the terminal and displayed to the user.

[0867] Specific operation: The server sends information including fashion-related advertisements and "gardening"-related advertisements to the terminal, which receives it and displays it to the user.

[0868] Step 5: Gather feedback

[0869] The user reacts to the displayed information (clicks, searches, etc.). The device records the user's reaction and sends it back to the server. The user's reaction data is used as input. As output, this reaction data is sent to the server and used to improve the AI ​​model.

[0870] How it works: When a user clicks on a gardening-related ad, the click data is recorded on the device and sent to the server, which uses this data to retrain the AI ​​model and improve the information provided next time.

[0871] Step 6: Emotion Recognition

[0872] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. As input, emotion data obtained through a webcam, microphone, and text input is used. The emotion recognition engine analyzes this data to identify the user's current emotional state. As output, the analyzed emotion data is sent to a server and used for information provision and feedback means.

[0873] Specific operation: While the user is browsing information using a webcam, the emotion recognition means collects facial expression data and detects an "excited" state. The server dynamically adjusts the information provided based on this data.

[0874] (Application example 2)

[0875] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0876] While conventional systems can provide information tailored to a user's interests, they lack the ability to intentionally provide different information to stimulate new interests. Furthermore, the information provided did not reflect the user's emotions, which could have further improved the quality of the user experience. In addition, the feedback loop for improving the AI ​​model based on collected data was not fully functional.

[0877] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0878] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, and an emotion engine for analyzing emotions from the user's voice, facial expressions, and text and providing information based on the emotion data. This makes it possible to provide advertisements that correspond to the user's emotions and discover new areas of interest.

[0879] The "data collection means" is a means for collecting user behavior data.

[0880] "Model generation means" means a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[0881] "Bug injection methods" are methods for intentionally incorporating statistical bugs into generated AI models to provide users with unexpected information.

[0882] "Information provision means" refers to a means for providing information obtained using an AI model incorporating a bug to a user terminal.

[0883] "Feedback means" refers to a means for recollecting user reactions and behavioral data and using it to improve the model.

[0884] The "emotion engine" is an engine that analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotion data.

[0885] This invention is a system that collects user behavioral data, intentionally incorporates statistical bugs into an AI model trained based on this data, and provides users with unexpected information, and further combines this with an emotion engine that recognizes the user's emotions.

[0886] The server implements the system using the following hardware and software.

[0887] Data collection method: Collect user behavior data (search history, ad click history, social media activity) using tools such as Google Analytics API and Mixpanel.

[0888] Model generation: Generate and train an AI model optimized for user interests based on collected behavioral data, using machine learning algorithms such as TensorFlow and Scikit-learn.

[0889] Bug injection: Intentionally injecting statistical bugs into the generated AI model to provide unexpected information. This involves using a specific algorithm to randomly generate noise or irrelevant information.

[0890] Information provision method: Information obtained using an AI model with a bug is provided to the user's device.

[0891] Feedback methods: Recollect user responses and behavioral data and use it to improve the AI ​​model. Specifically, use collection tools such as Firebase Analytics.

[0892] Emotion engine: Analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotional data. Specifically, it uses emotion recognition tools such as OpenCV and IBM Watson.

[0893] Example of User A (interested in fashion)

[0894] 1. Data collection: Collect data on User A's searches and clicks on fashion-related sites. The information is collected through the Google Analytics API.

[0895] 2. Model generation: Based on the collected data, the server trains an optimized fashion-related AI model using TensorFlow.

[0896] 3. Bug insertion: The server inserts information about "gardening," which is unrelated to fashion, into the AI ​​model as a bug.

[0897] 4. Information provision: The server generates fashion and some gardening-related advertisements and sends them to the user's terminal.

[0898] 5. Feedback: User A's response to the ad (click or ignore) is recorded and sent to the server via Firebase Analytics.

[0899] 6. Emotion recognition: While User A is browsing information, an emotion engine using OpenCV analyzes User A's facial expressions and voice and provides appropriate information based on that data.

[0900] Prompt Sentence Examples

[0901] "The user is interested in 'smartwatches,' so use an AI model to show them ads related to smartwatches. Also, intentionally show ads related to 'gardening,' and prioritize uplifting information because the user is emotionally excited."

[0902] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0903] Step 1:

[0904] The server uses data collection tools to collect user behavior data. Inputs include the user's search history, ad click history, and social media activity on web and mobile applications. This data is collected in real time through tools such as Google Analytics API and Mixpanel and stored in a database as behavioral data. The output is a compiled user behavior data.

[0905] Step 2:

[0906] The server generates and trains an AI model optimized for the user's interests based on the behavioral data collected using the model generation means. The behavioral data collected in step 1 is used as input. A machine learning algorithm (e.g., TensorFlow or Scikit-learn) is used to analyze the user's interests and behavioral patterns and generate a predictive model. The output is an optimized AI model.

[0907] Step 3:

[0908] The server intentionally embeds statistical bugs into the generated AI model using a bug injection method. The AI ​​model generated in step 2 is used as input. Random noise and irrelevant information (bug data) are generated using a specific algorithm and inserted into the model. The output is an AI model with embedded statistical bugs.

[0909] Step 4:

[0910] The server uses an information provision means to provide information generated based on the AI ​​model with the bug to the user's device. The AI ​​model obtained in step 3 is used as input. Based on the AI ​​model, advertisements and content containing unexpected information are generated and delivered to the device, taking into account the user's interests. The information displayed on the user's device is obtained as output.

[0911] Step 5:

[0912] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The inputs are provided by the user's webcam, microphone, and text input. For example, OpenCV or IBM Watson are used to collect emotion data and analyze it in real time. The output is the user's current emotional state, which is sent to the server.

[0913] Step 6:

[0914] The server reflects the emotional data obtained from the emotion engine and dynamically influences the information provided by the information providing means. For example, if the user is excited, it provides relaxing content, and if the user is uncomfortable, it provides information to excite the user. The emotional data obtained in step 5 is used as input. Information that matches the emotion is provided as output.

[0915] Step 7:

[0916] The server again collects user reactions and behavioral data through feedback channels and uses it to improve the model. The input is behavioral data, such as user click history and search history, and emotional data collected from the device. The data is accumulated and analyzed using collection tools such as Firebase Analytics. The output is an improved model that can be reflected in the next information provision.

[0917] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0918] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0919] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0920] [Third embodiment]

[0921] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0922] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0923] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0924] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0925] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0926] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0927] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0928] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0929] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0930] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0931] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0932] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0933] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user. Hereinafter, a specific embodiment of this invention will be described.

[0934] System Overview

[0935] The system includes the following elements:

[0936] Data collection methods

[0937] Model Generation Method

[0938] How to Introduce Bugs

[0939] Information provision means

[0940] Feedback Methods

[0941] Data collection methods

[0942] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[0943] Model Generation Method

[0944] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[0945] How to Introduce Bugs

[0946] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[0947] Information provision means

[0948] The server then sends the information generated by the bugged AI model to the user's device, which receives it and displays it to the user. This information includes both relevant information and intentionally inserted unexpected information.

[0949] Feedback Methods

[0950] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[0951] Specific examples

[0952] Example 1: User A (Interest: Fashion)

[0953] 1. Data Collection

[0954] User A frequently searches for fashion-related sites and clicks data is collected.

[0955] This information is sent from the terminal to the server.

[0956] 2. Model Generation

[0957] The server trains an AI model optimized for User A based on fashion-related data.

[0958] 3. Introducing bugs

[0959] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[0960] 4. Information provision

[0961] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[0962] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[0963] 5. Feedback

[0964] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[0965] The server uses this data to improve the model.

[0966] Example 2: User B (Interest: Technology)

[0967] 1. Data Collection

[0968] User B browses technology-related news and the browsing history is sent to the server.

[0969] 2. Model Generation

[0970] The server trains an optimized technology-related AI model based on User B's interests.

[0971] 3. Introducing bugs

[0972] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[0973] 4. Information provision

[0974] The server sends technology-related ads, as well as art-related ads, to the device.

[0975] 5. Feedback

[0976] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[0977] In this way, by providing both information that is in line with the user's interests and unexpected information, this system promotes diversity in user behavior and provides opportunities for users to discover new interests and behaviors.

[0978] The processing flow will be explained below.

[0979] Step 1:

[0980] Users conduct searches, click on ads, and post activities on social media.

[0981] Step 2:

[0982] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[0983] Step 3:

[0984] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[0985] Step 4:

[0986] The server collects user behavior data from the database and preprocesses it (noise removal, data standardization, etc.).

[0987] Step 5:

[0988] The server generates and trains an AI model optimized for each user based on the preprocessed data.

[0989] Step 6:

[0990] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[0991] Step 7:

[0992] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[0993] Step 8:

[0994] The terminal displays to the user the information sent from the server, including both information based on the user's interests and unexpected information based on bug data.

[0995] Step 9:

[0996] The user browses the displayed information, clicks on information that interests them, or searches further.

[0997] Step 10:

[0998] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[0999] Step 11:

[1000] The server collects new behavioral data sent from the terminal and stores it in a database.

[1001] Step 12:

[1002] The server analyzes the new behavioral data and analyzes which bugs affected user behavior.

[1003] Step 13:

[1004] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates in future iterations.

[1005] In this way, the system collects and analyzes data at each step and provides users with a variety of information, thereby promoting diversity in user behavior.

[1006] Example 1

[1007] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1008] Conventional information provision systems focused on providing information based on users' interests and behavioral patterns, but lacked mechanisms for eliciting new interests and behaviors. This resulted in users being confined to their existing interests and having few opportunities to explore new fields of interest. Furthermore, the system did not effectively utilize a feedback loop to improve the accuracy of models based on user behavioral data.

[1009] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1010] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, a data storage and preprocessing means for storing the data in a database and preprocessing it, and a random bug generation means for generating bug data. This makes it possible to build an effective feedback loop that not only provides information based on the user's existing interests but also promotes the exploration of new interests and behaviors and contributes to improving the accuracy of the model.

[1011] The "data collection means" is a means for collecting user behavior data.

[1012] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[1013] "Bug injection" is a method of intentionally incorporating statistical bugs into a generated AI model to provide unexpected information to the user.

[1014] "Information provision means" refers to a means of providing information obtained using an AI model with a bug incorporated into it to a user terminal.

[1015] "Feedback means" refers to a means of recollecting user reactions and behavioral data and using it to improve the model.

