System

Abstract

A system is provided.SOLUTION: A system comprising: means for collecting behavioral information; means for analyzing the collected behavioral information to determine user interests; means for using a generative AI to generate advertisement text and images based on the interest information; means for delivering the generated advertisements to user display terminals; and means for collecting feedback information to measure the effectiveness of the delivered advertisements and optimize the generative AI 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] While conventional targeted advertising systems deliver ads based on user behavioral data, the ad creative itself is uniform and not tailored to individual users' interests and preferences. This results in inappropriate ads being delivered to users, causing discomfort and disinterest, leading to increased use of ad blockers. In such situations, advertising effectiveness is low, resulting in reduced advertising revenue and a poor user experience. The present invention aims to solve these issues by generating personalized ads tailored to individual users' interests and improving advertising effectiveness. [Means for solving the problem]

[0005] The present invention solves the above-mentioned problems with a system including specific means. First, a means is provided for collecting behavioral data such as user browsing history, click history, and search keywords. Next, a means is provided for analyzing the collected behavioral data to identify user interests. Furthermore, a means is provided for using generative AI to generate advertising text and images based on the interest information. A means is also provided for delivering the generated advertisements to the user's display device. Finally, a means is provided for measuring the effectiveness of the delivered advertisements and collecting feedback data for optimizing the generative AI model. This provides a system that generates and delivers personalized advertisements optimized for each user's interests, maximizing the effectiveness of the advertisements and improving the user experience.

[0006] "Behavioral Data" refers to information about a user's web behavior, such as their browsing history, click history, and search keywords.

[0007] "Generative AI" refers to artificial intelligence technology that generates advertising text and images based on user interest information.

[0008] "User interests" refers to categories and topics that are identified as being of interest to users based on collected behavioral data.

[0009] "Ad text" refers to the written portion of the ad creative created by generative AI.

[0010] "Ad Image" refers to the visual portion of the ad creative created by generative AI.

[0011] "Feedback data" refers to information about user responses to delivered advertisements (such as click rates and skip rates).

[0012] "Advertising effectiveness" refers to an indicator that shows how much interest the delivered advertisement attracted in users and how many actions (clicks or conversions) it caused.

[0013] "User Profile" means a data set that contains the interests and characteristics identified for an individual user based on collected behavioral data and analytics.

[0014] "Advertising Content Database" refers to a database system for storing generated advertising text and images and delivering them as needed.

[0015] "Optimization" refers to the process of using feedback data to improve the performance of generative AI models and generate more effective ad creatives.

[0016] "Batch processing" refers to a processing method for analyzing collected data all at once.

[0017] "Click-through rate" refers to an indicator that indicates the percentage of clicks on a particular ad compared to the number of times it is displayed.

[0018] "HTML template" refers to the framework of a web page into which advertising creatives can be embedded. [Brief explanation of the drawings]

[0019] [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

[0020] 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.

[0021] First, the terms used in the following description will be explained.

[0022] 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).

[0023] 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.

[0024] 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.

[0025] 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.

[0026] 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."

[0027] [First embodiment]

[0028] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0029] 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.

[0030] 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).

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0036] 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.

[0037] 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.

[0038] 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.

[0039] 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."

[0040] System Overview

[0041] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. This system operates between a server and user devices, collecting and analyzing user behavior data to generate and deliver appropriate advertisements.

[0042] Program processing flow and operation

[0043] Data collection and analysis

[0044] server:

[0045] 1. When a user visits a website, the server obtains a user identifier using cookies, and also collects behavioral data such as browsing history, click history, and search keywords.

[0046] 2. The collected data is stored in a database.

[0047] 3. The server periodically analyzes the data in batches and uses machine learning algorithms and natural language processing techniques to identify user interests. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[0048] Ad generation

[0049] server:

[0050] 1. Based on the identified interest information, the server requests the ad text and images from the generating AI.

[0051] 2. For example, a generative AI model (e.g., GPT-3) generates advertising text like, "Special Offer on the Latest Fitness Equipment!" An image-generating AI like DALL-E generates images of fitness equipment.

[0052] 3. The generated ad text and images are stored in the ad content database.

[0053] Ad serving

[0054] server:

[0055] 1. When a user visits a web page, the server retrieves the user identifier from the cookie.

[0056] 2. The server reads the user's interest information from the user profile database and retrieves the relevant ad creative from the ad content database.

[0057] 3. The acquired ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[0058] Measurement and optimization

[0059] server:

[0060] 1. The server monitors user responses to the delivered advertisements (e.g., click-through rate and duration of visit) and records the data in real time.

[0061] 2. The recorded data is used to evaluate and optimize the generative AI model. The model's performance is evaluated, and retraining and tuning are carried out as necessary to further improve advertising effectiveness.

[0062] Specific examples

[0063] If User A reads many articles about pet supplies, their behavioral data is collected and analyzed. The server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of adorable pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[0064] In this way, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests, maximizing the effectiveness of the advertisements.

[0065] The processing flow will be explained below.

[0066] Step 1:

[0067] server:

[0068] Every time a user visits a website, the server obtains a user identifier using a cookie, and also records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[0069] Step 2:

[0070] server:

[0071] Periodically, collected data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify user interest categories (e.g., pets, travel, technology) and update the user profile database.

[0072] Step 3:

[0073] server:

[0074] Based on the interest information, a request is sent to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty pet food discount sale!". Furthermore, an image-generating AI (e.g., DALL-E) is used to generate attractive images of pet food.

[0075] Step 4:

[0076] server:

[0077] The generated advertisement text and images are stored in an advertisement content database.

[0078] Step 5:

[0079] server:

[0080] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from a user profile database, and retrieves relevant ad creative for that user from an ad content database.

[0081] Step 6:

[0082] server:

[0083] Ad creatives are dynamically embedded into HTML templates and displayed in the user's browser.

[0084] Step 7:

[0085] server:

[0086] It monitors user responses to delivered ads (click-through rate, duration of visit, etc.) in real time and records the data.

[0087] Step 8:

[0088] server:

[0089] The performance of the generative AI model is evaluated based on the collected effectiveness measurement data, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[0090] Step 9:

[0091] User device:

[0092] When an ad is displayed, the user's browser sends user actions such as clicks and skips to the server, which collects this data as feedback and uses it for performance measurement and optimization.

[0093] This series of steps allows personalized advertisements to be generated and delivered to users, maximizing advertising effectiveness.

[0094] Example 1

[0095] 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."

[0096] Conventional advertising systems struggled to generate and deliver personalized ads that accurately reflected user interests, resulting in a tendency for advertising effectiveness to decline. Furthermore, they lacked the functionality to measure the effectiveness of delivered ads in real time and optimize and retrain the AI ​​model, making it difficult to improve advertising accuracy. This could lead to dissatisfaction for both advertisers and users.

[0097] 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.

[0098] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generation AI to generate advertising text and images based on the interest information, means for recording the generated advertising text and images in a database, means for acquiring a user identifier each time the user visits a web page and delivering advertisements based on the user's interest information, and means for measuring the effectiveness of the delivered advertisements and collecting feedback data for optimizing the generation AI model. This enables the generation and delivery of highly accurate personalized advertisements that accurately reflect user interests and concerns, maximizing advertising effectiveness and improving user satisfaction.

[0099] "User browsing history" refers to the history of pages a user has viewed on a website, and is information about which pages a user has visited in the past.

[0100] "Click history" refers to the history of links and buttons that a user clicks on a web page, and is a record of the user's operational behavior.

[0101] "Search keywords" refer to keywords entered by users using search engines, and are data that indicate the user's interests and the information they are looking for.

[0102] "Behavioral data" refers to data about a series of actions a user takes on a website, such as the user's browsing history, click history, and search keywords.

[0103] "Means of collection" refers to the technologies and methods used to obtain and store user behavioral data.

[0104] "Means for analyzing and identifying user interests" refers to technologies and methods for analyzing collected behavioral data and identifying the interests of users.

[0105] "Interest information" refers to information about a user's interests and subjects of interest, and is data that falls into a specific category.

[0106] "Generative AI" refers to algorithms and models that use artificial intelligence technology to create advertising text and images.

[0107] "Dynamic embedding" refers to the real-time insertion of advertisements into web pages accessed by users.

[0108] "Means for measuring effectiveness" refers to technologies and methods for evaluating how users respond to the delivered advertisements, such as recording click-through rates and duration of visits.

[0109] "Feedback data" refers to data collected to evaluate the effectiveness of advertising and is information used to optimize and retrain generative AI models.

[0110] "Optimization means" refers to techniques and methods for improving the performance of generative AI models based on feedback data.

[0111] A "database" refers to a system for systematically managing stored data and retrieving data as needed.

[0112] The present invention provides a system for generating personalized advertisements based on user interests and improving advertising effectiveness. This system operates mainly between a server and a user terminal, and is realized using the following specific hardware and software.

[0113] Data collection and analysis

[0114] server:

[0115] When a user accesses a website, the server obtains a user identifier using cookies. It also collects behavioral data such as the user's browsing history, click history, and search keywords in real time. This data is initially stored in memory and periodically written to a database. Data analysis is performed using machine learning libraries such as Python's Scikit-learn library and TensorFlow. Based on the analysis results, the user's interest categories are identified and added to the user profile.

[0116] Ad generation

[0117] server:

[0118] Based on the identified user interest information, the server requests the generation AI to generate the text and images of the advertisement. The generation AI model uses a natural language processing algorithm (e.g., GPT-3) and an image generation algorithm (e.g., image generation AI). The server sends the following prompt to the generation AI:

[0119] Generate ad text with a "Fitness Equipment" theme. Include a short catchphrase to grab users' attention.

[0120] Example: Special deals on the latest fitness equipment!

[0121] The generated advertisement text and images are recorded in an advertisement content database.

[0122] Ad serving

[0123] server:

[0124] When a user visits a web page, the server retrieves the user's identifier using a cookie. The server reads the user's interests from the user profile database and retrieves the relevant ad creative from the ad content database. The retrieved ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[0125] Measurement and optimization

[0126] server:

[0127] The server monitors user responses to the delivered ads and records click information and ad viewing time in real time. The recorded data is used to evaluate and optimize the generative AI model. The server periodically uses this data to retrain the generative AI model and improve the accuracy of ad generation.

[0128] Specific examples

[0129] If User A reads many articles about pet supplies, that behavioral data is collected in real time. Based on this data, the server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of cute pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server records the click information and uses it as optimization data for the generation AI model. The next time the user visits, a more accurate ad is delivered, improving the effectiveness of the ad.

[0130] As described above, this system generates and delivers personalized advertisements based on user interests, maximizing advertising effectiveness and improving user satisfaction.

[0131] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0132] Step 1: Data collection

[0133] When a user accesses a website, the server obtains a user identifier using a cookie. Specifically, the server reads the cookie information included in the HTTP request and extracts the user identifier. The server then collects behavioral data such as the user's browsing history, click history, and search keywords. This data is temporarily stored in the server's memory and eventually written to a database.

[0134] Input: User access information (cookies, browsing history, click history, search keywords)

[0135] Output: Behavioral data record (database)

[0136] Step 2: Data analysis

[0137] The server periodically analyzes the collected behavioral data in batches. The server uses machine learning algorithms, such as Python's Scikit-learn library and TensorFlow, to identify user interest categories. This analysis allows the server to add interest categories based on the content the user frequently accesses.

[0138] Input: Behavioral data (database)

[0139] Output: Interest categories (user profile database)

[0140] Step 3: Request Ad Generation

[0141] The server sends a prompt to the AI ​​based on the identified interests. For example, if the user is interested in "health and fitness," the server sends the following prompt to the AI:

[0142] Generate ad text for "Fitness Equipment." Include a short catchphrase to grab users' attention.

[0143] Example: Special deals on the latest fitness equipment!

[0144] The AI ​​generator generates ad text based on this prompt, while the image generator algorithm simultaneously generates related images.

[0145] Input: Interest information (user profile database)

[0146] Output: Ad text and images (ad content database)

[0147] Step 4: Save your ad

[0148] The generated ad text and images are stored in an ad content database by the server, which does this by performing write operations to the database. Each ad is given a unique identifier and linked to user interest information.

[0149] Input: Ad text and image

[0150] Output: Record of advertising content (advertising content database)

[0151] Step 5: Serving Ads

[0152] Each time the user visits a new web page, the server retrieves the user identifier from the cookie and looks up the user's interests in a user profile database. The server then retrieves relevant advertisements from an advertising content database and dynamically inserts them into an HTML template that is sent to the browser, where the advertisements are displayed on the user's web page.

[0153] Input: User access information (cookies), interest information (user profile database), advertising content (advertising content database)

[0154] Output: A web page with dynamically generated ads

[0155] Step 6: Measure and optimize

[0156] The server monitors user responses to the delivered ads. Specifically, it records the ad click-through rate and duration in real time. This data is stored in a database and used to evaluate and optimize the generative AI model. The server uses this data to periodically retrain the generative AI model and improve the accuracy of ad generation.

[0157] Input: Ad response data (click-through rate, duration)

[0158] Output: Feedback data (database), optimized generative AI model

[0159] This is the flow of processing in the program for this system, which enables the generation and delivery of highly accurate personalized advertisements based on user interests.

[0160] (Application example 1)

[0161] 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."

[0162] Conventional ad delivery systems have difficulty generating personalized ads that accurately capture user interests, resulting in reduced advertising effectiveness. Additionally, it is difficult to measure the effectiveness of generated ads or optimize the AI ​​model in real time, making it difficult to maximize the effectiveness of ads.

[0163] 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.

[0164] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generative AI to generate advertising text and images based on the interest information, means for delivering the generated advertisement to the user's display device, means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generative AI model, and means for optimizing the generative AI model based on user response data. This makes it possible to generate and deliver advertisements individually personalized for each user and optimize the effectiveness of the advertisements in real time.

[0165] "User browsing history" refers to the history of web pages a user has accessed on the Internet.

[0166] "Click history" refers to the history of which links and ads a user clicked on a website.

[0167] "Search Keywords" refers to words or phrases entered by a user using a search engine.

[0168] "Behavioral data" is a collective term for data related to a series of operations or activities a user performs on the Internet.

[0169] "Generative AI" refers to systems or models that use artificial intelligence techniques to generate content for specific purposes.

[0170] "Advertising text" refers to the text and copy used in an advertisement.

[0171] "Image" means any illustration, photograph or graphic visually displayed in an Advertisement.

[0172] "User display device" refers to the device a user uses to view advertisements or web pages, including smartphones, tablets, and PCs.

[0173] "Delivery" refers to the process of displaying advertising content on a designated user's device.

[0174] "Measuring effectiveness" refers to analyzing users' responses and behavior to the ads delivered and evaluating their performance.

[0175] "Feedback Data" refers to user response data collected to evaluate the performance of an advertisement.

[0176] "User Response Data" means data regarding user clicks, view times, and other interactions with advertisements.

[0177] A "generative AI model" refers to a concrete implementation of an algorithm or system for generating advertising content using artificial intelligence technology.

[0178] "Optimization" refers to using collected data to adjust and refine generative AI models to improve their performance.

[0179] System Overview

[0180] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. The system operates between a server and user terminals, collecting and analyzing user behavior data to generate and deliver appropriate advertisements. Specifically, it includes the following means and processes:

[0181] Data collection and analysis

[0182] The server obtains a user identifier using cookies when a user visits a website. It also collects behavioral data, such as browsing history, click history, and search keywords. This behavioral data is stored in a database, and the server periodically analyzes the data in batches to identify user interests using machine learning algorithms and natural language processing techniques. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[0183] Ad generation

[0184] The server requests ad text and images from the generation AI based on the identified interest information. For example, the generative AI model generates ad text such as "Special Offer on the Latest Fitness Equipment!", and the image generation AI generates images of fitness equipment. The generated ad text and images are stored in the ad content database.

[0185] Ad serving

[0186] When a user visits a web page, the server retrieves the user identifier from the cookie. The server reads the user's interest information from the user profile database and retrieves relevant advertising content from the advertising content database. The retrieved advertising content is dynamically embedded into an HTML template and sent to the user's browser. This allows the user to see personalized ads in real time.

[0187] Measurement and optimization

[0188] The server monitors user responses to the delivered ads (e.g., click-through rate and dwell time) and records the data in real time. This collected data is used to evaluate and optimize the generative AI model. By evaluating the model's performance and re-training or tuning as necessary, the effectiveness of the ads can be further improved.

[0189] Specific examples of technology

[0190] For example, if a user reads many articles about "pet supplies," that behavioral data is collected and analyzed. The server identifies the user's interest category as "pets" and requests the generation AI to generate an ad. The generation AI model generates ad text such as "Specialty pet food discount sale!", and the image generation AI generates images of cute pet food. These ad creatives are embedded in real time into the web pages the user visits. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[0191] Prompt Sentence Examples

[0192] 1. "Create an ad for the latest pet products."

[0193] 2. "Pet supplies images"

[0194] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0195] Step 1:

[0196] Collects behavioral data such as user browsing history, click history, and search keywords.

[0197] Specific behavior:

[0198] Cookies are used in the user's browser to obtain a user identifier. Data on the user's browsing history when visiting websites, the links and ads clicked, and search keywords entered into search engines are collected. This behavioral data is stored in a database on the server.

[0199] Input: User behavior data on the website

[0200] Data processing: Use of cookies to identify user identifiers and collect behavioral data

[0201] Output: Behavioral data is saved to a database

[0202] Step 2:

[0203] Analyze collected behavioral data to identify user interests.

[0204] Specific behavior:

[0205] The server periodically extracts batches of behavioral data from the database and analyzes them using machine learning algorithms and natural language processing techniques. It extracts categories that users frequently access and adds interest categories, such as "health and fitness," to the profile database.

[0206] Input: Behavioral data stored in a database

[0207] Data processing: Analyzing data using machine learning algorithms and natural language processing techniques

[0208] Output: User interest categories are identified and stored in a profile database

[0209] Step 3:

[0210] Uses generative AI to generate ad text and images based on interest information.

[0211] Specific behavior:

[0212] The server sends the specified interest information as an input prompt to a generation AI (e.g., GPT-3 and DALL-E). The generation AI generates ad text such as "Special Offer on the Latest Fitness Equipment!" and a matching image. The generated ad content is stored in an ad content database.

[0213] Input: Interest information

[0214] Data processing: Send prompts to the generative AI to generate text and images

[0215] Output: The generated ad text and images are stored in the ad content database.

[0216] Step 4:

[0217] The generated advertisement is delivered to the user's display device.

[0218] Specific behavior:

[0219] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from the profile database, retrieves relevant advertising content from the advertising content database, dynamically embeds the retrieved advertisement into an HTML template, and sends it to the user's browser.

[0220] Input: User identifier, advertising content

[0221] Data processing: Dynamically embedding advertising content into HTML templates

[0222] Output: A personalized ad is displayed in the user's browser

[0223] Step 5:

[0224] Measure the effectiveness of delivered ads and collect feedback data to optimize generative AI models.

[0225] Specific behavior:

[0226] The server monitors user responses to the delivered ads, such as click-through rates and browser dwell times. The collected data is stored as feedback data and used to evaluate and optimize the generative AI model. The model's performance is analyzed and retrained or adjusted as necessary.

[0227] Input: User response data

[0228] Data processing: Collect and analyze reaction data, and retrain and adjust the model

[0229] Output: An optimized generative AI model

[0230] 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.

[0231] System Overview

[0232] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral and emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[0233] Program processing flow and operation

[0234] Data collection and analysis

[0235] server:

[0236] 1. The server obtains a user identifier using cookies each time a user visits a website, and records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[0237] 2. The server reads the user's facial expressions through the webcam on the web page the user is viewing, and analyzes the emotional data using the emotion engine. This emotional data is also stored in the database.

[0238] Ad generation

[0239] server:

[0240] 1. Periodically collected behavioral and emotional data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify the user's interest categories and emotional state.

[0241] 2. Based on the identified interests and real-time emotional state, the system sends a request to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty Pet Food Sale!", and an image-generating AI such as DALL-E is used to generate attractive images of the pet food.

[0242] 3. The emotion engine generates ads with bright colors and positive words when the user is in a positive emotional state, and conversely, generates ads with calming tones and soothing words when the user is in a negative emotional state.

[0243] Ad serving

[0244] server:

[0245] 1. When a user visits a web page, the server retrieves the user identifier from the cookie, reads the user's interests and emotional state from the user profile database, and retrieves the relevant ad creative for that user from the ad content database.

[0246] 2. The acquired ad creative is dynamically embedded into an HTML template and displayed in the user's browser.

[0247] Measurement and optimization

[0248] server:

[0249] 1. The server monitors user responses to delivered ads (click-through rate, length of stay, etc.) in real time and records the data.

[0250] 2. Based on the recorded data, evaluate the performance of the generative AI model and emotion engine, and retrain or tune it as needed to further improve the effectiveness of advertising.

[0251] Specific examples

[0252] For example, suppose User B reads many articles about pet supplies. While User B is browsing a web page, the emotion engine detects that the user is relaxed from their facial expression. Based on this data, the server generates ad text such as "Specialty Pet Food Discount Sale!" along with an image of pet food. The generated ad is created using bright colors and positive wording to match User B's relaxed state. This ad is then embedded in real time into the web pages visited by User B, attracting User B's interest. When the ad is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future ad generation.

[0253] As described above, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests and real-time emotional state, thereby maximizing advertising effectiveness.

[0254] The processing flow will be explained below.

[0255] Step 1:

[0256] server:

[0257] When a user visits a website, the server uses cookies to obtain a user identifier, and also obtains behavioral data such as the user's browsing history, click history, and search keywords in real time, which are then stored in a database.

[0258] Step 2:

[0259] User device:

[0260] While a user is browsing a website, their facial expressions are captured by a webcam, and this image data is instantly sent to a server.

[0261] Step 3:

[0262] server:

[0263] Using the transmitted image data, the emotion engine analyzes the user's facial expressions to identify their real-time emotional state (e.g., happy, sad, surprised), which is also recorded in a database.

[0264] Step 4:

[0265] server:

[0266] It periodically analyzes behavioral and emotional data through batch processing, and uses machine learning algorithms and natural language processing techniques to identify users' interest categories and emotional states, thereby revealing which categories users are interested in.

[0267] Step 5:

[0268] server:

[0269] Based on the identified interests and real-time emotional state, the system requests a generative AI (e.g., GPT-3, DALL-E) to generate ad text and images. For example, if the emotion engine determines that the user is relaxed, the generative AI will generate ad text that matches the relaxed state, such as "Specialty pet food discount sale!". DALL-E will then generate a bright, relaxing image of the pet food.

[0270] Step 6:

[0271] server:

[0272] The generated advertisement text and images are stored in an advertisement content database.

[0273] Step 7:

[0274] server:

[0275] Each time a user visits a web page, the system retrieves the user identifier from the cookie, reads their interests and emotional state from the user profile database, and retrieves ad creative from the ad content database based on this information.

[0276] Step 8:

[0277] server:

[0278] The resulting ad creative is dynamically embedded into an HTML template and displayed in the user's browser, with the tone and color of the ad adjusted based on the user's emotional state.

[0279] Step 9:

[0280] server:

[0281] It monitors user responses to delivered ads (whether the ad was clicked or skipped, etc.) in real time and records the data.

[0282] Step 10:

[0283] server:

[0284] Based on the collected effectiveness measurement data, the performance of the generative AI model and emotion engine is evaluated, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[0285] Step 11:

[0286] User device:

[0287] When an ad is displayed, the user's browser sends actions such as clicks or skips to the server, and this feedback data is used in the measurement and optimization process.

[0288] This series of steps allows for the generation and delivery of personalized ads that are optimized for the user's interests and emotional state, maximizing advertising effectiveness.

[0289] Example 2

[0290] 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."