[1016] "Data storage and preprocessing means" refers to the means for storing data in a database and preprocessing it.

[1017] The "random bug generation means" is a means for generating bug data.

[1018] "Behavioral data" refers to data such as a user's search history, ad click history, and social media activity.

[1019] An "AI model" is a model generated and trained by machine learning algorithms that analyzes and predicts user interests and behavioral patterns.

[1020] A "statistical bug" is noise or irrelevant information that is intentionally inserted to cause errors in the predictions of an AI model.

[1021] A "user terminal" is a terminal device such as a computer or mobile device used by a user.

[1022] MODE FOR CARRYING OUT THE INVENTION

[1023] This invention relates to a system that collects user behavioral data, intentionally incorporates statistical bugs into AI models trained based on this data, and provides users with unexpected information.

[1024] Data collection methods

[1025] The server collects user behavior data. Specifically, when a user uses a web or mobile application, the server sends data such as search history, ad click history, and social media activity from the device in real time. This allows the server to collect detailed information about the user's behavioral patterns.

[1026] Model Generation Method

[1027] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This process uses machine learning algorithms and performs data preprocessing such as filling in missing values, removing noise, and normalizing the data. For example, for User A, who is interested in fashion, an AI model is generated that prioritizes providing fashion-related information.

[1028] How to Introduce Bugs

[1029] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's hobbies and preferences, as well as random noise. A specific algorithm is applied to generate bug data, randomly generating unrelated data and inserting it into the AI ​​model. For example, "gardening" information is incorporated into the model of User A, who is interested in fashion.

[1030] Information provision means

[1031] The server then sends the information generated using the bugged AI model to the user's device, which receives it and displays it to the user. The user's device displays both information relevant to their interests and intentionally inserted unexpected information. For example, a gardening ad might appear alongside a fashion ad.

[1032] Feedback Methods

[1033] The device records behavioral data on the user's reactions to the displayed information (e.g., clicks, searches, etc.). This data is then sent back to the server and used to improve the AI ​​model. This feedback makes future information provision more suitable for the user. For example, if User A clicks on a gardening-related advertisement, the data is sent to the server and the model is adjusted.

[1034] Specific examples

[1035] Example 1: User A (Interest: Fashion)

[1036] 1. Data Collection

[1037] User A frequently searches for fashion-related sites and clicks data is collected.

[1038] This information is sent from the terminal to the server.

[1039] 2. Model Generation

[1040] The server trains an AI model optimized for User A based on fashion-related data.

[1041] 3. Introducing bugs

[1042] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[1043] 4. Information provision

[1044] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[1045] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[1046] 5. Feedback

[1047] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the terminal and sent to the server.

[1048] The server uses this data to improve the model.

[1049] Example 2: User B (Interest: Technology)

[1050] 1. Data Collection

[1051] User B browses technology-related news and the browsing history is sent to the server.

[1052] 2. Model Generation

[1053] The server trains an optimized technology-related AI model based on User B's interests.

[1054] 3. Introducing bugs

[1055] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[1056] 4. Information provision

[1057] The server sends technology-related ads, as well as art-related ads, to the device.

[1058] 5. Feedback

[1059] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[1060] Prompt Sentence Examples

[1061] Example prompt for user A

[1062] "User A is interested in fashion, but the next information we show them should also include gardening information."

[1063] Example prompt for User B

[1064] "User B prefers technology-related information, but please also include some information about art."

[1065] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1066] Specific explanation of processing steps

[1067] Step 1: Data collection

[1068] The server collects user behavior data from the device. Input includes search terms performed by the user, ads clicked, social media posts and comments, etc. Using data collection tools, the server receives this data in real time and stores it in a database. Specifically, if a user searches for "latest fashion," the search term data is sent to the server. The output is the stored behavioral data.

[1069] Step 2: Data storage and preprocessing

[1070] The server stores the collected data in a database and performs preprocessing. The behavioral data collected in step 1 is used as input. The data storage and preprocessing means performs missing value completion, noise removal, and data normalization. For example, error values ​​are removed from the collected data to create up-to-date normalized fashion-related data. The output is the preprocessed data.

[1071] Step 3: Training the AI ​​model

[1072] The server trains an AI model based on the preprocessed data. The input is the preprocessed data generated in step 2. Using the model generation means, a machine learning algorithm generates a model that predicts the user's interests and behavioral patterns. Specifically, the server trains the AI ​​model based on the fashion data and creates a predictive model optimized for user A. The output is the trained AI model.

[1073] Step 4: Generate random bugs

[1074] The server generates random bugs for the trained AI model. The input is the AI ​​model generated in step 3. Using the random bug generation means, irrelevant data and noise are generated and incorporated into the model. For example, for a model of user A who is interested in fashion, irrelevant data related to gardening is randomly generated. The output is an AI model containing bugs.

[1075] Step 5: Incorporating bug data

[1076] The server incorporates the generated random bug data into the AI ​​model. The input is the random bug data generated in step 4. The bug injection means is used to insert irrelevant data into the model. Specifically, the server inserts "gardening" information into the AI ​​model. The output is an AI model with a bug injected.

[1077] Step 6: Generate customization information

[1078] The server generates customized information using the AI ​​model with the bug. The input is the AI ​​model with the bug created in step 5. The information to be displayed to the user is generated by predictions made through the model. Specifically, fashion advertisements and some gardening advertisements are generated. The output is customized information.

[1079] Step 7: Submit your information

[1080] The server sends the generated customization information to the user's terminal. The input is the customization information generated in step 6. The information is sent to the user's terminal using the information providing means. The terminal displays the received information on the screen. As a specific operation, a fashion advertisement and a gardening advertisement are displayed on User A's terminal. The output is the information displayed on the user's terminal.

[1081] Step 8: Recording behavioral data

[1082] The terminal records the user's reactions to the displayed information (clicks, searches, etc.). The input is the user's actions. This data is sent back to the server using feedback means. For example, if user A clicks on a gardening advertisement, the click information is recorded and sent to the server. The output is the recorded behavioral data.

[1083] Step 9: Improve the model

[1084] The server retrains the AI ​​model based on the transmitted behavioral data and improves the model. The input is the behavioral data transmitted in step 8. The model is adjusted and retrained using the model generation means. Specifically, the server improves the AI ​​model based on User A's click data and reflects this in the next information provision. The output is the improved AI model.

[1085] (Application example 1)

[1086] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1087] Conventional advertising delivery systems have focused on displaying advertisements based on users' interests, but this has the problem of narrowing the diversity of users' interests and behaviors. It also limits opportunities to discover unexpected information or new interests. Therefore, there is a need for systems that allow users to discover new interests and promote the diversity of their behaviors.

[1088] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1089] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected advertisements, an information provision means for providing the user's smartphone with advertisement information obtained using the AI ​​model with the bug injected, and a feedback means for re-collecting the user's response to advertisements and behavioral data and using it to improve the model, thereby enabling users to discover new interests and information and promoting behavioral diversity.

[1090] The "data collection means" is a means for collecting user behavior data.

[1091] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[1092] "Bug injection" refers to the intentional incorporation of statistical bugs into the generated AI model to provide users with unexpected advertisements.

[1093] The "information provision means" is a means of providing advertising information obtained using an AI model with a built-in bug to the user's smartphone.

[1094] "Feedback means" refers to a means of recollecting user responses to advertisements and behavioral data and using it to improve the model.

[1095] The "advertising delivery system" is a system that generates an AI model based on user behavior data and delivers advertisements that intentionally incorporate bugs to attract new user interest.

[1096] An "AI model" is an artificial intelligence model that is generated and trained based on user behavior data.

[1097] A "statistical bug" is unexpected information or random noise that is intentionally incorporated into an AI model.

[1098] "User behavioral data" refers to information such as a user's search history, ad click history, and social media activity.

[1099] A "smartphone" is a portable information terminal that is primarily used by users.

[1100] This invention is an advertising distribution system that collects user behavior data and intentionally incorporates statistical bugs into a trained AI model to provide users with unexpected advertisements. Specific embodiments of this system are described below.

[1101] System Overview

[1102] The system includes the following elements:

[1103] Data collection methods

[1104] Model Generation Method

[1105] How to Introduce Bugs

[1106] Information provision means

[1107] Feedback Methods

[1108] Data collection methods

[1109] The server collects user behavioral data, including search history, ad click history, social network activity, etc. This data is sent from the device to the server in real time when the user uses the smartphone.

[1110] Model Generation Method

[1111] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This AI model uses machine learning algorithms to analyze the user's interests and behavioral patterns and predict which advertisements should be displayed next.

[1112] How to Introduce Bugs

[1113] The server intentionally embeds statistical bugs into the generated AI model. This bug data includes advertising information unrelated to the user's interests and random noise. For this purpose, the server generates the bug data using a statistical algorithm.

[1114] Information provision means

[1115] The server then uses the bugged AI model to generate advertising information, which is then sent to the user's smartphone, which then receives the information and displays the ads to the user. This advertising information includes both ads tailored to the user's interests and intentionally inserted unexpected ads.

[1116] Feedback Methods

[1117] It records how users respond to the ads they see (for example, clicks, viewing time, etc.). This behavioral data is sent back to the server and used to improve the model. This feedback makes future ad serving more effective.

[1118] Specific examples

[1119] Example 1: User C (Interest: Technology)

[1120] 1. Data Collection

[1121] User C frequently browses technology-related content and their behavioral data is collected.

[1122] The data is sent from the smartphone to the server in real time.

[1123] 2. Model Generation

[1124] The server trains an AI model optimized for User C based on technology-related data.

[1125] 3. Introducing bugs

[1126] The server inserts advertising information for "sports," which differs from the technology, into the AI ​​model as a bug.

[1127] 4. Information provision

[1128] The server generates technology-related ads, along with some sports-related ads, and sends them to the smartphone.

[1129] When the smartphone displays this information, User C sees sports ads along with technology ads.

[1130] 5. Feedback

[1131] If User C is interested in a sports-related advertisement and clicks on it, his / her behavior is recorded and sent to the server.

[1132] The server uses this data to improve the model.

[1133] Prompt Sentence Examples

[1134] "Generate an AI model based on the categories of web pages frequently visited by users, and add ads for irrelevant categories with a set probability."

[1135] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1136] Step 1:

[1137] The server collects user behavior data. Specifically, it collects data in real time when users use their smartphones to browse websites or click on ads (search history, ad click history, social network activity, etc.). The input is user behavior data, and the output is the user behavior data stored on the server.

[1138] Step 2:

[1139] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. Specifically, it analyzes this behavioral data using a machine learning algorithm (e.g., RandomForestClassifier) ​​to build a model that predicts the next advertisement the user is likely to be interested in. The input is the collected behavioral data, and the output is the trained AI model.

[1140] Step 3:

[1141] The server intentionally incorporates statistical bugs into the generated AI model. Specifically, it applies an algorithm that adds random noise or advertising information unrelated to the user's interests (e.g., sports-related information) to the AI ​​model. The input is the trained AI model and the added bug data, and the output is an AI model with the bug incorporated.

[1142] Step 4:

[1143] The server uses the AI ​​model with the bug to generate advertising information and send it to the user's smartphone. Specifically, the advertising information generated by the AI ​​model is sent as a data packet to the smartphone, which receives and displays it. The input is the AI ​​model with the bug, and the output is the advertising information sent to the user's smartphone.

[1144] Step 5:

[1145] The device records how users respond to displayed ads (clicks, viewing time, etc.). Specifically, the smartphone application captures user responses as events and sends the data to a server. The input is the user's response to the ad, and the output is the feedback data sent to the server.

[1146] Step 6:

[1147] The server retrains and improves the AI ​​model based on the collected feedback data. Specifically, it adds the user's new behavioral data to the training dataset and updates the AI ​​model. The input is the new behavioral data including the feedback data, and the output is the improved AI model.

[1148] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1149] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with an emotion engine that recognizes the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[1150] System Overview

[1151] The system includes the following elements:

[1152] Data collection methods

[1153] Model Generation Method

[1154] How to Introduce Bugs

[1155] Information provision means

[1156] Feedback Methods

[1157] Emotion Engine

[1158] Data collection methods

[1159] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[1160] Model Generation Method

[1161] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[1162] How to Introduce Bugs

[1163] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[1164] Information provision means

[1165] The server then provides the user with information generated using the AI ​​model with the bugs. The device receives the information and displays it to the user. This information includes both information relevant to the user's interests and unexpected information based on the bug data.

[1166] Feedback Methods

[1167] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[1168] Emotion Engine

[1169] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The emotion engine acquires emotion data from the user's webcam, microphone, and text input. The server analyzes this data to determine the user's current emotional state.

[1170] The emotional data obtained by the emotion engine influences how information is presented to users. For example, if a user is excited, ads with content that matches that emotion will be dynamically served. Emotional data is also used as a feedback tool to help train AI models.

[1171] Specific examples

[1172] Example 1: User A (Interest: Fashion)

[1173] 1. Data Collection

[1174] User A frequently searches for fashion-related sites and clicks data is collected.

[1175] This information is sent from the terminal to the server.

[1176] 2. Model Generation

[1177] The server trains an AI model optimized for User A based on fashion-related data.

[1178] 3. Introducing bugs

[1179] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[1180] 4. Information provision

[1181] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[1182] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[1183] 5. Feedback

[1184] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[1185] The server uses this data to improve the model.

[1186] 6. Emotion recognition

[1187] While User A is browsing the information, the emotion engine analyzes User A's emotions from their facial expressions and voice and sends the results to the server.

[1188] The server provides appropriate information to User A based on the emotion data.

[1189] Example 2: User B (Interest: Technology)

[1190] 1. Data Collection

[1191] User B browses technology-related news and the browsing history is sent to the server.

[1192] 2. Model Generation

[1193] The server trains an optimized technology-related AI model based on User B's interests.

[1194] 3. Introducing bugs

[1195] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[1196] 4. Information provision

[1197] The server sends technology-related ads, as well as art-related ads, to the device.

[1198] 5. Feedback

[1199] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[1200] 6. Emotion recognition

[1201] While User B is browsing the information, the emotion engine analyzes User B's emotions from his / her tone of voice and text input and sends the results to the server.

[1202] The server dynamically changes the advertisement content it provides based on the analyzed emotion data.

[1203] In this way, this system not only provides users with information that is in line with their interests and unexpected information, but also uses emotional data to provide more personalized information, thereby simultaneously improving the diversity of user behavior and satisfaction.

[1204] The processing flow will be explained below.

[1205] Step 1:

[1206] Users conduct searches, click on ads, and post activities on social media.

[1207] Step 2:

[1208] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[1209] Step 3:

[1210] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[1211] Step 4:

[1212] The server collects user behavior data from the database and preprocesses the data (noise removal, data standardization, etc.).

[1213] Step 5:

[1214] The server uses the preprocessed data to generate and train an AI model optimized for each user.

[1215] Step 6:

[1216] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[1217] Step 7:

[1218] Users express their emotions through a webcam and microphone.

[1219] Step 8:

[1220] The device collects the user's voice, facial expressions, and text in real time and sends them to the emotion engine.

[1221] Step 9:

[1222] The emotion engine analyzes the received data, identifies the user's emotional state, and sends the results to the server.

[1223] Step 10:

[1224] The server analyzes the emotion data sent from the emotion engine and dynamically adjusts the information and advertisements provided based on the user's emotional state.

[1225] Step 11:

[1226] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[1227] Step 12:

[1228] The terminal displays the information sent from the server to the user, including both information of interest to the user and unexpected information from bug data.

[1229] Step 13:

[1230] The user browses the displayed information, clicks on the information that interests them, or searches further.

[1231] Step 14:

[1232] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[1233] Step 15:

[1234] The server collects new behavioral data sent from the terminal and stores it in a database.

[1235] Step 16:

[1236] The server analyzes the new behavioral data and determines which bugs affected user behavior.

[1237] Step 17:

[1238] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates from the next time onwards.

[1239] In this way, the system collects and analyzes data at each step and provides users with a variety of information, promoting diverse user behavior. Furthermore, it uses an emotion engine to understand the user's emotional state and provide more personalized information.

[1240] Example 2

[1241] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1242] Modern information delivery systems are required to accurately predict user interests. However, typical systems lack the flexibility to respond to unexpected user behaviors and emotions, limiting the improvement of user experience. Furthermore, current systems have difficulty acquiring user emotional data in real time and providing information based on that data. Therefore, there is a need for a system that can simultaneously analyze a user's diverse behaviors and the emotions behind them, and dynamically provide appropriate information.

[1243] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1244] In this invention, the server includes means for collecting user behavioral data, means for generating and training a generative AI model optimized for the user's interests based on the collected behavioral data, means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected information, means for providing information obtained using the AI ​​model with the bugs incorporated to the user terminal, means for re-collecting user reaction and behavioral data and using it to improve the model, and emotion recognition means for analyzing emotions from the user's voice, facial expressions, and text to obtain emotion data. This makes it possible to efficiently analyze the user's diverse behavioral data and emotion data and provide the user with unexpected and useful information.

[1245] "User behavioral data" refers to data generated when a user uses the web or mobile applications, such as search history, ad click history, and social network service activity.

[1246] A "generative AI model" is a machine learning model built to analyze a user's interests and behavioral patterns based on collected user behavioral data and predict what information should be displayed next.

[1247] A "statistical bug" is data that contains information unrelated to the user's preferences or random noise that is intentionally incorporated into the generated AI model.

[1248] The "emotion recognition means" is an engine or algorithm for analyzing emotions from the user's voice, facial expressions, and text, and obtaining emotion data based on this.

[1249] "Information provision means" refers to a system or process that provides information generated using an AI model with a bug to a user's device and displays it to the user.

[1250] "Feedback means" refers to a means of recollecting the user's reactions to the displayed information (clicks, searches, etc.) and using this information to improve the AI ​​model.

[1251] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into a generative AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with emotion recognition means for recognizing the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[1252] System Overview

[1253] The system includes the following elements:

[1254] Behavioral data collection methods

[1255] Model Generation Method

[1256] How to Introduce Bugs

[1257] Information provision means

[1258] Feedback Methods

[1259] emotion recognition means

[1260] Behavioral data collection methods

[1261] The server collects behavioral data generated when users use web and mobile applications, such as search history, ad click history, social network service activity, etc. This behavioral data is collected in real time and transmitted to the server.

[1262] Model Generation Method

[1263] The server generates and trains a generative AI model optimized for the user's interests based on the collected behavioral data. This generative AI model uses machine learning algorithms (e.g., TensorFlow, PyTorch) to analyze the user's behavioral data and predict what information should be displayed next.

[1264] How to Introduce Bugs

[1265] The server intentionally incorporates statistical bugs into the trained generative AI model. This bug data includes information unrelated to the user's preferences and random noise, and uses a specific algorithm (e.g., a noise generation algorithm) to generate it.

[1266] Information provision means

[1267] The server provides information generated using the generative AI model with the bug to the user's device. The device receives this information and displays it to the user. The displayed information includes both information relevant to the user's interests and unexpected information based on the bug data.

[1268] Feedback Methods

[1269] The device records the user's reactions to the displayed information (clicks, searches, etc.) and sends the behavioral data back to the server. This feedback data is used to improve the AI ​​model, making future information provision even more effective.

[1270] emotion recognition means

[1271] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. Emotional data is acquired through the user's webcam, microphone, and text input, and the server analyzes this data to identify the user's current emotional state. The analyzed emotional data is also used for information provision and feedback means, making it possible to dynamically adjust information.

[1272] Examples of concrete examples and prompts

[1273] Example 1: User A (Interest: Fashion)

[1274] 1. Data Collection

[1275] User A frequently searches for fashion-related sites and clicks data is collected.

[1276] This information is sent from the terminal to the server.

[1277] 2. Model Generation

[1278] The server trains a generative AI model optimized for User A based on fashion-related data.

[1279] 3. Introducing bugs

[1280] The server inserts "gardening" information, which is different from fashion, into the generative AI model as a bug.

[1281] 4. Information provision

[1282] The server generates some gardening-related advertisements along with fashion-related advertisements and sends them to the terminal. When the terminal displays this information, User A sees the gardening advertisements along with the fashion advertisements.

[1283] 5. Feedback

[1284] If User A is interested and clicks on a gardening-related ad, that behavior is recorded by the device and sent to the server, which uses this data to improve the model.

[1285] 6. Emotion recognition

[1286] While User A is browsing information, the emotion recognition means analyzes User A's emotions from their facial expressions and voice and sends the results to the server. The server then provides appropriate information to User A based on the emotion data.