[0291] Conventional personalized advertising systems generate ads based solely on user behavioral data and lack the ability to generate ads that take into account the user's real-time emotional state. This makes it difficult to optimize ads based on the user's emotional state and maximize advertising effectiveness. Furthermore, the AI ​​model for generating ads is not adequately optimized based on the effectiveness measurement data of the generated ads, making it difficult to expect continuous improvement in advertising effectiveness.

[0292] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify the user's interests, means for analyzing the collected facial expression data to identify the user's emotional state, means for using a generation AI to generate advertisement text and images based on the interest information and emotional state, means for delivering the generated advertisement to the user's display terminal, and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating the user's behavioral data and emotional data.

[0293] "User browsing history" refers to historical data about the pages a user visits on a website.

[0294] "Click history" is historical data about the links and buttons a user clicks on a website.

[0295] "Search Keywords" are search terms entered by a user into a search engine.

[0296] "Behavioral data" refers to data related to a user's behavior on a website, such as their browsing history, click history, and search keywords.

[0297] "Facial expression data" is data collected based on a user's facial expressions.

[0298] "Emotional state" is information indicating the emotional state of the user, obtained by analyzing the user's facial expression data.

[0299] "Interest Information" is information about a user's interests and categories of interest that are identified through analysis of collected behavioral data.

[0300] "Generative AI" is an artificial intelligence technology that generates text and images based on a given prompt.

[0301] The "user's display terminal" refers to a display device such as a computer or smartphone used by the user.

[0302] "Advertising text" is the text displayed as an advertising message.

[0303] "Advertising image" is image data that is displayed as an advertising message.

[0304] "Feedback data" is data that records user responses to delivered advertisements and is used to optimize generative AI models.

[0305] "Delivery" refers to displaying the generated advertisement on the user's display device.

[0306] "Optimizing generative AI models" refers to making adjustments and improvements to improve the performance of generative AI based on collected feedback data.

[0307] MODE FOR CARRYING OUT THE INVENTION

[0308] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates primarily between a server and a user's device and is designed to provide a rich, personalized advertising experience.

[0309] Data collection and analysis

[0310] server:

[0311] 1. The server obtains a user identifier using a cookie each time a user visits the website. The user identifier is generated as a unique ID and stored in a database.

[0312] 2. The server then records users' browsing history, click history, search keywords, and other behavioral data in real time, which is then stored in a database such as MongoDB or MySQL along with a timestamp.

[0313] 3. The server reads the user's facial expressions through the camera installed on the user's device and analyzes the emotional data using an emotion engine (for example, Microsoft Azure's Emotion API). The resulting emotion score (for example, joy, sadness, surprise, etc.) is also stored in a database.

[0314] Ad generation

[0315] server:

[0316] 1. The server processes the collected behavioral and emotional data in batches, analyzes the data using a batch processing framework such as Apache Spark, and applies a machine learning model (e.g., a clustering algorithm using scikit-learn) to classify the user's interest categories.

[0317] 2. Send a request to a generative AI (e.g., GPT-3 and DALL-E using OpenAI APIs) to generate ad text and images based on the identified interests and real-time emotional state. Create a prompt like the following: "Generate ad copy with a light tone that corresponds to a relaxed emotional state for users interested in pet products."

[0318] 3. The emotion engine is used to appropriately adjust the tone of the generated ad: if the emotional state is positive, the ad content will be generated with bright images and positive language.

[0319] Ad serving

[0320] server:

[0321] 1. Every time a user visits a web page, the cookie information is analyzed to obtain the user identifier again, and an SQL query is issued to obtain the necessary interest information and emotional state from the user profile database.

[0322] 2. The retrieved ad creative is dynamically embedded into an HTML template and displayed in the user's browser using a web server framework such as Node.js. A template engine (e.g., EJS or Handlebars) is used to embed ad data into HTML.

[0323] Measurement and optimization

[0324] server:

[0325] 1. User responses to delivered ads (click-through rate, duration, etc.) are monitored in real time and the data is recorded. This is done using the Google Analytics API, which collects user behavior data and stores it in a database.

[0326] 2. Evaluate the performance of the generative AI model and emotion engine based on the recorded data. Use Python scripts to calculate the precision and recall of the model, and retrain or tune it as needed to improve the overall advertising effectiveness of the system.

[0327] Specific examples

[0328] For example, if User B reads many articles about pet supplies, the server will use the emotion engine to detect from User B's facial expression that he or she is relaxed while browsing the webpage. Based on this data, the server generates advertising text such as "Specialty Pet Food Discount Sale!" and an image of pet food. The generated advertisement is created using bright colors and positive wording to match User B's relaxed state. This advertisement is then embedded in real time into the webpages visited by User B, attracting User B's interest. When the advertisement is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future advertisement generation.

[0329] This system makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating user behavioral and emotional data.

[0330] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0331] Program processing flow

[0332] Step 1:

[0333] The server uses cookies to obtain a user identifier each time a user visits a website. It analyzes the information in the cookie (input) and extracts the user identifier (output). It stores this identifier in a database. Specifically, when an HTTP request is received, it reads the value of the HTTP cookie and records it in the database.

[0334] Step 2:

[0335] The server records user behavioral data such as browsing history, click history, and search keywords in real time. First, it monitors user behavior (input), activates an event logger accordingly, and saves the behavioral data (output) in a database. Specifically, it captures user events that occur within a web page (such as clicks and page transitions) and stores that information in MongoDB or MySQL.

[0336] Step 3:

[0337] The server reads the user's facial expression through the camera installed on the user's device and analyzes the emotional data using the emotion engine. The server obtains the user's facial expression image (input), analyzes it, and generates an emotion score (output). Specifically, it sends the image data to the Emotion API in real time and stores the results in a database.

[0338] Step 4:

[0339] The server processes the collected behavioral and emotional data in batches and analyzes the data. The behavioral and emotional data (input) is processed using a batch processing framework such as Apache Spark to identify the user's interest categories (output). Specifically, the server periodically runs a batch job and clusters the data using a machine learning model.

[0340] Step 5:

[0341] The server sends a request to the generation AI to generate ad text and images based on the identified interest information and real-time emotional state. The interest information and emotional state (input) are processed, and ad creative (output) is obtained by sending a prompt to the generation AI. Specifically, the server sends a prompt to the generation AI, such as "Please generate ad copy with a bright tone that corresponds to a relaxed emotional state, targeted at users interested in pet products."

[0342] Step 6:

[0343] The server uses the emotion engine to appropriately adjust the tone of the generated advertisement. It receives the ad creative (input) and modifies the tone depending on the emotional state. Specifically, if the emotional state is positive, it generates an advertisement with bright images and positive wording.

[0344] Step 7:

[0345] The server analyzes the cookie information again each time the user visits a web page to obtain the user identifier. It receives the HTTP request (input) when the web page is visited, analyzes the cookie information, and extracts the user identifier (output). Specifically, it reads the cookie value included in the HTTP request and uses that information to retrieve it from the user profile database.

[0346] Step 8:

[0347] The server dynamically embeds the acquired ad creative into an HTML template and displays it in the user's browser. By embedding the ad creative (input) into the HTML template, an ad-embedded page (output) is generated. Specifically, it uses a template engine (EJS or Handlebars) to insert ad data into the HTML code and sends the generated HTML to the client.

[0348] Step 9:

[0349] The server monitors user responses to delivered ads (click-through rate, duration, etc.) in real time and records the data. It analyzes user behavior data (input) and generates response data (output). Specifically, it uses the Google Analytics API to monitor user activity and stores that data in a database.

[0350] Step 10:

[0351] The server evaluates the performance of the generative AI model and emotion engine based on the recorded data. It analyzes the reaction data (input) and obtains the model evaluation results (output). Specifically, it uses a Python script to calculate the precision and recall of the model and retrain or tune the generative AI model and emotion engine as needed.

[0352] (Application example 2)

[0353] 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."

[0354] Conventional advertising systems typically generate personalized ads based on user behavior data, but they are unable to reflect the user's real-time emotional state. As a result, ads that do not match the user's emotions are sometimes displayed, making effective ad delivery difficult. Furthermore, there was no established method for collecting real-time user emotional data using mobile devices such as smart glasses and reflecting that data in ad generation. Therefore, a new system was needed to maximize the user experience and improve advertising effectiveness.

[0355] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords; means for analyzing the collected behavioral data to identify the user's interests; means for using a generation AI to generate advertising text and images based on the interest information and real-time emotional state; means for reading the user's facial expressions from a mobile device such as smart glasses and analyzing the emotional data; means for delivering the generated advertisement to the user's display device; and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to combine the user's behavioral data and real-time emotional data to generate and deliver personalized advertisements that are best suited to the user's situation.

[0356] "User browsing history" refers to historical data about the websites and pages a user visits on the Internet.

[0357] "Click history" is historical data about the links and buttons a user clicks on the Internet.

[0358] "Search keywords" are historical data of words and phrases entered by users into search engines, etc.

[0359] "Behavioral data" is data that records a user's activities on the Internet, such as their browsing history, click history, and search keywords.

[0360] "Collection methods" are the technical methods and systems used to collect user behavior data.

[0361] "Means for analyzing and identifying user interests" refers to technical methods or systems that analyze collected behavioral data to identify areas or products in which users are interested.

[0362] "Generative AI" is an artificial intelligence technology that uses machine learning models to generate new text or images based on specified conditions.

[0363] "Real-time emotional state" refers to data that instantly analyzes the user's current emotional state based on facial expressions, voice, etc.

[0364] "Smart glasses" are wearable devices equipped with cameras and sensors that can detect the user's gaze and facial expressions.

[0365] "Means for reading facial expressions and analyzing emotional data" refers to technical methods or systems that use a device to capture a user's facial expressions and use that data to analyze the user's emotions.

[0366] "Display terminal" refers to a device on which a generated advertisement is displayed, such as a smartphone or smart glasses.

[0367] "Delivery means" refers to the technical method or system by which the generated advertisement is transmitted to the user's display device via the Internet.

[0368] "Feedback data" refers to data that is collected from users' reactions and behavior to delivered advertisements and used to optimize the system.

[0369] "Means for measuring effectiveness and optimizing generative AI models" refers to technical methods and systems that analyze the performance of delivered advertisements and improve the performance of generative AI models based on the collected data.

[0370] System Overview

[0371] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral data and real-time emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[0372] Hardware and Software Configuration

[0373] The system mainly uses the following hardware and software:

[0374] Server: Hosts and processes databases, analytics engines, and ad generation AI.

[0375] Smart glasses: Captures the user's facial expressions and gaze and transmits the emotional data to a server.

[0376] Emotion Recognition Model: Recognizing emotions from facial expressions using Keras.

[0377] Face detector: Detects the user's face using dlib.

[0378] Generative AI: Generate ad text and images using GPT-3 and DALL-E.

[0379] User device: the device on which the generated advertisement is displayed (e.g. smartphone, PC, smart glasses).

[0380] Data collection and analysis

[0381] The servers collect the websites and pages you visit on the Internet (browsing history), the links and buttons you click (click history), and the words and phrases you enter into search engines (search keywords). This behavioral data is analyzed to identify your interests.

[0382] The smart glasses use built-in cameras and sensors to capture the user's facial expressions and send the data to a server, which then uses an emotion recognition model to analyze the user's emotional state in real time.

[0383] Ad generation

[0384] The server uses generative AI to generate personalized ad text and images based on the collected behavioral and emotional data. For example, if a user is interested in pet supplies and has a relaxed expression, the server generates ad text saying "Specialty pet food discount sale!" and an attractive image of pet food.

[0385] Ad serving

[0386] The generated ads are then delivered to the user's display device, such as the smart glasses display or smartphone screen, and personalized ads are presented in real time based on the user's gaze and emotional state.

[0387] Measurement and optimization

[0388] The server monitors user responses to the delivered ads and collects the data. The collected data is used to evaluate the performance of the generative AI model, and retraining and tuning are performed as necessary. This allows for optimization to improve the effectiveness of the ads.