[1287] Example prompt sentence:

[1288] "User A is interested in fashion, but how will he react if we mix in random gardening information? Analyze the sentiment data as well."

[1289] Example 2: User B (Interest: Technology)

[1290] 1. Data Collection

[1291] User B browses technology-related news and the browsing history is sent to the server.

[1292] 2. Model Generation

[1293] The server trains an optimized generative AI model related to technology based on User B's interests.

[1294] 3. Introducing bugs

[1295] The server inserts "art" information, which is different from technology, into the generative AI model as bug data.

[1296] 4. Information provision

[1297] The server sends technology-related ads, as well as art-related ads, to the device.

[1298] 5. Feedback

[1299] If User B shows interest in an art-related ad and clicks on it, that data is recorded on the device and sent to the server, which uses this data to retrain the generative AI model.

[1300] 6. Emotion recognition

[1301] The emotion recognition means analyzes emotions from the voice and text input of User B and sends the data to the server. The server changes the advertisement content to be provided based on the analyzed emotion data.

[1302] Example prompt sentence:

[1303] "User B is interested in technology, but if you add some random art information, how will they behave? We'll take their sentiment data into account and do some analysis."

[1304] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1305] System program processing flow and processing steps

[1306] Step 1: Collect behavioral data

[1307] The server collects user behavioral data. As input, it obtains search history, ad click history, social network service activity, and other data in real time when users use web and mobile applications. This data is automatically sent from the device to the server when the user visits a specific website or clicks on a product. As output, the collected behavioral data is stored in a database.

[1308] Specific operation: A user searches a fashion-related website, enters the keyword "fall fashion," and clicks on several items. The data is sent from the device to the server in real time.

[1309] Step 2: Creating a generative AI model

[1310] The server creates and trains a generative AI model based on the collected behavioral data. It uses the behavioral data stored in the database as input. It uses a machine learning framework (e.g., TensorFlow, PyTorch) to analyze the user's interests and behavioral patterns. This analysis generates an AI model optimized for the user's interests. The output is a trained generative AI model.

[1311] How it works: The server analyzes fashion-related behavioral data and determines that "this user is interested in fall fashion." It then trains a generative AI model based on this interest.

[1312] Step 3: Incorporating the bug

[1313] The server intentionally incorporates bug data into a trained generative AI model. As input, it uses the trained generative AI model and a specific algorithm (e.g., a noise generation algorithm) to generate bug data. The bug data includes information unrelated to the user's preferences and random noise. As output, it obtains a generative AI model incorporating the bug data.

[1314] Specific operation: The server adds "gardening" information as a bug to the generative AI model, which generates "gardening"-related information along with fashion-related information.

[1315] Step 4: Provide information

[1316] The server uses a generative AI model with bugs built in to provide generated information to the terminal. As input, the server uses a generative AI model with bugs built in. The server uses this model to generate information based on the user's interests and unexpected information based on bug data. As output, the generated information is sent to the terminal and displayed to the user.

[1317] Specific operation: The server sends information including fashion-related advertisements and "gardening"-related advertisements to the terminal, which receives it and displays it to the user.

[1318] Step 5: Gather feedback

[1319] The user reacts to the displayed information (clicks, searches, etc.). The device records the user's reaction and sends it back to the server. The user's reaction data is used as input. As output, this reaction data is sent to the server and used to improve the AI ​​model.

[1320] How it works: When a user clicks on a gardening-related ad, the click data is recorded on the device and sent to the server, which uses this data to retrain the AI ​​model and improve the information provided next time.

[1321] Step 6: Emotion Recognition

[1322] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. As input, emotion data obtained through a webcam, microphone, and text input is used. The emotion recognition engine analyzes this data to identify the user's current emotional state. As output, the analyzed emotion data is sent to a server and used for information provision and feedback means.

[1323] Specific operation: While the user is browsing information using a webcam, the emotion recognition means collects facial expression data and detects an "excited" state. The server dynamically adjusts the information provided based on this data.

[1324] (Application example 2)

[1325] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1326] While conventional systems can provide information tailored to a user's interests, they lack the ability to intentionally provide different information to stimulate new interests. Furthermore, the information provided did not reflect the user's emotions, which could have further improved the quality of the user experience. In addition, the feedback loop for improving the AI ​​model based on collected data was not fully functional.

[1327] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1328] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, and an emotion engine for analyzing emotions from the user's voice, facial expressions, and text and providing information based on the emotion data. This makes it possible to provide advertisements that correspond to the user's emotions and discover new areas of interest.

[1329] The "data collection means" is a means for collecting user behavior data.

[1330] "Model generation means" means a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[1331] "Bug injection methods" are methods for intentionally incorporating statistical bugs into generated AI models to provide users with unexpected information.

[1332] "Information provision means" refers to a means for providing information obtained using an AI model incorporating a bug to a user terminal.

[1333] "Feedback means" refers to a means for recollecting user reactions and behavioral data and using it to improve the model.

[1334] The "emotion engine" is an engine that analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotion data.

[1335] This invention is a system that collects user behavioral data, intentionally incorporates statistical bugs into an AI model trained based on this data, and provides users with unexpected information, and further combines this with an emotion engine that recognizes the user's emotions.

[1336] The server implements the system using the following hardware and software.

[1337] Data collection method: Collect user behavior data (search history, ad click history, social media activity) using tools such as Google Analytics API and Mixpanel.

[1338] Model generation method: Generate and train an AI model optimized for user interests based on collected behavioral data, using machine learning algorithms such as TensorFlow and Scikit-learn.

[1339] Bug injection: Intentionally injecting statistical bugs into the generated AI model to provide unexpected information. This involves using a specific algorithm to randomly generate noise or irrelevant information.

[1340] Information provision method: Information obtained using an AI model with a bug is provided to the user's device.

[1341] Feedback methods: Recollect user reactions and behavioral data and use it to improve the AI ​​model. Specifically, use collection tools such as Firebase Analytics.

[1342] Emotion engine: Analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotional data. Specifically, it uses emotion recognition tools such as OpenCV and IBM Watson.

[1343] Example of User A (interested in fashion)

[1344] 1. Data collection: Collect data on User A's searches and clicks on fashion-related sites. The information is collected through the Google Analytics API.

[1345] 2. Model generation: Based on the collected data, the server trains an optimized fashion-related AI model using TensorFlow.

[1346] 3. Bug insertion: The server inserts information about "gardening," which is unrelated to fashion, into the AI ​​model as a bug.

[1347] 4. Information provision: The server generates fashion and some gardening-related advertisements and sends them to the user's terminal.

[1348] 5. Feedback: User A's response to the ad (click or ignore) is recorded and sent to the server via Firebase Analytics.

[1349] 6. Emotion recognition: While User A is browsing information, an emotion engine using OpenCV analyzes User A's facial expressions and voice and provides appropriate information based on that data.

[1350] Prompt Sentence Examples

[1351] "The user is interested in 'smartwatches,' so use an AI model to show them ads related to smartwatches. Also, intentionally show ads related to 'gardening,' and prioritize uplifting information because the user is emotionally excited."

[1352] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1353] Step 1:

[1354] The server uses data collection tools to collect user behavior data. Inputs include the user's search history, ad click history, and social media activity on web and mobile applications. This data is collected in real time through tools such as Google Analytics API and Mixpanel and stored in a database as behavioral data. The output is a compiled user behavior data.

[1355] Step 2:

[1356] The server generates and trains an AI model optimized for the user's interests based on the behavioral data collected using the model generation means. The behavioral data collected in step 1 is used as input. A machine learning algorithm (e.g., TensorFlow or Scikit-learn) is used to analyze the user's interests and behavioral patterns and generate a predictive model. The output is an optimized AI model.

[1357] Step 3:

[1358] The server intentionally embeds statistical bugs into the generated AI model using a bug injection method. The AI ​​model generated in step 2 is used as input. Random noise and irrelevant information (bug data) are generated using a specific algorithm and inserted into the model. The output is an AI model with embedded statistical bugs.

[1359] Step 4:

[1360] The server uses an information provision means to provide information generated based on the AI ​​model with the bug to the user's device. The AI ​​model obtained in step 3 is used as input. Based on the AI ​​model, advertisements and content containing unexpected information are generated and delivered to the device, taking into account the user's interests. The information displayed on the user's device is obtained as output.

[1361] Step 5:

[1362] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The inputs are provided by the user's webcam, microphone, and text input. For example, OpenCV or IBM Watson are used to collect emotion data and analyze it in real time. The output is the user's current emotional state, which is sent to the server.

[1363] Step 6:

[1364] The server reflects the emotional data obtained from the emotion engine and dynamically influences the information provided by the information providing means. For example, if the user is excited, it provides relaxing content, and if the user is uncomfortable, it provides information to excite the user. The emotional data obtained in step 5 is used as input. Information that matches the emotion is provided as output.

[1365] Step 7:

[1366] The server again collects user reactions and behavioral data through feedback channels and uses it to improve the model. The input is behavioral data, such as user click history and search history, and emotional data collected from the device. The data is accumulated and analyzed using collection tools such as Firebase Analytics. The output is an improved model that can be reflected in the next information provision.

[1367] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1368] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1369] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1370] [Fourth embodiment]

[1371] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1372] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1373] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1374] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1375] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1376] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1377] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1378] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1379] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1380] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1381] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1382] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1383] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1384] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user. Hereinafter, a specific embodiment of this invention will be described.

[1385] System Overview

[1386] The system includes the following elements:

[1387] Data collection methods

[1388] Model Generation Method

[1389] How to Introduce Bugs

[1390] Information provision means

[1391] Feedback Methods

[1392] Data collection methods

[1393] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[1394] Model Generation Method

[1395] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[1396] How to Introduce Bugs

[1397] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[1398] Information provision means

[1399] The server then sends the information generated by the bugged AI model to the user's device, which receives it and displays it to the user. This information includes both relevant information and intentionally inserted unexpected information.

[1400] Feedback Methods

[1401] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[1402] Specific examples

[1403] Example 1: User A (Interest: Fashion)

[1404] 1. Data Collection

[1405] User A frequently searches for fashion-related sites and clicks data is collected.

[1406] This information is sent from the terminal to the server.

[1407] 2. Model Generation

[1408] The server trains an AI model optimized for User A based on fashion-related data.