[0389] Specific examples

[0390] For example, when a user visits a pet shop, the smart glasses identify the user's gaze. If the user shows interest in pet food and has a relaxed expression, the server uses generative AI to generate advertising text such as "Specialty Pet Food Discount Sale!" along with an image of attractive pet food. This advertisement is displayed in real time on the smart glasses' display, hoping to attract the user's attention.

[0391] Example prompt for a generative AI model:

[0392] User behavior data:

[0393] User ID: user123

[0394] Interests: Pet supplies

[0395] Recent activity: Reading a lot of articles about pet supplies

[0396] Real-time sentiment data:

[0397] Emotion: Relaxed

[0398] Generated ad text:

[0399] Text: Special discount sale on select pet foods!

[0400] Color: Bright

[0401] In this way, ads are dynamically generated that are tailored to the user's circumstances and interests, providing an optimal advertising experience.

[0402] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0403] Step 1:

[0404] When a user visits a website on the Internet, the server obtains a user identifier using cookies. The input is the website the user visits and their activities within it, and the output is the user identifier and behavioral data. This allows the user's browsing history, click history, search keywords, etc. to be collected and stored in a database.

[0405] Step 2:

[0406] When a user wears the smart glasses, the camera and sensors in the smart glasses capture the user's facial expressions. The input is the user's real-time facial expression data, and the output is the captured facial image, which is sent to the server.

[0407] Step 3:

[0408] The server runs an emotion recognition model using the received facial expression data. The input is the captured facial expression image, and the output is data indicating the user's emotional state. The emotion recognition model (Keras model) analyzes this facial expression data and identifies the user's real-time emotional state.

[0409] Step 4:

[0410] The server sends an ad generation request to the generative AI based on the collected behavioral and emotional data. The input is user interest data and real-time emotional data, and the output is ad text and images. The generative AI (GPT-3 and DALL-E) generates personalized ads based on the user's interests and emotions.

[0411] Step 5:

[0412] The generated advertisement is dynamically embedded into an HTML template by the server. The input is the advertisement text and images, and the output is an advertisement creative appropriate for the user's display device. This allows the advertisement to be displayed where the user sees it through smart glasses or a smartphone.

[0413] Step 6:

[0414] The server monitors how users respond to the displayed ads. The input is user reaction data to the ads (number of clicks, time spent, etc.), and the output is effectiveness measurement data. This allows the performance of the ads to be evaluated in real time.

[0415] Step 7:

[0416] The server retrains or tunes the generative AI model based on the collected effectiveness measurement data. The input is the advertising effectiveness measurement data, and the output is an optimized generative AI model. This allows for optimization that can be expected to produce even greater effectiveness in the next ad generation.

[0417] 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.

[0418] 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.

[0419] 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.

[0420] [Second embodiment]

[0421] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0422] 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.

[0423] 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).

[0424] 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.

[0425] 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.

[0426] 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).

[0427] 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.

[0428] 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.

[0429] 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.

[0430] 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.

[0431] 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.

[0432] 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."

[0433] System Overview

[0434] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. This system operates between a server and user devices, collecting and analyzing user behavior data to generate and deliver appropriate advertisements.

[0435] Program processing flow and operation

[0436] Data collection and analysis

[0437] server:

[0438] 1. When a user visits a website, the server obtains a user identifier using cookies, and also collects behavioral data such as browsing history, click history, and search keywords.

[0439] 2. The collected data is stored in a database.

[0440] 3. The server periodically analyzes the data in batches and uses machine learning algorithms and natural language processing techniques to identify user interests. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[0441] Ad generation

[0442] server:

[0443] 1. Based on the identified interest information, the server requests the ad text and images from the generating AI.

[0444] 2. For example, a generative AI model (e.g., GPT-3) generates advertising text like, "Special Offer on the Latest Fitness Equipment!" An image-generating AI like DALL-E generates images of fitness equipment.

[0445] 3. The generated ad text and images are stored in the ad content database.

[0446] Ad serving

[0447] server:

[0448] 1. When a user visits a web page, the server retrieves the user identifier from the cookie.

[0449] 2. The server reads the user's interest information from the user profile database and retrieves the relevant ad creative from the ad content database.

[0450] 3. The acquired ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[0451] Measurement and optimization

[0452] server:

[0453] 1. The server monitors user responses to the delivered advertisements (e.g., click-through rate and duration of visit) and records the data in real time.

[0454] 2. The recorded data is used to evaluate and optimize the generative AI model. The model's performance is evaluated, and retraining and tuning are carried out as necessary to further improve advertising effectiveness.

[0455] Specific examples

[0456] If User A reads many articles about pet supplies, their behavioral data is collected and analyzed. The server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of adorable pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[0457] In this way, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests, maximizing the effectiveness of the advertisements.

[0458] The processing flow will be explained below.

[0459] Step 1:

[0460] server:

[0461] Every time a user visits a website, the server obtains a user identifier using a cookie, and also records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[0462] Step 2:

[0463] server:

[0464] Periodically, collected data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify user interest categories (e.g., pets, travel, technology) and update the user profile database.

[0465] Step 3:

[0466] server:

[0467] Based on the interest information, a request is sent to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty pet food discount sale!". Furthermore, an image-generating AI (e.g., DALL-E) is used to generate attractive images of pet food.

[0468] Step 4:

[0469] server:

[0470] The generated advertisement text and images are stored in an advertisement content database.

[0471] Step 5:

[0472] server:

[0473] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from a user profile database, and retrieves relevant ad creative for that user from an ad content database.

[0474] Step 6:

[0475] server:

[0476] Ad creatives are dynamically embedded into HTML templates and displayed in the user's browser.

[0477] Step 7:

[0478] server:

[0479] It monitors user responses to delivered ads (click-through rate, duration of visit, etc.) in real time and records the data.

[0480] Step 8:

[0481] server:

[0482] The performance of the generative AI model is evaluated based on the collected effectiveness measurement data, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[0483] Step 9:

[0484] User device:

[0485] When an ad is displayed, the user's browser sends user actions such as clicks and skips to the server, which collects this data as feedback and uses it for performance measurement and optimization.

[0486] This series of steps allows personalized advertisements to be generated and delivered to users, maximizing advertising effectiveness.

[0487] Example 1

[0488] 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."

[0489] Conventional advertising systems struggled to generate and deliver personalized ads that accurately reflected user interests, resulting in a tendency for advertising effectiveness to decline. Furthermore, they lacked the functionality to measure the effectiveness of delivered ads in real time and optimize and retrain the AI ​​model, making it difficult to improve advertising accuracy. This could lead to dissatisfaction for both advertisers and users.

[0490] 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.

[0491] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generation AI to generate advertising text and images based on the interest information, means for recording the generated advertising text and images in a database, means for acquiring a user identifier each time the user visits a web page and delivering advertisements based on the user's interest information, and means for measuring the effectiveness of the delivered advertisements and collecting feedback data for optimizing the generation AI model. This enables the generation and delivery of highly accurate personalized advertisements that accurately reflect user interests and concerns, maximizing advertising effectiveness and improving user satisfaction.

[0492] "User browsing history" refers to the history of pages a user has viewed on a website, and is information about which pages a user has visited in the past.

[0493] "Click history" refers to the history of links and buttons that a user clicks on a web page, and is a record of the user's operational behavior.

[0494] "Search keywords" refer to keywords entered by users using search engines, and are data that indicate the user's interests and the information they are looking for.

[0495] "Behavioral data" refers to data about a series of actions a user takes on a website, such as the user's browsing history, click history, and search keywords.

[0496] "Means of collection" refers to the technologies and methods used to obtain and store user behavioral data.

[0497] "Means for analyzing and identifying user interests" refers to technologies and methods for analyzing collected behavioral data and identifying the interests of users.

[0498] "Interest information" refers to information about a user's interests and subjects of interest, and is data that falls into a specific category.

[0499] "Generative AI" refers to algorithms and models that use artificial intelligence technology to create advertising text and images.

[0500] "Dynamic embedding" refers to the real-time insertion of advertisements into web pages accessed by users.

[0501] "Means for measuring effectiveness" refers to technologies and methods for evaluating how users respond to the delivered advertisements, such as recording click-through rates and duration of visits.

[0502] "Feedback data" refers to data collected to evaluate the effectiveness of advertising and is information used to optimize and retrain generative AI models.

[0503] "Optimization means" refers to techniques and methods for improving the performance of generative AI models based on feedback data.

[0504] A "database" refers to a system for systematically managing stored data and retrieving data as needed.

[0505] The present invention provides a system for generating personalized advertisements based on user interests and improving advertising effectiveness. This system operates mainly between a server and a user terminal, and is realized using the following specific hardware and software.

[0506] Data collection and analysis

[0507] server:

[0508] When a user accesses a website, the server obtains a user identifier using cookies. It also collects behavioral data such as the user's browsing history, click history, and search keywords in real time. This data is initially stored in memory and periodically written to a database. Data analysis is performed using machine learning libraries such as Python's Scikit-learn library and TensorFlow. Based on the analysis results, the user's interest categories are identified and added to the user profile.

[0509] Ad generation

[0510] server:

[0511] Based on the identified user interest information, the server requests the generation AI to generate the text and images of the advertisement. The generation AI model uses a natural language processing algorithm (e.g., GPT-3) and an image generation algorithm (e.g., image generation AI). The server sends the following prompt to the generation AI:

[0512] Generate ad text with a "Fitness Equipment" theme. Include a short catchphrase to grab users' attention.

[0513] Example: Special deals on the latest fitness equipment!

[0514] The generated advertisement text and images are recorded in an advertisement content database.

[0515] Ad serving

[0516] server:

[0517] When a user visits a web page, the server retrieves the user's identifier using a cookie. The server reads the user's interests from the user profile database and retrieves the relevant ad creative from the ad content database. The retrieved ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[0518] Measurement and optimization

[0519] server:

[0520] The server monitors user responses to the delivered ads and records click information and ad viewing time in real time. The recorded data is used to evaluate and optimize the generative AI model. The server periodically uses this data to retrain the generative AI model and improve the accuracy of ad generation.

[0521] Specific examples

[0522] If User A reads many articles about pet supplies, that behavioral data is collected in real time. Based on this data, the server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of cute pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server records the click information and uses it as optimization data for the generation AI model. The next time the user visits, a more accurate ad is delivered, improving the effectiveness of the ad.

[0523] As described above, this system generates and delivers personalized advertisements based on user interests, maximizing advertising effectiveness and improving user satisfaction.

[0524] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0525] Step 1: Data collection

[0526] When a user accesses a website, the server obtains a user identifier using a cookie. Specifically, the server reads the cookie information included in the HTTP request and extracts the user identifier. The server then collects behavioral data such as the user's browsing history, click history, and search keywords. This data is temporarily stored in the server's memory and eventually written to a database.

[0527] Input: User access information (cookies, browsing history, click history, search keywords)

[0528] Output: Behavioral data record (database)

[0529] Step 2: Data analysis

[0530] The server periodically analyzes the collected behavioral data in batches. The server uses machine learning algorithms, such as Python's Scikit-learn library and TensorFlow, to identify user interest categories. This analysis allows the server to add interest categories based on the content the user frequently accesses.

[0531] Input: Behavioral data (database)

[0532] Output: Interest categories (user profile database)

[0533] Step 3: Request Ad Generation

[0534] The server sends a prompt to the AI ​​based on the identified interests. For example, if the user is interested in "health and fitness," the server sends the following prompt to the AI:

[0535] Generate ad text for "Fitness Equipment." Include a short catchphrase to grab users' attention.

[0536] Example: Special deals on the latest fitness equipment!

[0537] The AI ​​generator generates ad text based on this prompt, while the image generator algorithm simultaneously generates related images.

[0538] Input: Interest information (user profile database)

[0539] Output: Ad text and images (ad content database)

[0540] Step 4: Save your ad

[0541] The generated ad text and images are stored in an ad content database by the server, which does this by performing write operations to the database. Each ad is given a unique identifier and linked to user interest information.

[0542] Input: Ad text and image

[0543] Output: Record of advertising content (advertising content database)

[0544] Step 5: Serving Ads

[0545] Each time the user visits a new web page, the server retrieves the user identifier from the cookie and looks up the user's interests in a user profile database. The server then retrieves relevant advertisements from an advertising content database and dynamically inserts them into an HTML template that is sent to the browser, where the advertisements are displayed on the user's web page.

[0546] Input: User access information (cookies), interest information (user profile database), advertising content (advertising content database)

[0547] Output: A web page with dynamically generated ads

[0548] Step 6: Measure and optimize

[0549] The server monitors user responses to the delivered ads. Specifically, it records the ad click-through rate and duration in real time. This data is stored in a database and used to evaluate and optimize the generative AI model. The server uses this data to periodically retrain the generative AI model and improve the accuracy of ad generation.

[0550] Input: Ad response data (click-through rate, duration)

[0551] Output: Feedback data (database), optimized generative AI model

[0552] This is the flow of processing in the program for this system, which enables the generation and delivery of highly accurate personalized advertisements based on user interests.

[0553] (Application example 1)

[0554] 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."

[0555] Conventional ad delivery systems have difficulty generating personalized ads that accurately capture user interests, resulting in reduced advertising effectiveness. Additionally, it is difficult to measure the effectiveness of generated ads or optimize the AI ​​model in real time, making it difficult to maximize the effectiveness of ads.

[0556] 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.

[0557] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generative AI to generate advertising text and images based on the interest information, means for delivering the generated advertisement to the user's display device, means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generative AI model, and means for optimizing the generative AI model based on user response data. This makes it possible to generate and deliver advertisements individually personalized for each user and optimize the effectiveness of the advertisements in real time.

[0558] "User browsing history" refers to the history of web pages a user has accessed on the Internet.

[0559] "Click history" refers to the history of which links and ads a user clicked on a website.

[0560] "Search Keywords" refers to words or phrases entered by a user using a search engine.

[0561] "Behavioral data" is a collective term for data related to a series of operations or activities a user performs on the Internet.

[0562] "Generative AI" refers to systems or models that use artificial intelligence techniques to generate content for specific purposes.

[0563] "Advertising text" refers to the text and copy used in an advertisement.

[0564] "Image" means any illustration, photograph or graphic visually displayed in an Advertisement.

[0565] "User display device" refers to the device a user uses to view advertisements or web pages, including smartphones, tablets, and PCs.

[0566] "Delivery" refers to the process of displaying advertising content on a designated user's device.

[0567] "Measuring effectiveness" refers to analyzing users' responses and behavior to the ads delivered and evaluating their performance.

[0568] "Feedback Data" refers to user response data collected to evaluate the performance of an advertisement.

[0569] "User Response Data" means data regarding user clicks, view times, and other interactions with advertisements.

[0570] A "generative AI model" refers to a concrete implementation of an algorithm or system for generating advertising content using artificial intelligence technology.

[0571] "Optimization" refers to using collected data to adjust and refine generative AI models to improve their performance.

[0572] System Overview

[0573] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. The system operates between a server and user terminals, collecting and analyzing user behavior data to generate and deliver appropriate advertisements. Specifically, it includes the following means and processes:

[0574] Data collection and analysis

[0575] The server obtains a user identifier using cookies when a user visits a website. It also collects behavioral data, such as browsing history, click history, and search keywords. This behavioral data is stored in a database, and the server periodically analyzes the data in batches to identify user interests using machine learning algorithms and natural language processing techniques. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[0576] Ad generation

[0577] The server requests ad text and images from the generation AI based on the identified interest information. For example, the generative AI model generates ad text such as "Special Offer on the Latest Fitness Equipment!", and the image generation AI generates images of fitness equipment. The generated ad text and images are stored in the ad content database.

[0578] Ad serving

[0579] When a user visits a web page, the server retrieves the user identifier from the cookie. The server reads the user's interest information from the user profile database and retrieves relevant advertising content from the advertising content database. The retrieved advertising content is dynamically embedded into an HTML template and sent to the user's browser. This allows the user to see personalized ads in real time.

[0580] Measurement and optimization

[0581] The server monitors user responses to the delivered ads (e.g., click-through rate and dwell time) and records the data in real time. This collected data is used to evaluate and optimize the generative AI model. By evaluating the model's performance and re-training or tuning as necessary, the effectiveness of the ads can be further improved.

[0582] Specific examples of technology

[0583] For example, if a user reads many articles about "pet supplies," that behavioral data is collected and analyzed. The server identifies the user's interest category as "pets" and requests the generation AI to generate an ad. The generation AI model generates ad text such as "Specialty pet food discount sale!", and the image generation AI generates images of cute pet food. These ad creatives are embedded in real time into the web pages the user visits. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[0584] Prompt Sentence Examples

[0585] 1. "Create an ad for the latest pet products."

[0586] 2. "Pet supplies images"

[0587] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0588] Step 1:

[0589] Collects behavioral data such as user browsing history, click history, and search keywords.

[0590] Specific behavior:

[0591] Cookies are used in the user's browser to obtain a user identifier. Data on the user's browsing history when visiting websites, the links and ads clicked, and search keywords entered into search engines are collected. This behavioral data is stored in a database on the server.

[0592] Input: User behavior data on the website

[0593] Data processing: Use of cookies to identify user identifiers and collect behavioral data

[0594] Output: Behavioral data is saved to a database

[0595] Step 2:

[0596] Analyze collected behavioral data to identify user interests.

[0597] Specific behavior:

[0598] The server periodically extracts batches of behavioral data from the database and analyzes them using machine learning algorithms and natural language processing techniques. It extracts categories that users frequently access and adds interest categories, such as "health and fitness," to the profile database.

[0599] Input: Behavioral data stored in a database

[0600] Data processing: Analyzing data using machine learning algorithms and natural language processing techniques

[0601] Output: User interest categories are identified and stored in a profile database

[0602] Step 3:

[0603] Uses generative AI to generate ad text and images based on interest information.

[0604] Specific behavior:

[0605] The server sends the specified interest information as an input prompt to a generation AI (e.g., GPT-3 and DALL-E). The generation AI generates ad text such as "Special Offer on the Latest Fitness Equipment!" and a matching image. The generated ad content is stored in an ad content database.

[0606] Input: Interest information

[0607] Data processing: Send prompts to the generative AI to generate text and images

[0608] Output: The generated ad text and images are stored in the ad content database.

[0609] Step 4:

[0610] The generated advertisement is delivered to the user's display device.

[0611] Specific behavior:

[0612] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from the profile database, retrieves relevant advertising content from the advertising content database, dynamically embeds the retrieved advertisement into an HTML template, and sends it to the user's browser.

[0613] Input: User identifier, advertising content

[0614] Data processing: Dynamically embedding advertising content into HTML templates

[0615] Output: A personalized ad is displayed in the user's browser

[0616] Step 5:

[0617] Measure the effectiveness of delivered ads and collect feedback data to optimize generative AI models.

[0618] Specific behavior:

[0619] The server monitors user responses to the delivered ads, such as click-through rates and browser dwell times. The collected data is stored as feedback data and used to evaluate and optimize the generative AI model. The model's performance is analyzed and retrained or adjusted as necessary.

[0620] Input: User response data

[0621] Data processing: Collect and analyze reaction data, and retrain and adjust the model

[0622] Output: An optimized generative AI model

[0623] 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.

[0624] System Overview

[0625] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral and emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[0626] Program processing flow and operation

[0627] Data collection and analysis

[0628] server:

[0629] 1. The server obtains a user identifier using cookies each time a user visits a website, and records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[0630] 2. The server reads the user's facial expressions through the webcam on the web page the user is viewing, and analyzes the emotional data using the emotion engine. This emotional data is also stored in the database.

[0631] Ad generation

[0632] server:

[0633] 1. Periodically collected behavioral and emotional data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify the user's interest categories and emotional state.

[0634] 2. Based on the identified interests and real-time emotional state, the system sends a request to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty Pet Food Sale!", and an image-generating AI such as DALL-E is used to generate attractive images of the pet food.

[0635] 3. The emotion engine generates ads with bright colors and positive words when the user is in a positive emotional state, and conversely, generates ads with calming tones and soothing words when the user is in a negative emotional state.

[0636] Ad serving

[0637] server:

[0638] 1. When a user visits a web page, the server retrieves the user identifier from the cookie, reads the user's interests and emotional state from the user profile database, and retrieves the relevant ad creative for that user from the ad content database.

[0639] 2. The acquired ad creative is dynamically embedded into an HTML template and displayed in the user's browser.

[0640] Measurement and optimization

[0641] server:

[0642] 1. The server monitors user responses to delivered ads (click-through rate, length of stay, etc.) in real time and records the data.

[0643] 2. Based on the recorded data, evaluate the performance of the generative AI model and emotion engine, and retrain or tune it as needed to further improve the effectiveness of advertising.

[0644] Specific examples

[0645] For example, suppose User B reads many articles about pet supplies. While User B is browsing a web page, the emotion engine detects that the user is relaxed from their facial expression. Based on this data, the server generates ad text such as "Specialty Pet Food Discount Sale!" along with an image of pet food. The generated ad is created using bright colors and positive wording to match User B's relaxed state. This ad is then embedded in real time into the web pages visited by User B, attracting User B's interest. When the ad is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future ad generation.

[0646] As described above, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests and real-time emotional state, thereby maximizing advertising effectiveness.

[0647] The processing flow will be explained below.

[0648] Step 1:

[0649] server:

[0650] When a user visits a website, the server uses cookies to obtain a user identifier, and also obtains behavioral data such as the user's browsing history, click history, and search keywords in real time, which are then stored in a database.

[0651] Step 2:

[0652] User device:

[0653] While a user is browsing a website, their facial expressions are captured by a webcam, and this image data is instantly sent to a server.

[0654] Step 3:

[0655] server:

[0656] Using the transmitted image data, the emotion engine analyzes the user's facial expressions to identify their real-time emotional state (e.g., happy, sad, surprised), which is also recorded in a database.

[0657] Step 4:

[0658] server:

[0659] It periodically analyzes behavioral and emotional data through batch processing, and uses machine learning algorithms and natural language processing techniques to identify users' interest categories and emotional states, thereby revealing which categories users are interested in.

[0660] Step 5:

[0661] server:

[0662] Based on the identified interests and real-time emotional state, the system requests a generative AI (e.g., GPT-3, DALL-E) to generate ad text and images. For example, if the emotion engine determines that the user is relaxed, the generative AI will generate ad text that matches the relaxed state, such as "Specialty pet food discount sale!". DALL-E will then generate a bright, relaxing image of the pet food.

[0663] Step 6:

[0664] server:

[0665] The generated advertisement text and images are stored in an advertisement content database.

[0666] Step 7:

[0667] server:

[0668] Each time a user visits a web page, the system retrieves the user identifier from the cookie, reads their interests and emotional state from the user profile database, and retrieves ad creative from the ad content database based on this information.

[0669] Step 8:

[0670] server:

[0671] The resulting ad creative is dynamically embedded into an HTML template and displayed in the user's browser, with the tone and color of the ad adjusted based on the user's emotional state.

[0672] Step 9:

[0673] server:

[0674] It monitors user responses to delivered ads (whether the ad was clicked or skipped, etc.) in real time and records the data.

[0675] Step 10:

[0676] server:

[0677] Based on the collected effectiveness measurement data, the performance of the generative AI model and emotion engine is evaluated, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[0678] Step 11:

[0679] User device:

[0680] When an ad is displayed, the user's browser sends actions such as clicks or skips to the server, and this feedback data is used in the measurement and optimization process.

[0681] This series of steps allows for the generation and delivery of personalized ads that are optimized for the user's interests and emotional state, maximizing advertising effectiveness.

[0682] Example 2

[0683] 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."

[0684] Conventional personalized advertising systems generate ads based solely on user behavioral data and lack the ability to generate ads that take into account the user's real-time emotional state. This makes it difficult to optimize ads based on the user's emotional state and maximize advertising effectiveness. Furthermore, the AI ​​model for generating ads is not adequately optimized based on the effectiveness measurement data of the generated ads, making it difficult to expect continuous improvement in advertising effectiveness.