[1409] 3. Introducing bugs

[1410] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[1411] 4. Information provision

[1412] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[1413] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[1414] 5. Feedback

[1415] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[1416] The server uses this data to improve the model.

[1417] Example 2: User B (Interest: Technology)

[1418] 1. Data Collection

[1419] User B browses technology-related news and the browsing history is sent to the server.

[1420] 2. Model Generation

[1421] The server trains an optimized technology-related AI model based on User B's interests.

[1422] 3. Introducing bugs

[1423] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[1424] 4. Information provision

[1425] The server sends technology-related ads, as well as art-related ads, to the device.

[1426] 5. Feedback

[1427] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[1428] In this way, by providing both information that is in line with the user's interests and unexpected information, this system promotes diversity in user behavior and provides opportunities for users to discover new interests and behaviors.

[1429] The processing flow will be explained below.

[1430] Step 1:

[1431] Users conduct searches, click on ads, and post activities on social media.

[1432] Step 2:

[1433] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[1434] Step 3:

[1435] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[1436] Step 4:

[1437] The server collects user behavior data from the database and preprocesses it (noise removal, data standardization, etc.).

[1438] Step 5:

[1439] The server generates and trains an AI model optimized for each user based on the preprocessed data.

[1440] Step 6:

[1441] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[1442] Step 7:

[1443] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[1444] Step 8:

[1445] The terminal displays to the user the information sent from the server, including both information based on the user's interests and unexpected information based on bug data.

[1446] Step 9:

[1447] The user browses the displayed information, clicks on information that interests them, or searches further.

[1448] Step 10:

[1449] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[1450] Step 11:

[1451] The server collects new behavioral data sent from the terminal and stores it in a database.

[1452] Step 12:

[1453] The server analyzes the new behavioral data and analyzes which bugs affected user behavior.

[1454] Step 13:

[1455] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates in future iterations.

[1456] In this way, the system collects and analyzes data at each step and provides users with a variety of information, thereby promoting diversity in user behavior.

[1457] Example 1

[1458] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1459] Conventional information provision systems focused on providing information based on users' interests and behavioral patterns, but lacked mechanisms for eliciting new interests and behaviors. This resulted in users being confined to their existing interests and having few opportunities to explore new fields of interest. Furthermore, the system did not effectively utilize a feedback loop to improve the accuracy of models based on user behavioral data.

[1460] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1461] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, a data storage and preprocessing means for storing the data in a database and preprocessing it, and a random bug generation means for generating bug data. This makes it possible to build an effective feedback loop that not only provides information based on the user's existing interests but also promotes the exploration of new interests and behaviors and contributes to improving the accuracy of the model.

[1462] The "data collection means" is a means for collecting user behavior data.

[1463] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[1464] "Bug injection" is a method of intentionally incorporating statistical bugs into a generated AI model to provide unexpected information to the user.

[1465] "Information provision means" refers to a means of providing information obtained using an AI model with a bug incorporated into it to a user terminal.

[1466] "Feedback means" refers to a means of recollecting user reactions and behavioral data and using it to improve the model.

[1467] "Data storage and preprocessing means" refers to the means for storing data in a database and preprocessing it.

[1468] The "random bug generation means" is a means for generating bug data.

[1469] "Behavioral data" refers to data such as a user's search history, ad click history, and social media activity.

[1470] An "AI model" is a model generated and trained by machine learning algorithms that analyzes and predicts user interests and behavioral patterns.

[1471] A "statistical bug" is noise or irrelevant information that is intentionally inserted to cause errors in the predictions of an AI model.

[1472] A "user terminal" is a terminal device such as a computer or mobile device used by a user.

[1473] MODE FOR CARRYING OUT THE INVENTION

[1474] This invention relates to a system that collects user behavioral data, intentionally incorporates statistical bugs into AI models trained based on this data, and provides users with unexpected information.

[1475] Data collection methods

[1476] The server collects user behavior data. Specifically, when a user uses a web or mobile application, the server sends data such as search history, ad click history, and social media activity from the device in real time. This allows the server to collect detailed information about the user's behavioral patterns.

[1477] Model Generation Method

[1478] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This process uses machine learning algorithms and performs data preprocessing such as filling in missing values, removing noise, and normalizing the data. For example, for User A, who is interested in fashion, an AI model is generated that prioritizes providing fashion-related information.

[1479] How to Introduce Bugs

[1480] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's hobbies and preferences, as well as random noise. A specific algorithm is applied to generate bug data, randomly generating unrelated data and inserting it into the AI ​​model. For example, "gardening" information is incorporated into the model of User A, who is interested in fashion.

[1481] Information provision means

[1482] The server then sends the information generated using the bugged AI model to the user's device, which receives it and displays it to the user. The user's device displays both information relevant to their interests and intentionally inserted unexpected information. For example, a gardening ad might appear alongside a fashion ad.

[1483] Feedback Methods

[1484] The device records behavioral data on the user's reactions to the displayed information (e.g., clicks, searches, etc.). This data is then sent back to the server and used to improve the AI ​​model. This feedback makes future information provision more suitable for the user. For example, if User A clicks on a gardening-related advertisement, the data is sent to the server and the model is adjusted.

[1485] Specific examples

[1486] Example 1: User A (Interest: Fashion)

[1487] 1. Data Collection

[1488] User A frequently searches for fashion-related sites and clicks data is collected.

[1489] This information is sent from the terminal to the server.

[1490] 2. Model Generation

[1491] The server trains an AI model optimized for User A based on fashion-related data.

[1492] 3. Introducing bugs

[1493] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[1494] 4. Information provision

[1495] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[1496] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[1497] 5. Feedback

[1498] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the terminal and sent to the server.

[1499] The server uses this data to improve the model.

[1500] Example 2: User B (Interest: Technology)

[1501] 1. Data Collection

[1502] User B browses technology-related news and the browsing history is sent to the server.

[1503] 2. Model Generation

[1504] The server trains an optimized technology-related AI model based on User B's interests.

[1505] 3. Introducing bugs

[1506] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[1507] 4. Information provision

[1508] The server sends technology-related ads, as well as art-related ads, to the device.

[1509] 5. Feedback

[1510] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[1511] Prompt Sentence Examples

[1512] Example prompt for user A

[1513] "User A is interested in fashion, but the next information we show them should also include gardening information."

[1514] Example prompt for User B

[1515] "User B prefers technology-related information, but please also include some information about art."

[1516] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1517] Specific explanation of processing steps

[1518] Step 1: Data collection

[1519] The server collects user behavior data from the device. Input includes search terms performed by the user, ads clicked, social media posts and comments, etc. Using data collection tools, the server receives this data in real time and stores it in a database. Specifically, if a user searches for "latest fashion," the search term data is sent to the server. The output is the stored behavioral data.

[1520] Step 2: Data storage and preprocessing

[1521] The server stores the collected data in a database and performs preprocessing. The behavioral data collected in step 1 is used as input. The data storage and preprocessing means performs missing value completion, noise removal, and data normalization. For example, error values ​​are removed from the collected data to create up-to-date normalized fashion-related data. The output is the preprocessed data.

[1522] Step 3: Training the AI ​​model

[1523] The server trains an AI model based on the preprocessed data. The input is the preprocessed data generated in step 2. Using the model generation means, a machine learning algorithm generates a model that predicts the user's interests and behavioral patterns. Specifically, the server trains the AI ​​model based on the fashion data and creates a predictive model optimized for user A. The output is the trained AI model.

[1524] Step 4: Generate random bugs

[1525] The server generates random bugs for the trained AI model. The input is the AI ​​model generated in step 3. Using the random bug generation means, irrelevant data and noise are generated and incorporated into the model. For example, for a model of user A who is interested in fashion, irrelevant data related to gardening is randomly generated. The output is an AI model containing bugs.

[1526] Step 5: Incorporating bug data

[1527] The server incorporates the generated random bug data into the AI ​​model. The input is the random bug data generated in step 4. The bug injection means is used to insert irrelevant data into the model. Specifically, the server inserts "gardening" information into the AI ​​model. The output is an AI model with a bug injected.

[1528] Step 6: Generate customization information

[1529] The server generates customized information using the AI ​​model with the bug. The input is the AI ​​model with the bug created in step 5. The information to be displayed to the user is generated by predictions made through the model. Specifically, fashion advertisements and some gardening advertisements are generated. The output is customized information.

[1530] Step 7: Submit your information

[1531] The server sends the generated customization information to the user's terminal. The input is the customization information generated in step 6. The information is sent to the user's terminal using the information providing means. The terminal displays the received information on the screen. As a specific operation, a fashion advertisement and a gardening advertisement are displayed on User A's terminal. The output is the information displayed on the user's terminal.

[1532] Step 8: Recording behavioral data

[1533] The terminal records the user's reactions to the displayed information (clicks, searches, etc.). The input is the user's actions. This data is sent back to the server using feedback means. For example, if user A clicks on a gardening advertisement, the click information is recorded and sent to the server. The output is the recorded behavioral data.

[1534] Step 9: Improve the model

[1535] The server retrains the AI ​​model based on the transmitted behavioral data and improves the model. The input is the behavioral data transmitted in step 8. The model is adjusted and retrained using the model generation means. Specifically, the server improves the AI ​​model based on User A's click data and reflects this in the next information provision. The output is the improved AI model.

[1536] (Application example 1)

[1537] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1538] Conventional advertising delivery systems have focused on displaying advertisements based on users' interests, but this has the problem of narrowing the diversity of users' interests and behaviors. It also limits opportunities to discover unexpected information or new interests. Therefore, there is a need for systems that allow users to discover new interests and promote the diversity of their behaviors.

[1539] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1540] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected advertisements, an information provision means for providing the user's smartphone with advertisement information obtained using the AI ​​model with the bug injected, and a feedback means for re-collecting the user's response to advertisements and behavioral data and using it to improve the model, thereby enabling users to discover new interests and information and promoting behavioral diversity.

[1541] The "data collection means" is a means for collecting user behavior data.

[1542] "Model generation means" refers to a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[1543] "Bug injection" refers to the intentional incorporation of statistical bugs into the generated AI model to provide users with unexpected advertisements.

[1544] The "information provision means" is a means of providing advertising information obtained using an AI model with a built-in bug to the user's smartphone.

[1545] "Feedback means" refers to a means of recollecting user responses to advertisements and behavioral data and using it to improve the model.