[0685] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify the user's interests, means for analyzing the collected facial expression data to identify the user's emotional state, means for using a generation AI to generate advertisement text and images based on the interest information and emotional state, means for delivering the generated advertisement to the user's display terminal, and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating the user's behavioral data and emotional data.

[0686] "User browsing history" refers to historical data about the pages a user visits on a website.

[0687] "Click history" is historical data about the links and buttons a user clicks on a website.

[0688] "Search Keywords" are search terms entered by a user into a search engine.

[0689] "Behavioral data" refers to data related to a user's behavior on a website, such as their browsing history, click history, and search keywords.

[0690] "Facial expression data" is data collected based on a user's facial expressions.

[0691] "Emotional state" is information indicating the emotional state of the user, obtained by analyzing the user's facial expression data.

[0692] "Interest Information" is information about a user's interests and categories of interest that are identified through analysis of collected behavioral data.

[0693] "Generative AI" is an artificial intelligence technology that generates text and images based on a given prompt.

[0694] The "user's display terminal" refers to a display device such as a computer or smartphone used by the user.

[0695] "Advertising text" is the text displayed as an advertising message.

[0696] "Advertising image" is image data that is displayed as an advertising message.

[0697] "Feedback data" is data that records user responses to delivered advertisements and is used to optimize generative AI models.

[0698] "Delivery" refers to displaying the generated advertisement on the user's display device.

[0699] "Optimizing generative AI models" refers to making adjustments and improvements to improve the performance of generative AI based on collected feedback data.

[0700] MODE FOR CARRYING OUT THE INVENTION

[0701] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates primarily between a server and a user's device and is designed to provide a rich, personalized advertising experience.

[0702] Data collection and analysis

[0703] server:

[0704] 1. The server obtains a user identifier using a cookie each time a user visits the website. The user identifier is generated as a unique ID and stored in a database.

[0705] 2. The server then records users' browsing history, click history, search keywords, and other behavioral data in real time, which is then stored in a database such as MongoDB or MySQL along with a timestamp.

[0706] 3. The server reads the user's facial expressions through the camera installed on the user's device and analyzes the emotional data using an emotion engine (for example, Microsoft Azure's Emotion API). The resulting emotion score (for example, joy, sadness, surprise, etc.) is also stored in a database.

[0707] Ad generation

[0708] server:

[0709] 1. The server processes the collected behavioral and emotional data in batches, analyzes the data using a batch processing framework such as Apache Spark, and applies a machine learning model (e.g., a clustering algorithm using scikit-learn) to classify the user's interest categories.

[0710] 2. Send a request to a generative AI (e.g., GPT-3 and DALL-E using OpenAI APIs) to generate ad text and images based on the identified interests and real-time emotional state. Create a prompt like the following: "Generate ad copy with a light tone that corresponds to a relaxed emotional state for users interested in pet products."

[0711] 3. The emotion engine is used to appropriately adjust the tone of the generated ad: if the emotional state is positive, the ad content will be generated with bright images and positive language.

[0712] Ad serving

[0713] server:

[0714] 1. Every time a user visits a web page, the cookie information is analyzed to obtain the user identifier again, and an SQL query is issued to obtain the necessary interest information and emotional state from the user profile database.

[0715] 2. The retrieved ad creative is dynamically embedded into an HTML template and displayed in the user's browser using a web server framework such as Node.js. A template engine (e.g., EJS or Handlebars) is used to embed ad data into HTML.

[0716] Measurement and optimization

[0717] server:

[0718] 1. User responses to delivered ads (click-through rate, duration, etc.) are monitored in real time and the data is recorded. This is done using the Google Analytics API, which collects user behavior data and stores it in a database.

[0719] 2. Evaluate the performance of the generative AI model and emotion engine based on the recorded data. Use Python scripts to calculate the precision and recall of the model, and retrain or tune it as needed to improve the overall advertising effectiveness of the system.

[0720] Specific examples

[0721] For example, if User B reads many articles about pet supplies, the server will use the emotion engine to detect from User B's facial expression that he or she is relaxed while browsing the webpage. Based on this data, the server generates advertising text such as "Specialty Pet Food Discount Sale!" and an image of pet food. The generated advertisement is created using bright colors and positive wording to match User B's relaxed state. This advertisement is then embedded in real time into the webpages visited by User B, attracting User B's interest. When the advertisement is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future advertisement generation.

[0722] This system makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating user behavioral and emotional data.

[0723] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0724] Program processing flow

[0725] Step 1:

[0726] The server uses cookies to obtain a user identifier each time a user visits a website. It analyzes the information in the cookie (input) and extracts the user identifier (output). It stores this identifier in a database. Specifically, when an HTTP request is received, it reads the value of the HTTP cookie and records it in the database.

[0727] Step 2:

[0728] The server records user behavioral data such as browsing history, click history, and search keywords in real time. First, it monitors user behavior (input), activates an event logger accordingly, and saves the behavioral data (output) in a database. Specifically, it captures user events that occur within a web page (such as clicks and page transitions) and stores that information in MongoDB or MySQL.

[0729] Step 3:

[0730] The server reads the user's facial expression through the camera installed on the user's device and analyzes the emotional data using the emotion engine. The server obtains the user's facial expression image (input), analyzes it, and generates an emotion score (output). Specifically, it sends the image data to the Emotion API in real time and stores the results in a database.

[0731] Step 4:

[0732] The server processes the collected behavioral and emotional data in batches and analyzes the data. The behavioral and emotional data (input) is processed using a batch processing framework such as Apache Spark to identify the user's interest categories (output). Specifically, the server periodically runs a batch job and clusters the data using a machine learning model.

[0733] Step 5:

[0734] The server sends a request to the generation AI to generate ad text and images based on the identified interest information and real-time emotional state. The interest information and emotional state (input) are processed, and ad creative (output) is obtained by sending a prompt to the generation AI. Specifically, the server sends a prompt to the generation AI, such as "Please generate ad copy with a bright tone that corresponds to a relaxed emotional state, targeted at users interested in pet products."

[0735] Step 6:

[0736] The server uses the emotion engine to appropriately adjust the tone of the generated advertisement. It receives the ad creative (input) and modifies the tone depending on the emotional state. Specifically, if the emotional state is positive, it generates an advertisement with bright images and positive wording.

[0737] Step 7:

[0738] The server analyzes the cookie information again each time the user visits a web page to obtain the user identifier. It receives the HTTP request (input) when the web page is visited, analyzes the cookie information, and extracts the user identifier (output). Specifically, it reads the cookie value included in the HTTP request and uses that information to retrieve it from the user profile database.

[0739] Step 8:

[0740] The server dynamically embeds the acquired ad creative into an HTML template and displays it in the user's browser. By embedding the ad creative (input) into the HTML template, an ad-embedded page (output) is generated. Specifically, it uses a template engine (EJS or Handlebars) to insert ad data into the HTML code and sends the generated HTML to the client.

[0741] Step 9:

[0742] The server monitors user responses to delivered ads (click-through rate, duration, etc.) in real time and records the data. It analyzes user behavior data (input) and generates response data (output). Specifically, it uses the Google Analytics API to monitor user activity and stores that data in a database.

[0743] Step 10:

[0744] The server evaluates the performance of the generative AI model and emotion engine based on the recorded data. It analyzes the reaction data (input) and obtains the model evaluation results (output). Specifically, it uses a Python script to calculate the precision and recall of the model and retrain or tune the generative AI model and emotion engine as needed.

[0745] (Application example 2)

[0746] 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."

[0747] Conventional advertising systems typically generate personalized ads based on user behavior data, but they are unable to reflect the user's real-time emotional state. As a result, ads that do not match the user's emotions are sometimes displayed, making effective ad delivery difficult. Furthermore, there was no established method for collecting real-time user emotional data using mobile devices such as smart glasses and reflecting that data in ad generation. Therefore, a new system was needed to maximize the user experience and improve advertising effectiveness.

[0748] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords; means for analyzing the collected behavioral data to identify the user's interests; means for using a generation AI to generate advertising text and images based on the interest information and real-time emotional state; means for reading the user's facial expressions from a mobile device such as smart glasses and analyzing the emotional data; means for delivering the generated advertisement to the user's display device; and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to combine the user's behavioral data and real-time emotional data to generate and deliver personalized advertisements that are best suited to the user's situation.

[0749] "User browsing history" refers to historical data about the websites and pages a user visits on the Internet.

[0750] "Click history" is historical data about the links and buttons a user clicks on the Internet.

[0751] "Search keywords" are historical data of words and phrases entered by users into search engines, etc.

[0752] "Behavioral data" is data that records a user's activities on the Internet, such as their browsing history, click history, and search keywords.

[0753] "Collection methods" are the technical methods and systems used to collect user behavior data.

[0754] "Means for analyzing and identifying user interests" refers to technical methods or systems that analyze collected behavioral data to identify areas or products in which users are interested.

[0755] "Generative AI" is an artificial intelligence technology that uses machine learning models to generate new text or images based on specified conditions.

[0756] "Real-time emotional state" refers to data that instantly analyzes the user's current emotional state based on facial expressions, voice, etc.

[0757] "Smart glasses" are wearable devices equipped with cameras and sensors that can detect the user's gaze and facial expressions.

[0758] "Means for reading facial expressions and analyzing emotional data" refers to technical methods or systems that use a device to capture a user's facial expressions and use that data to analyze the user's emotions.

[0759] "Display terminal" refers to a device on which a generated advertisement is displayed, such as a smartphone or smart glasses.

[0760] "Delivery means" refers to the technical method or system by which the generated advertisement is transmitted to the user's display device via the Internet.

[0761] "Feedback data" refers to data that is collected from users' reactions and behavior to delivered advertisements and used to optimize the system.

[0762] "Means for measuring effectiveness and optimizing generative AI models" refers to technical methods and systems that analyze the performance of delivered advertisements and improve the performance of generative AI models based on the collected data.

[0763] System Overview

[0764] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral data and real-time emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[0765] Hardware and Software Configuration

[0766] The system mainly uses the following hardware and software:

[0767] Server: Hosts and processes databases, analytics engines, and ad generation AI.

[0768] Smart glasses: Captures the user's facial expressions and gaze and transmits the emotional data to a server.

[0769] Emotion Recognition Model: Recognizing emotions from facial expressions using Keras.

[0770] Face detector: Detects the user's face using dlib.

[0771] Generative AI: Generate ad text and images using GPT-3 and DALL-E.

[0772] User device: the device on which the generated advertisement is displayed (e.g. smartphone, PC, smart glasses).

[0773] Data collection and analysis

[0774] The servers collect the websites and pages you visit on the Internet (browsing history), the links and buttons you click (click history), and the words and phrases you enter into search engines (search keywords). This behavioral data is analyzed to identify your interests.

[0775] The smart glasses use built-in cameras and sensors to capture the user's facial expressions and send the data to a server, which then uses an emotion recognition model to analyze the user's emotional state in real time.

[0776] Ad generation

[0777] The server uses generative AI to generate personalized ad text and images based on the collected behavioral and emotional data. For example, if a user is interested in pet supplies and has a relaxed expression, the server generates ad text saying "Specialty pet food discount sale!" and an attractive image of pet food.

[0778] Ad serving

[0779] The generated ads are then delivered to the user's display device, such as the smart glasses display or smartphone screen, and personalized ads are presented in real time based on the user's gaze and emotional state.

[0780] Measurement and optimization

[0781] The server monitors user responses to the delivered ads and collects the data. The collected data is used to evaluate the performance of the generative AI model, and retraining and tuning are performed as necessary. This allows for optimization to improve the effectiveness of the ads.

[0782] Specific examples

[0783] For example, when a user visits a pet shop, the smart glasses identify the user's gaze. If the user shows interest in pet food and has a relaxed expression, the server uses generative AI to generate advertising text such as "Specialty Pet Food Discount Sale!" along with an image of attractive pet food. This advertisement is displayed in real time on the smart glasses' display, hoping to attract the user's attention.