[1546] The "advertising delivery system" is a system that generates an AI model based on user behavior data and delivers advertisements that intentionally incorporate bugs to attract new user interest.

[1547] An "AI model" is an artificial intelligence model that is generated and trained based on user behavior data.

[1548] A "statistical bug" is unexpected information or random noise that is intentionally incorporated into an AI model.

[1549] "User behavioral data" refers to information such as a user's search history, ad click history, and social media activity.

[1550] A "smartphone" is a portable information terminal that is primarily used by users.

[1551] This invention is an advertising distribution system that collects user behavior data and intentionally incorporates statistical bugs into a trained AI model to provide users with unexpected advertisements. Specific embodiments of this system are described below.

[1552] System Overview

[1553] The system includes the following elements:

[1554] Data collection methods

[1555] Model Generation Method

[1556] How to Introduce Bugs

[1557] Information provision means

[1558] Feedback Methods

[1559] Data collection methods

[1560] The server collects user behavioral data, including search history, ad click history, social network activity, etc. This data is sent from the device to the server in real time when the user uses the smartphone.

[1561] Model Generation Method

[1562] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This AI model uses machine learning algorithms to analyze the user's interests and behavioral patterns and predict which advertisements should be displayed next.

[1563] How to Introduce Bugs

[1564] The server intentionally embeds statistical bugs into the generated AI model. This bug data includes advertising information unrelated to the user's interests and random noise. For this purpose, the server generates the bug data using a statistical algorithm.

[1565] Information provision means

[1566] The server then uses the bugged AI model to generate advertising information, which is then sent to the user's smartphone, which then receives the information and displays the ads to the user. This advertising information includes both ads tailored to the user's interests and intentionally inserted unexpected ads.

[1567] Feedback Methods

[1568] It records how users respond to the ads they see (for example, clicks, viewing time, etc.). This behavioral data is sent back to the server and used to improve the model. This feedback makes future ad serving more effective.

[1569] Specific examples

[1570] Example 1: User C (Interest: Technology)

[1571] 1. Data Collection

[1572] User C frequently browses technology-related content and their behavioral data is collected.

[1573] The data is sent from the smartphone to the server in real time.

[1574] 2. Model Generation

[1575] The server trains an AI model optimized for User C based on technology-related data.

[1576] 3. Introducing bugs

[1577] The server inserts advertising information for "sports," which differs from the technology, into the AI ​​model as a bug.

[1578] 4. Information provision

[1579] The server generates technology-related ads, along with some sports-related ads, and sends them to the smartphone.

[1580] When the smartphone displays this information, User C sees sports ads along with technology ads.

[1581] 5. Feedback

[1582] If User C is interested in a sports-related advertisement and clicks on it, his / her behavior is recorded and sent to the server.

[1583] The server uses this data to improve the model.

[1584] Prompt Sentence Examples

[1585] "Generate an AI model based on the categories of web pages frequently visited by users, and add ads for irrelevant categories with a set probability."

[1586] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1587] Step 1:

[1588] The server collects user behavior data. Specifically, it collects data in real time when users use their smartphones to browse websites or click on ads (search history, ad click history, social network activity, etc.). The input is user behavior data, and the output is the user behavior data stored on the server.

[1589] Step 2:

[1590] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. Specifically, it analyzes this behavioral data using a machine learning algorithm (e.g., RandomForestClassifier) ​​to build a model that predicts the next advertisement the user is likely to be interested in. The input is the collected behavioral data, and the output is the trained AI model.

[1591] Step 3:

[1592] The server intentionally incorporates statistical bugs into the generated AI model. Specifically, it applies an algorithm that adds random noise or advertising information unrelated to the user's interests (e.g., sports-related information) to the AI ​​model. The input is the trained AI model and the added bug data, and the output is an AI model with the bug incorporated.

[1593] Step 4:

[1594] The server uses the AI ​​model with the bug to generate advertising information and send it to the user's smartphone. Specifically, the advertising information generated by the AI ​​model is sent as a data packet to the smartphone, which receives and displays it. The input is the AI ​​model with the bug, and the output is the advertising information sent to the user's smartphone.

[1595] Step 5:

[1596] The device records how users respond to displayed ads (clicks, viewing time, etc.). Specifically, the smartphone application captures user responses as events and sends the data to a server. The input is the user's response to the ad, and the output is the feedback data sent to the server.

[1597] Step 6:

[1598] The server retrains and improves the AI ​​model based on the collected feedback data. Specifically, it adds the user's new behavioral data to the training dataset and updates the AI ​​model. The input is the new behavioral data including the feedback data, and the output is the improved AI model.

[1599] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1600] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into an AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with an emotion engine that recognizes the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[1601] System Overview

[1602] The system includes the following elements:

[1603] Data collection methods

[1604] Model Generation Method

[1605] How to Introduce Bugs

[1606] Information provision means

[1607] Feedback Methods

[1608] Emotion Engine

[1609] Data collection methods

[1610] The server collects user behavioral data, including search history, ad click history, and social media activity, which is transmitted from the device to the server in real time as the user uses the web or mobile application.

[1611] Model Generation Method

[1612] The server generates and trains an AI model optimized for the user's interests based on the collected behavioral data. This model analyzes the user's interests and behavioral patterns using, for example, machine learning algorithms to predict what information should be displayed next.

[1613] How to Introduce Bugs

[1614] The server intentionally incorporates bug data into the generated AI model. This bug data includes information unrelated to the user's preferences and random noise. For this purpose, the server generates bug data using a specific algorithm.

[1615] Information provision means

[1616] The server then provides the user with information generated using the AI ​​model with the bugs. The device receives the information and displays it to the user. This information includes both information relevant to the user's interests and unexpected information based on the bug data.

[1617] Feedback Methods

[1618] The device records how the user responds to the displayed information (e.g., clicks, searches, etc.). This behavioral data is then sent back to the server and used to improve the model. This feedback makes future information provision even more effective.

[1619] Emotion Engine

[1620] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The emotion engine acquires emotion data from the user's webcam, microphone, and text input. The server analyzes this data to determine the user's current emotional state.

[1621] The emotional data obtained by the emotion engine influences how information is presented to users. For example, if a user is excited, ads with content that matches that emotion will be dynamically served. Emotional data is also used as a feedback tool to help train AI models.

[1622] Specific examples

[1623] Example 1: User A (Interest: Fashion)

[1624] 1. Data Collection

[1625] User A frequently searches for fashion-related sites and clicks data is collected.

[1626] This information is sent from the terminal to the server.

[1627] 2. Model Generation

[1628] The server trains an AI model optimized for User A based on fashion-related data.

[1629] 3. Introducing bugs

[1630] The server inserts information about "gardening," which is different from fashion, into the AI ​​model as a bug.

[1631] 4. Information provision

[1632] The server generates some gardening-related advertisements along with fashion-related advertisements and transmits them to the terminal.

[1633] When the device displays this information, User A sees gardening advertisements along with fashion advertisements.

[1634] 5. Feedback

[1635] If User A becomes interested and clicks on a gardening-related advertisement, the action is recorded by the device and sent to the server.

[1636] The server uses this data to improve the model.

[1637] 6. Emotion recognition

[1638] While User A is browsing the information, the emotion engine analyzes User A's emotions from their facial expressions and voice and sends the results to the server.

[1639] The server provides appropriate information to User A based on the emotion data.

[1640] Example 2: User B (Interest: Technology)

[1641] 1. Data Collection

[1642] User B browses technology-related news and the browsing history is sent to the server.

[1643] 2. Model Generation

[1644] The server trains an optimized technology-related AI model based on User B's interests.

[1645] 3. Introducing bugs

[1646] Information about "art," which is different from technology, is inserted into the AI ​​model as a bug.

[1647] 4. Information provision

[1648] The server sends technology-related ads, as well as art-related ads, to the device.

[1649] 5. Feedback

[1650] If User B shows interest in an art-related advertisement and clicks on it, the data is sent to the server and reflected in the next information provided.

[1651] 6. Emotion recognition

[1652] While User B is browsing the information, the emotion engine analyzes User B's emotions from his / her tone of voice and text input and sends the results to the server.

[1653] The server dynamically changes the advertisement content it provides based on the analyzed emotion data.

[1654] In this way, this system not only provides users with information that is in line with their interests and unexpected information, but also uses emotional data to provide more personalized information, thereby simultaneously improving the diversity of user behavior and satisfaction.

[1655] The processing flow will be explained below.

[1656] Step 1:

[1657] Users conduct searches, click on ads, and post activities on social media.

[1658] Step 2:

[1659] The device collects behavioral data such as the user's search history, ad click history, and social media activity in real time and sends it to a server.

[1660] Step 3:

[1661] The server stores the behavioral data sent from the terminal in a database and updates the user profile.

[1662] Step 4:

[1663] The server collects user behavior data from the database and preprocesses the data (noise removal, data standardization, etc.).

[1664] Step 5:

[1665] The server uses the preprocessed data to generate and train an AI model optimized for each user.

[1666] Step 6:

[1667] The server intentionally adds random noise and irrelevant information as bugs to a trained AI model using a specific algorithm.

[1668] Step 7:

[1669] Users express their emotions through a webcam and microphone.

[1670] Step 8:

[1671] The device collects the user's voice, facial expressions, and text in real time and sends them to the emotion engine.

[1672] Step 9:

[1673] The emotion engine analyzes the received data, identifies the user's emotional state, and sends the results to the server.

[1674] Step 10:

[1675] The server analyzes the emotion data sent from the emotion engine and dynamically adjusts the information and advertisements provided based on the user's emotional state.

[1676] Step 11:

[1677] The server uses a bugged AI model to generate information and advertisements that are then provided to the device.

[1678] Step 12:

[1679] The terminal displays the information sent from the server to the user, including both information of interest to the user and unexpected information from bug data.

[1680] Step 13:

[1681] The user browses the displayed information, clicks on the information that interests them, or searches further.

[1682] Step 14:

[1683] The device again records the user's browsing, clicks, and other behavioral data and sends this data to the server.

[1684] Step 15:

[1685] The server collects new behavioral data sent from the terminal and stores it in a database.

[1686] Step 16:

[1687] The server analyzes the new behavioral data and determines which bugs affected user behavior.

[1688] Step 17:

[1689] The server retrains the model based on the feedback data obtained, improving the quality of the bug data it incorporates from the next time onwards.