[0784] Example prompt for a generative AI model:

[0785] User behavior data:

[0786] User ID: user123

[0787] Interests: Pet supplies

[0788] Recent activity: Reading a lot of articles about pet supplies

[0789] Real-time sentiment data:

[0790] Emotion: Relaxed

[0791] Generated ad text:

[0792] Text: Special discount sale on select pet foods!

[0793] Color: Bright

[0794] In this way, ads are dynamically generated that are tailored to the user's circumstances and interests, providing an optimal advertising experience.

[0795] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0796] Step 1:

[0797] When a user visits a website on the Internet, the server obtains a user identifier using cookies. The input is the website the user visits and their activities within it, and the output is the user identifier and behavioral data. This allows the user's browsing history, click history, search keywords, etc. to be collected and stored in a database.

[0798] Step 2:

[0799] When a user wears the smart glasses, the camera and sensors in the smart glasses capture the user's facial expressions. The input is the user's real-time facial expression data, and the output is the captured facial image, which is sent to the server.

[0800] Step 3:

[0801] The server runs an emotion recognition model using the received facial expression data. The input is the captured facial expression image, and the output is data indicating the user's emotional state. The emotion recognition model (Keras model) analyzes this facial expression data and identifies the user's real-time emotional state.

[0802] Step 4:

[0803] The server sends an ad generation request to the generative AI based on the collected behavioral and emotional data. The input is user interest data and real-time emotional data, and the output is ad text and images. The generative AI (GPT-3 and DALL-E) generates personalized ads based on the user's interests and emotions.

[0804] Step 5:

[0805] The generated advertisement is dynamically embedded into an HTML template by the server. The input is the advertisement text and images, and the output is an advertisement creative appropriate for the user's display device. This allows the advertisement to be displayed where the user sees it through smart glasses or a smartphone.

[0806] Step 6:

[0807] The server monitors how users respond to the displayed ads. The input is user reaction data to the ads (number of clicks, time spent, etc.), and the output is effectiveness measurement data. This allows the performance of the ads to be evaluated in real time.

[0808] Step 7:

[0809] The server retrains or tunes the generative AI model based on the collected effectiveness measurement data. The input is the advertising effectiveness measurement data, and the output is an optimized generative AI model. This allows for optimization that can be expected to produce even greater effectiveness in the next ad generation.

[0810] 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.

[0811] 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.

[0812] 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.

[0813] [Third embodiment]

[0814] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0815] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0816] 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).

[0817] 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.

[0818] 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.

[0819] 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).

[0820] 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.

[0821] 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.

[0822] 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.

[0823] 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.

[0824] 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.

[0825] 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."

[0826] System Overview

[0827] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. This system operates between a server and user devices, collecting and analyzing user behavior data to generate and deliver appropriate advertisements.

[0828] Program processing flow and operation

[0829] Data collection and analysis

[0830] server:

[0831] 1. When a user visits a website, the server obtains a user identifier using cookies, and also collects behavioral data such as browsing history, click history, and search keywords.

[0832] 2. The collected data is stored in a database.

[0833] 3. The server periodically analyzes the data in batches and uses machine learning algorithms and natural language processing techniques to identify user interests. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[0834] Ad generation

[0835] server:

[0836] 1. Based on the identified interest information, the server requests the ad text and images from the generating AI.

[0837] 2. For example, a generative AI model (e.g., GPT-3) generates advertising text like, "Special Offer on the Latest Fitness Equipment!" An image-generating AI like DALL-E generates images of fitness equipment.

[0838] 3. The generated ad text and images are stored in the ad content database.

[0839] Ad serving

[0840] server:

[0841] 1. When a user visits a web page, the server retrieves the user identifier from the cookie.

[0842] 2. The server reads the user's interest information from the user profile database and retrieves the relevant ad creative from the ad content database.

[0843] 3. The acquired ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[0844] Measurement and optimization

[0845] server:

[0846] 1. The server monitors user responses to the delivered advertisements (e.g., click-through rate and duration of visit) and records the data in real time.

[0847] 2. The recorded data is used to evaluate and optimize the generative AI model. The model's performance is evaluated, and retraining and tuning are carried out as necessary to further improve advertising effectiveness.

[0848] Specific examples

[0849] If User A reads many articles about pet supplies, their behavioral data is collected and analyzed. The server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of adorable pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[0850] In this way, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests, maximizing the effectiveness of the advertisements.

[0851] The processing flow will be explained below.

[0852] Step 1:

[0853] server:

[0854] Every time a user visits a website, the server obtains a user identifier using a cookie, and also records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[0855] Step 2:

[0856] server:

[0857] Periodically, collected data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify user interest categories (e.g., pets, travel, technology) and update the user profile database.

[0858] Step 3:

[0859] server:

[0860] Based on the interest information, a request is sent to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty pet food discount sale!". Furthermore, an image-generating AI (e.g., DALL-E) is used to generate attractive images of pet food.

[0861] Step 4:

[0862] server:

[0863] The generated advertisement text and images are stored in an advertisement content database.

[0864] Step 5:

[0865] server:

[0866] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from a user profile database, and retrieves relevant ad creative for that user from an ad content database.

[0867] Step 6:

[0868] server:

[0869] Ad creatives are dynamically embedded into HTML templates and displayed in the user's browser.

[0870] Step 7:

[0871] server:

[0872] It monitors user responses to delivered ads (click-through rate, duration of visit, etc.) in real time and records the data.

[0873] Step 8:

[0874] server:

[0875] The performance of the generative AI model is evaluated based on the collected effectiveness measurement data, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[0876] Step 9:

[0877] User device:

[0878] When an ad is displayed, the user's browser sends user actions such as clicks and skips to the server, which collects this data as feedback and uses it for performance measurement and optimization.

[0879] This series of steps allows personalized advertisements to be generated and delivered to users, maximizing advertising effectiveness.

[0880] Example 1

[0881] 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."

[0882] Conventional advertising systems struggled to generate and deliver personalized ads that accurately reflected user interests, resulting in a tendency for advertising effectiveness to decline. Furthermore, they lacked the functionality to measure the effectiveness of delivered ads in real time and optimize and retrain the AI ​​model, making it difficult to improve advertising accuracy. This could lead to dissatisfaction for both advertisers and users.

[0883] 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.

[0884] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generation AI to generate advertising text and images based on the interest information, means for recording the generated advertising text and images in a database, means for acquiring a user identifier each time the user visits a web page and delivering advertisements based on the user's interest information, and means for measuring the effectiveness of the delivered advertisements and collecting feedback data for optimizing the generation AI model. This enables the generation and delivery of highly accurate personalized advertisements that accurately reflect user interests and concerns, maximizing advertising effectiveness and improving user satisfaction.

[0885] "User browsing history" refers to the history of pages a user has viewed on a website, and is information about which pages a user has visited in the past.

[0886] "Click history" refers to the history of links and buttons that a user clicks on a web page, and is a record of the user's operational behavior.

[0887] "Search keywords" refer to keywords entered by users using search engines, and are data that indicate the user's interests and the information they are looking for.

[0888] "Behavioral data" refers to data about a series of actions a user takes on a website, such as the user's browsing history, click history, and search keywords.

[0889] "Means of collection" refers to the technologies and methods used to obtain and store user behavioral data.

[0890] "Means for analyzing and identifying user interests" refers to technologies and methods for analyzing collected behavioral data and identifying the interests of users.

[0891] "Interest information" refers to information about a user's interests and subjects of interest, and is data that falls into a specific category.

[0892] "Generative AI" refers to algorithms and models that use artificial intelligence technology to create advertising text and images.

[0893] "Dynamic embedding" refers to the real-time insertion of advertisements into web pages accessed by users.

[0894] "Means for measuring effectiveness" refers to technologies and methods for evaluating how users respond to the delivered advertisements, such as recording click-through rates and duration of visits.

[0895] "Feedback data" refers to data collected to evaluate the effectiveness of advertising and is information used to optimize and retrain generative AI models.

[0896] "Optimization means" refers to techniques and methods for improving the performance of generative AI models based on feedback data.

[0897] A "database" refers to a system for systematically managing stored data and retrieving data as needed.

[0898] The present invention provides a system for generating personalized advertisements based on user interests and improving advertising effectiveness. This system operates mainly between a server and a user terminal, and is realized using the following specific hardware and software.

[0899] Data collection and analysis

[0900] server:

[0901] When a user accesses a website, the server obtains a user identifier using cookies. It also collects behavioral data such as the user's browsing history, click history, and search keywords in real time. This data is initially stored in memory and periodically written to a database. Data analysis is performed using machine learning libraries such as Python's Scikit-learn library and TensorFlow. Based on the analysis results, the user's interest categories are identified and added to the user profile.

[0902] Ad generation

[0903] server:

[0904] Based on the identified user interest information, the server requests the generation AI to generate the text and images of the advertisement. The generation AI model uses a natural language processing algorithm (e.g., GPT-3) and an image generation algorithm (e.g., image generation AI). The server sends the following prompt to the generation AI:

[0905] Generate ad text with a "Fitness Equipment" theme. Include a short catchphrase to grab users' attention.

[0906] Example: Special deals on the latest fitness equipment!

[0907] The generated advertisement text and images are recorded in an advertisement content database.

[0908] Ad serving

[0909] server:

[0910] When a user visits a web page, the server retrieves the user's identifier using a cookie. The server reads the user's interests from the user profile database and retrieves the relevant ad creative from the ad content database. The retrieved ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[0911] Measurement and optimization

[0912] server:

[0913] The server monitors user responses to the delivered ads and records click information and ad viewing time in real time. The recorded data is used to evaluate and optimize the generative AI model. The server periodically uses this data to retrain the generative AI model and improve the accuracy of ad generation.

[0914] Specific examples

[0915] If User A reads many articles about pet supplies, that behavioral data is collected in real time. Based on this data, the server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of cute pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server records the click information and uses it as optimization data for the generation AI model. The next time the user visits, a more accurate ad is delivered, improving the effectiveness of the ad.

[0916] As described above, this system generates and delivers personalized advertisements based on user interests, maximizing advertising effectiveness and improving user satisfaction.

[0917] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0918] Step 1: Data collection

[0919] When a user accesses a website, the server obtains a user identifier using a cookie. Specifically, the server reads the cookie information included in the HTTP request and extracts the user identifier. The server then collects behavioral data such as the user's browsing history, click history, and search keywords. This data is temporarily stored in the server's memory and eventually written to a database.

[0920] Input: User access information (cookies, browsing history, click history, search keywords)

[0921] Output: Behavioral data record (database)

[0922] Step 2: Data analysis

[0923] The server periodically analyzes the collected behavioral data in batches. The server uses machine learning algorithms, such as Python's Scikit-learn library and TensorFlow, to identify user interest categories. This analysis allows the server to add interest categories based on the content the user frequently accesses.

[0924] Input: Behavioral data (database)

[0925] Output: Interest categories (user profile database)

[0926] Step 3: Request Ad Generation

[0927] The server sends a prompt to the AI ​​based on the identified interests. For example, if the user is interested in "health and fitness," the server sends the following prompt to the AI:

[0928] Generate ad text for "Fitness Equipment." Include a short catchphrase to grab users' attention.

[0929] Example: Special deals on the latest fitness equipment!

[0930] The AI ​​generator generates ad text based on this prompt, while the image generator algorithm simultaneously generates related images.

[0931] Input: Interest information (user profile database)

[0932] Output: Ad text and images (ad content database)

[0933] Step 4: Save your ad

[0934] The generated ad text and images are stored in an ad content database by the server, which does this by performing write operations to the database. Each ad is given a unique identifier and linked to user interest information.

[0935] Input: Ad text and image

[0936] Output: Record of advertising content (advertising content database)

[0937] Step 5: Serving Ads

[0938] Each time the user visits a new web page, the server retrieves the user identifier from the cookie and looks up the user's interests in a user profile database. The server then retrieves relevant advertisements from an advertising content database and dynamically inserts them into an HTML template that is sent to the browser, where the advertisements are displayed on the user's web page.