[1690] In this way, the system collects and analyzes data at each step and provides users with a variety of information, promoting diverse user behavior. Furthermore, it uses an emotion engine to understand the user's emotional state and provide more personalized information.

[1691] Example 2

[1692] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1693] Modern information delivery systems are required to accurately predict user interests. However, typical systems lack the flexibility to respond to unexpected user behaviors and emotions, limiting the improvement of user experience. Furthermore, current systems have difficulty acquiring user emotional data in real time and providing information based on that data. Therefore, there is a need for a system that can simultaneously analyze a user's diverse behaviors and the emotions behind them, and dynamically provide appropriate information.

[1694] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1695] In this invention, the server includes means for collecting user behavioral data, means for generating and training a generative AI model optimized for the user's interests based on the collected behavioral data, means for intentionally incorporating statistical bugs into the generated AI model to provide the user with unexpected information, means for providing information obtained using the AI ​​model with the bugs incorporated to the user terminal, means for re-collecting user reaction and behavioral data and using it to improve the model, and emotion recognition means for analyzing emotions from the user's voice, facial expressions, and text to obtain emotion data. This makes it possible to efficiently analyze the user's diverse behavioral data and emotion data and provide the user with unexpected and useful information.

[1696] "User behavioral data" refers to data generated when a user uses the web or mobile applications, such as search history, ad click history, and social network service activity.

[1697] A "generative AI model" is a machine learning model built to analyze a user's interests and behavioral patterns based on collected user behavioral data and predict what information should be displayed next.

[1698] A "statistical bug" is data that contains information unrelated to the user's preferences or random noise that is intentionally incorporated into the generated AI model.

[1699] The "emotion recognition means" is an engine or algorithm for analyzing emotions from the user's voice, facial expressions, and text, and obtaining emotion data based on this.

[1700] "Information provision means" refers to a system or process that provides information generated using an AI model with a bug to a user's device and displays it to the user.

[1701] "Feedback means" refers to a means of recollecting the user's reactions to the displayed information (clicks, searches, etc.) and using this information to improve the AI ​​model.

[1702] This invention relates to a system that collects user behavior data, intentionally incorporates statistical bugs into a generative AI model trained based on the collected data, and provides unexpected information to the user, and further combines this with emotion recognition means for recognizing the user's emotions. Hereinafter, specific embodiments of the invention will be described.

[1703] System Overview

[1704] The system includes the following elements:

[1705] Behavioral data collection methods

[1706] Model Generation Method

[1707] How to Introduce Bugs

[1708] Information provision means

[1709] Feedback Methods

[1710] emotion recognition means

[1711] Behavioral data collection methods

[1712] The server collects behavioral data generated when users use web and mobile applications, such as search history, ad click history, social network service activity, etc. This behavioral data is collected in real time and transmitted to the server.

[1713] Model Generation Method

[1714] The server generates and trains a generative AI model optimized for the user's interests based on the collected behavioral data. This generative AI model uses machine learning algorithms (e.g., TensorFlow, PyTorch) to analyze the user's behavioral data and predict what information should be displayed next.

[1715] How to Introduce Bugs

[1716] The server intentionally incorporates statistical bugs into the trained generative AI model. This bug data includes information unrelated to the user's preferences and random noise, and uses a specific algorithm (e.g., a noise generation algorithm) to generate it.

[1717] Information provision means

[1718] The server provides information generated using the generative AI model with the bug to the user's device. The device receives this information and displays it to the user. The displayed information includes both information relevant to the user's interests and unexpected information based on the bug data.

[1719] Feedback Methods

[1720] The device records the user's reactions to the displayed information (clicks, searches, etc.) and sends the behavioral data back to the server. This feedback data is used to improve the AI ​​model, making future information provision even more effective.

[1721] emotion recognition means

[1722] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. Emotional data is acquired through the user's webcam, microphone, and text input, and the server analyzes this data to identify the user's current emotional state. The analyzed emotional data is also used for information provision and feedback means, making it possible to dynamically adjust information.

[1723] Examples of concrete examples and prompts

[1724] Example 1: User A (Interest: Fashion)

[1725] 1. Data Collection

[1726] User A frequently searches for fashion-related sites and clicks data is collected.

[1727] This information is sent from the terminal to the server.

[1728] 2. Model Generation

[1729] The server trains a generative AI model optimized for User A based on fashion-related data.

[1730] 3. Introducing bugs

[1731] The server inserts "gardening" information, which is different from fashion, into the generative AI model as a bug.

[1732] 4. Information provision

[1733] The server generates some gardening-related advertisements along with fashion-related advertisements and sends them to the terminal. When the terminal displays this information, User A sees the gardening advertisements along with the fashion advertisements.

[1734] 5. Feedback

[1735] If User A is interested and clicks on a gardening-related ad, that behavior is recorded by the device and sent to the server, which uses this data to improve the model.

[1736] 6. Emotion recognition

[1737] While User A is browsing information, the emotion recognition means analyzes User A's emotions from their facial expressions and voice and sends the results to the server. The server then provides appropriate information to User A based on the emotion data.

[1738] Example prompt sentence:

[1739] "User A is interested in fashion, but how will he react if we mix in random gardening information? Analyze the sentiment data as well."

[1740] Example 2: User B (Interest: Technology)

[1741] 1. Data Collection

[1742] User B browses technology-related news and the browsing history is sent to the server.

[1743] 2. Model Generation

[1744] The server trains an optimized generative AI model related to technology based on User B's interests.

[1745] 3. Introducing bugs

[1746] The server inserts "art" information, which is different from technology, into the generative AI model as bug data.

[1747] 4. Information provision

[1748] The server sends technology-related ads, as well as art-related ads, to the device.

[1749] 5. Feedback

[1750] If User B shows interest in an art-related ad and clicks on it, that data is recorded on the device and sent to the server, which uses this data to retrain the generative AI model.

[1751] 6. Emotion recognition

[1752] The emotion recognition means analyzes emotions from the voice and text input of User B and sends the data to the server. The server changes the advertisement content to be provided based on the analyzed emotion data.

[1753] Example prompt sentence:

[1754] "User B is interested in technology, but if you add some random art information, how will they behave? We'll take their sentiment data into account and do some analysis."

[1755] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1756] System program processing flow and processing steps

[1757] Step 1: Collect behavioral data

[1758] The server collects user behavioral data. As input, it obtains search history, ad click history, social network service activity, and other data in real time when users use web and mobile applications. This data is automatically sent from the device to the server when the user visits a specific website or clicks on a product. As output, the collected behavioral data is stored in a database.

[1759] Specific operation: A user searches a fashion-related website, enters the keyword "fall fashion," and clicks on several items. The data is sent from the device to the server in real time.

[1760] Step 2: Creating a generative AI model

[1761] The server creates and trains a generative AI model based on the collected behavioral data. It uses the behavioral data stored in the database as input. It uses a machine learning framework (e.g., TensorFlow, PyTorch) to analyze the user's interests and behavioral patterns. This analysis generates an AI model optimized for the user's interests. The output is a trained generative AI model.

[1762] How it works: The server analyzes fashion-related behavioral data and determines that "this user is interested in fall fashion." It then trains a generative AI model based on this interest.

[1763] Step 3: Incorporating the bug

[1764] The server intentionally incorporates bug data into a trained generative AI model. As input, it uses the trained generative AI model and a specific algorithm (e.g., a noise generation algorithm) to generate bug data. The bug data includes information unrelated to the user's preferences and random noise. As output, it obtains a generative AI model incorporating the bug data.

[1765] Specific operation: The server adds "gardening" information as a bug to the generative AI model, which generates "gardening"-related information along with fashion-related information.

[1766] Step 4: Provide information

[1767] The server uses a generative AI model with bugs built in to provide generated information to the terminal. As input, the server uses a generative AI model with bugs built in. The server uses this model to generate information based on the user's interests and unexpected information based on bug data. As output, the generated information is sent to the terminal and displayed to the user.

[1768] Specific operation: The server sends information including fashion-related advertisements and "gardening"-related advertisements to the terminal, which receives it and displays it to the user.

[1769] Step 5: Gather feedback

[1770] The user reacts to the displayed information (clicks, searches, etc.). The device records the user's reaction and sends it back to the server. The user's reaction data is used as input. As output, this reaction data is sent to the server and used to improve the AI ​​model.

[1771] How it works: When a user clicks on a gardening-related ad, the click data is recorded on the device and sent to the server, which uses this data to retrain the AI ​​model and improve the information provided next time.

[1772] Step 6: Emotion Recognition

[1773] The emotion recognition means analyzes emotions from the user's voice, facial expressions, and text. As input, emotion data obtained through a webcam, microphone, and text input is used. The emotion recognition engine analyzes this data to identify the user's current emotional state. As output, the analyzed emotion data is sent to a server and used for information provision and feedback means.

[1774] Specific operation: While the user is browsing information using a webcam, the emotion recognition means collects facial expression data and detects an "excited" state. The server dynamically adjusts the information provided based on this data.

[1775] (Application example 2)

[1776] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1777] While conventional systems can provide information tailored to a user's interests, they lack the ability to intentionally provide different information to stimulate new interests. Furthermore, the information provided did not reflect the user's emotions, which could have further improved the quality of the user experience. In addition, the feedback loop for improving the AI ​​model based on collected data was not fully functional.

[1778] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1779] In this invention, the server includes a data collection means for collecting user behavioral data, a model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data, a bug injection means for intentionally injecting statistical bugs into the generated AI model to provide the user with unexpected information, an information provision means for providing information obtained using the AI ​​model with the bug injected to the user terminal, a feedback means for re-collecting user reaction and behavioral data and using it to improve the model, and an emotion engine for analyzing emotions from the user's voice, facial expressions, and text and providing information based on the emotion data. This makes it possible to provide advertisements that correspond to the user's emotions and discover new areas of interest.

[1780] The "data collection means" is a means for collecting user behavior data.

[1781] "Model generation means" means a means for generating and training an AI model optimized for the user's interests based on collected behavioral data.

[1782] "Bug injection methods" are methods for intentionally incorporating statistical bugs into generated AI models to provide users with unexpected information.

[1783] "Information provision means" refers to a means for providing information obtained using an AI model incorporating a bug to a user terminal.

[1784] "Feedback means" refers to a means for recollecting user reactions and behavioral data and using it to improve the model.

[1785] The "emotion engine" is an engine that analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotion data.

[1786] This invention is a system that collects user behavioral data, intentionally incorporates statistical bugs into an AI model trained based on this data, and provides users with unexpected information, and further combines this with an emotion engine that recognizes the user's emotions.

[1787] The server implements the system using the following hardware and software.

[1788] Data collection method: Collect user behavior data (search history, ad click history, social media activity) using tools such as Google Analytics API and Mixpanel.

[1789] Model generation method: Generate and train an AI model optimized for user interests based on collected behavioral data, using machine learning algorithms such as TensorFlow and Scikit-learn.

[1790] Bug injection: Intentionally injecting statistical bugs into the generated AI model to provide unexpected information. This involves using a specific algorithm to randomly generate noise or irrelevant information.

[1791] Information provision method: Information obtained using an AI model with a bug is provided to the user's device.

[1792] Feedback methods: Recollect user reactions and behavioral data and use it to improve the AI ​​model. Specifically, use collection tools such as Firebase Analytics.

[1793] Emotion engine: Analyzes emotions from the user's voice, facial expressions, and text, and provides information based on the emotional data. Specifically, it uses emotion recognition tools such as OpenCV and IBM Watson.

[1794] Example of User A (interested in fashion)

[1795] 1. Data collection: Collect data on User A's searches and clicks on fashion-related sites. The information is collected through the Google Analytics API.

[1796] 2. Model generation: Based on the collected data, the server trains an optimized fashion-related AI model using TensorFlow.

[1797] 3. Bug insertion: The server inserts information about "gardening," which is unrelated to fashion, into the AI ​​model as a bug.

[1798] 4. Information provision: The server generates fashion and some gardening-related advertisements and sends them to the user's terminal.

[1799] 5. Feedback: User A's response to the ad (click or ignore) is recorded and sent to the server via Firebase Analytics.

[1800] 6. Emotion recognition: While User A is browsing information, an emotion engine using OpenCV analyzes User A's facial expressions and voice and provides appropriate information based on that data.

[1801] Prompt Sentence Examples

[1802] "The user is interested in 'smartwatches,' so use an AI model to show them ads related to smartwatches. Also, intentionally show ads related to 'gardening,' and prioritize uplifting information because the user is emotionally excited."

[1803] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1804] Step 1:

[1805] The server uses data collection tools to collect user behavior data. Inputs include the user's search history, ad click history, and social media activity on web and mobile applications. This data is collected in real time through tools such as Google Analytics API and Mixpanel and stored in a database as behavioral data. The output is a compiled user behavior data.

[1806] Step 2:

[1807] The server generates and trains an AI model optimized for the user's interests based on the behavioral data collected using the model generation means. The behavioral data collected in step 1 is used as input. A machine learning algorithm (e.g., TensorFlow or Scikit-learn) is used to analyze the user's interests and behavioral patterns and generate a predictive model. The output is an optimized AI model.

[1808] Step 3:

[1809] The server intentionally embeds statistical bugs into the generated AI model using a bug injection method. The AI ​​model generated in step 2 is used as input. Random noise and irrelevant information (bug data) are generated using a specific algorithm and inserted into the model. The output is an AI model with embedded statistical bugs.

[1810] Step 4:

[1811] The server uses an information provision means to provide information generated based on the AI ​​model with the bug to the user's device. The AI ​​model obtained in step 3 is used as input. Based on the AI ​​model, advertisements and content containing unexpected information are generated and delivered to the device, taking into account the user's interests. The information displayed on the user's device is obtained as output.

[1812] Step 5:

[1813] The emotion engine analyzes emotions from the user's voice, facial expressions, and text. The inputs are provided by the user's webcam, microphone, and text input. For example, OpenCV or IBM Watson are used to collect emotion data and analyze it in real time. The output is the user's current emotional state, which is sent to the server.

[1814] Step 6:

[1815] The server reflects the emotional data obtained from the emotion engine and dynamically influences the information provided by the information providing means. For example, if the user is excited, it provides relaxing content, and if the user is uncomfortable, it provides information to excite the user. The emotional data obtained in step 5 is used as input. Information that matches the emotion is provided as output.

[1816] Step 7:

[1817] The server again collects user reactions and behavioral data through feedback channels and uses it to improve the model. The input is behavioral data, such as user click history and search history, and emotional data collected from the device. The data is accumulated and analyzed using collection tools such as Firebase Analytics. The output is an improved model that can be reflected in the next information provision.

[1818] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1819] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1820] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1821] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1822] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1823] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1824] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1825] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1826] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1827] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1828] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1829] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1830] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1831] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1832] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1833] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1834] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1835] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1836] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1837] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1838] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1839] The following is further disclosed regarding the above embodiment.

[1840] (Claim 1)

[1841] a data collection means for collecting user behavior data;

[1842] A model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data;

[1843] A bug injection method that intentionally incorporates statistical bugs into the generated AI model to provide unexpected information to the user.

[1844] an information providing means for providing information obtained using the AI ​​model incorporating the bug to a user terminal;

[1845] A feedback method to collect user reactions and behavioral data again and use it to improve the model.

[1846] A system including:

[1847] (Claim 2)

[1848] 10. The system of claim 1, wherein the data collection means collects behavioral data including user search history, ad click history, and social media activity.

[1849] (Claim 3)

[1850] 2. The system of claim 1, wherein the bug injection means randomly generates and inserts noise or irrelevant information into the AI ​​model.

[1851] "Example 1"

[1852] (Claim 1)

[1853] a data collection means for collecting user behavior data;

[1854] A model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data;

[1855] A bug injection method that intentionally incorporates statistical bugs into the generated AI model to provide unexpected information to the user.

[1856] an information providing means for providing information obtained using the AI ​​model incorporating the bug to a user terminal;

[1857] A feedback method to collect user reactions and behavioral data again and use it to improve the model.

[1858] a data storage and preprocessing means for storing data in a database and preprocessing the data;

[1859] a random bug generation means for generating bug data;

[1860] A system including:

[1861] (Claim 2)

[1862] 10. The system of claim 1, wherein the data collection means collects behavioral data including user search history, ad click history, and social media activity.

[1863] (Claim 3)

[1864] 2. The system of claim 1, wherein the bug injection means utilizes a random bug generation means that randomly generates noise or irrelevant information and inserts it into the AI ​​model.

[1865] "Application Example 1"

[1866] Reflecting new invention content in claims

[1867] Original Claims:

[1868] (Claim 1)

[1869] a data collection means for collecting user behavior data;

[1870] A model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data;

[1871] A bug injection method that intentionally incorporates statistical bugs into the generated AI model to provide unexpected information to the user.

[1872] an information providing means for providing information obtained using the AI ​​model incorporating the bug to a user terminal;

[1873] A feedback method to collect user reactions and behavioral data again and use it to improve the model.

[1874] A system including:

[1875] (Claim 2)

[1876] 10. The system of claim 1, wherein the data collection means collects behavioral data including user search history, ad click history, and social media activity.

[1877] (Claim 3)

[1878] 2. The system of claim 1, wherein the bug injection means randomly generates and inserts noise or irrelevant information into the AI ​​model.

[1879] New invention claims:

[1880] (Claim 1)

[1881] a data collection means for collecting user behavior data;

[1882] A model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data;

[1883] A bug-injection method that intentionally incorporates statistical bugs into the generated AI model to provide users with unexpected advertisements;

[1884] An information providing means for providing advertising information obtained using the AI ​​model incorporating the bug to a user's smartphone;

[1885] A feedback mechanism to collect user responses to advertisements and behavioral data again and use it to improve the model.

[1886] An advertising distribution system including:

[1887] (Claim 2)

[1888] 2. The advertising distribution system according to claim 1, wherein the data collection means collects behavioral data including a user's search history, ad click history, and social media activity.

[1889] (Claim 3)

[1890] 2. The advertising distribution system of claim 1, wherein the bug injection means randomly generates noise or irrelevant information and inserts it into the AI ​​model.

[1891] "Example 2: Combining Emotion Engines"

[1892] (Claim 1)

[1893] a means for collecting user behavior data;

[1894] A means for generating and training a generative AI model optimized for user interests based on collected behavioral data; and

[1895] A method to intentionally incorporate statistical bugs into the generated AI model to provide unexpected information to the user, and

[1896] A means for providing information obtained using the AI ​​model incorporating the bug to a user terminal;

[1897] A means to re-collect user reaction and behavior data and use it to improve the model,

[1898] emotion recognition means for analyzing emotions from a user's voice, facial expression, and text to acquire emotion data;

[1899] A system including:

[1900] (Claim 2)

[1901] 10. The system of claim 1, wherein the data collection means collects behavioral data including a user's search history, ad click history, and social network service activity.

[1902] (Claim 3)

[1903] 2. The system of claim 1, wherein the bug injection means randomly generates and inserts noise or irrelevant information into the AI ​​model.

[1904] "Application example 2 when combining emotion engines"

[1905] (Claim 1)

[1906] a data collection means for collecting user behavior data;

[1907] A model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data;

[1908] A bug injection method that intentionally incorporates statistical bugs into the generated AI model to provide unexpected information to the user.

[1909] an information providing means for providing information obtained using the AI ​​model incorporating the bug to a user terminal;

[1910] A feedback method to collect user reactions and behavioral data again and use it to improve the model.

[1911] an emotion engine that analyzes emotions from the user's voice, facial expressions, and text and provides information based on the emotion data;

[1912] A system including:

[1913] (Claim 2)

[1914] 10. The system of claim 1, wherein the data collection means collects behavioral data including user search history, ad click history, and social media activity.

[1915] (Claim 3)

[1916] 2. The system of claim 1, wherein the bug injection means randomly generates and inserts noise or irrelevant information into the AI ​​model. [Explanation of symbols]

[1917] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a data collection means for collecting user behavior data; A model generation means for generating and training an AI model optimized for the user's interests based on the collected behavioral data; A bug injection method that intentionally incorporates statistical bugs into the generated AI model to provide unexpected information to the user. an information providing means for providing information obtained using the AI ​​model incorporating the bug to a user terminal; A feedback method to collect user reactions and behavioral data again and use it to improve the model. A system including:

2. The system of claim 1 , wherein the data collection means collects behavioral data including a user's search history, ad click history, and social networking activity.

3. The system of claim 1 , wherein the bug injection means randomly generates noise or irrelevant information and inserts it into the AI ​​model.

Images