[0939] Input: User access information (cookies), interest information (user profile database), advertising content (advertising content database)

[0940] Output: A web page with dynamically generated ads

[0941] Step 6: Measure and optimize

[0942] The server monitors user responses to the delivered ads. Specifically, it records the ad click-through rate and duration in real time. This data is stored in a database and used to evaluate and optimize the generative AI model. The server uses this data to periodically retrain the generative AI model and improve the accuracy of ad generation.

[0943] Input: Ad response data (click-through rate, duration)

[0944] Output: Feedback data (database), optimized generative AI model

[0945] This is the flow of processing in the program for this system, which enables the generation and delivery of highly accurate personalized advertisements based on user interests.

[0946] (Application example 1)

[0947] 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."

[0948] Conventional ad delivery systems have difficulty generating personalized ads that accurately capture user interests, resulting in reduced advertising effectiveness. Additionally, it is difficult to measure the effectiveness of generated ads or optimize the AI ​​model in real time, making it difficult to maximize the effectiveness of ads.

[0949] 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.

[0950] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generative AI to generate advertising text and images based on the interest information, means for delivering the generated advertisement to the user's display device, means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generative AI model, and means for optimizing the generative AI model based on user response data. This makes it possible to generate and deliver advertisements individually personalized for each user and optimize the effectiveness of the advertisements in real time.

[0951] "User browsing history" refers to the history of web pages a user has accessed on the Internet.

[0952] "Click history" refers to the history of which links and ads a user clicked on a website.

[0953] "Search Keywords" refers to words or phrases entered by a user using a search engine.

[0954] "Behavioral data" is a collective term for data related to a series of operations or activities a user performs on the Internet.

[0955] "Generative AI" refers to systems or models that use artificial intelligence techniques to generate content for specific purposes.

[0956] "Advertising text" refers to the text and copy used in an advertisement.

[0957] "Image" means any illustration, photograph or graphic visually displayed in an Advertisement.

[0958] "User display device" refers to the device a user uses to view advertisements or web pages, including smartphones, tablets, and PCs.

[0959] "Delivery" refers to the process of displaying advertising content on a designated user's device.

[0960] "Measuring effectiveness" refers to analyzing users' responses and behavior to the ads delivered and evaluating their performance.

[0961] "Feedback Data" refers to user response data collected to evaluate the performance of an advertisement.

[0962] "User Response Data" means data regarding user clicks, view times, and other interactions with advertisements.

[0963] A "generative AI model" refers to a concrete implementation of an algorithm or system for generating advertising content using artificial intelligence technology.

[0964] "Optimization" refers to using collected data to adjust and refine generative AI models to improve their performance.

[0965] System Overview

[0966] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. The system operates between a server and user terminals, collecting and analyzing user behavior data to generate and deliver appropriate advertisements. Specifically, it includes the following means and processes:

[0967] Data collection and analysis

[0968] The server obtains a user identifier using cookies when a user visits a website. It also collects behavioral data, such as browsing history, click history, and search keywords. This behavioral data is stored in a database, and the server periodically analyzes the data in batches to identify user interests using machine learning algorithms and natural language processing techniques. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[0969] Ad generation

[0970] The server requests ad text and images from the generation AI based on the identified interest information. For example, the generative AI model generates ad text such as "Special Offer on the Latest Fitness Equipment!", and the image generation AI generates images of fitness equipment. The generated ad text and images are stored in the ad content database.

[0971] Ad serving

[0972] When a user visits a web page, the server retrieves the user identifier from the cookie. The server reads the user's interest information from the user profile database and retrieves relevant advertising content from the advertising content database. The retrieved advertising content is dynamically embedded into an HTML template and sent to the user's browser. This allows the user to see personalized ads in real time.

[0973] Measurement and optimization

[0974] The server monitors user responses to the delivered ads (e.g., click-through rate and dwell time) and records the data in real time. This collected data is used to evaluate and optimize the generative AI model. By evaluating the model's performance and re-training or tuning as necessary, the effectiveness of the ads can be further improved.

[0975] Specific examples of technology

[0976] For example, if a user reads many articles about "pet supplies," that behavioral data is collected and analyzed. The server identifies the user's interest category as "pets" and requests the generation AI to generate an ad. The generation AI model generates ad text such as "Specialty pet food discount sale!", and the image generation AI generates images of cute pet food. These ad creatives are embedded in real time into the web pages the user visits. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[0977] Prompt Sentence Examples

[0978] 1. "Create an ad for the latest pet products."

[0979] 2. "Pet supplies images"

[0980] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0981] Step 1:

[0982] Collects behavioral data such as user browsing history, click history, and search keywords.

[0983] Specific behavior:

[0984] Cookies are used in the user's browser to obtain a user identifier. Data on the user's browsing history when visiting websites, the links and ads clicked, and search keywords entered into search engines are collected. This behavioral data is stored in a database on the server.

[0985] Input: User behavior data on the website

[0986] Data processing: Use of cookies to identify user identifiers and collect behavioral data

[0987] Output: Behavioral data is saved to a database

[0988] Step 2:

[0989] Analyze collected behavioral data to identify user interests.

[0990] Specific behavior:

[0991] The server periodically extracts batches of behavioral data from the database and analyzes them using machine learning algorithms and natural language processing techniques. It extracts categories that users frequently access and adds interest categories, such as "health and fitness," to the profile database.

[0992] Input: Behavioral data stored in a database

[0993] Data processing: Analyzing data using machine learning algorithms and natural language processing techniques

[0994] Output: User interest categories are identified and stored in a profile database

[0995] Step 3:

[0996] Uses generative AI to generate ad text and images based on interest information.

[0997] Specific behavior:

[0998] The server sends the specified interest information as an input prompt to a generation AI (e.g., GPT-3 and DALL-E). The generation AI generates ad text such as "Special Offer on the Latest Fitness Equipment!" and a matching image. The generated ad content is stored in an ad content database.

[0999] Input: Interest information

[1000] Data processing: Send prompts to the generative AI to generate text and images

[1001] Output: The generated ad text and images are stored in the ad content database.

[1002] Step 4:

[1003] The generated advertisement is delivered to the user's display device.

[1004] Specific behavior:

[1005] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from the profile database, retrieves relevant advertising content from the advertising content database, dynamically embeds the retrieved advertisement into an HTML template, and sends it to the user's browser.

[1006] Input: User identifier, advertising content

[1007] Data processing: Dynamically embedding advertising content into HTML templates

[1008] Output: A personalized ad is displayed in the user's browser

[1009] Step 5:

[1010] Measure the effectiveness of delivered ads and collect feedback data to optimize generative AI models.

[1011] Specific behavior:

[1012] The server monitors user responses to the delivered ads, such as click-through rates and browser dwell times. The collected data is stored as feedback data and used to evaluate and optimize the generative AI model. The model's performance is analyzed and retrained or adjusted as necessary.

[1013] Input: User response data

[1014] Data processing: Collect and analyze reaction data, and retrain and adjust the model

[1015] Output: An optimized generative AI model

[1016] 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.

[1017] System Overview

[1018] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral and emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[1019] Program processing flow and operation

[1020] Data collection and analysis

[1021] server:

[1022] 1. The server obtains a user identifier using cookies each time a user visits a website, and records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[1023] 2. The server reads the user's facial expressions through the webcam on the web page the user is viewing, and analyzes the emotional data using the emotion engine. This emotional data is also stored in the database.

[1024] Ad generation

[1025] server:

[1026] 1. Periodically collected behavioral and emotional data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify the user's interest categories and emotional state.

[1027] 2. Based on the identified interests and real-time emotional state, the system sends a request to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty Pet Food Sale!", and an image-generating AI such as DALL-E is used to generate attractive images of the pet food.

[1028] 3. The emotion engine generates ads with bright colors and positive words when the user is in a positive emotional state, and conversely, generates ads with calming tones and soothing words when the user is in a negative emotional state.

[1029] Ad serving

[1030] server:

[1031] 1. When a user visits a web page, the server retrieves the user identifier from the cookie, reads the user's interests and emotional state from the user profile database, and retrieves the relevant ad creative for that user from the ad content database.

[1032] 2. The acquired ad creative is dynamically embedded into an HTML template and displayed in the user's browser.

[1033] Measurement and optimization

[1034] server:

[1035] 1. The server monitors user responses to delivered ads (click-through rate, length of stay, etc.) in real time and records the data.

[1036] 2. Based on the recorded data, evaluate the performance of the generative AI model and emotion engine, and retrain or tune it as needed to further improve the effectiveness of advertising.

[1037] Specific examples

[1038] For example, suppose User B reads many articles about pet supplies. While User B is browsing a web page, the emotion engine detects that the user is relaxed from their facial expression. Based on this data, the server generates ad text such as "Specialty Pet Food Discount Sale!" along with an image of pet food. The generated ad is created using bright colors and positive wording to match User B's relaxed state. This ad is then embedded in real time into the web pages visited by User B, attracting User B's interest. When the ad is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future ad generation.

[1039] As described above, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests and real-time emotional state, thereby maximizing advertising effectiveness.

[1040] The processing flow will be explained below.

[1041] Step 1:

[1042] server:

[1043] When a user visits a website, the server uses cookies to obtain a user identifier, and also obtains behavioral data such as the user's browsing history, click history, and search keywords in real time, which are then stored in a database.

[1044] Step 2:

[1045] User device:

[1046] While a user is browsing a website, their facial expressions are captured by a webcam, and this image data is instantly sent to a server.

[1047] Step 3:

[1048] server:

[1049] Using the transmitted image data, the emotion engine analyzes the user's facial expressions to identify their real-time emotional state (e.g., happy, sad, surprised), which is also recorded in a database.

[1050] Step 4:

[1051] server:

[1052] It periodically analyzes behavioral and emotional data through batch processing, and uses machine learning algorithms and natural language processing techniques to identify users' interest categories and emotional states, thereby revealing which categories users are interested in.

[1053] Step 5:

[1054] server:

[1055] Based on the identified interests and real-time emotional state, the system requests a generative AI (e.g., GPT-3, DALL-E) to generate ad text and images. For example, if the emotion engine determines that the user is relaxed, the generative AI will generate ad text that matches the relaxed state, such as "Specialty pet food discount sale!". DALL-E will then generate a bright, relaxing image of the pet food.

[1056] Step 6:

[1057] server:

[1058] The generated advertisement text and images are stored in an advertisement content database.

[1059] Step 7:

[1060] server:

[1061] Each time a user visits a web page, the system retrieves the user identifier from the cookie, reads their interests and emotional state from the user profile database, and retrieves ad creative from the ad content database based on this information.

[1062] Step 8:

[1063] server:

[1064] The resulting ad creative is dynamically embedded into an HTML template and displayed in the user's browser, with the tone and color of the ad adjusted based on the user's emotional state.

[1065] Step 9:

[1066] server:

[1067] It monitors user responses to delivered ads (whether the ad was clicked or skipped, etc.) in real time and records the data.

[1068] Step 10:

[1069] server:

[1070] Based on the collected effectiveness measurement data, the performance of the generative AI model and emotion engine is evaluated, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[1071] Step 11:

[1072] User device:

[1073] When an ad is displayed, the user's browser sends actions such as clicks or skips to the server, and this feedback data is used in the measurement and optimization process.

[1074] This series of steps allows for the generation and delivery of personalized ads that are optimized for the user's interests and emotional state, maximizing advertising effectiveness.

[1075] Example 2

[1076] 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."

[1077] Conventional personalized advertising systems generate ads based solely on user behavioral data and lack the ability to generate ads that take into account the user's real-time emotional state. This makes it difficult to optimize ads based on the user's emotional state and maximize advertising effectiveness. Furthermore, the AI ​​model for generating ads is not adequately optimized based on the effectiveness measurement data of the generated ads, making it difficult to expect continuous improvement in advertising effectiveness.

[1078] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify the user's interests, means for analyzing the collected facial expression data to identify the user's emotional state, means for using a generation AI to generate advertisement text and images based on the interest information and emotional state, means for delivering the generated advertisement to the user's display terminal, and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating the user's behavioral data and emotional data.

[1079] "User browsing history" refers to historical data about the pages a user visits on a website.

[1080] "Click history" is historical data about the links and buttons a user clicks on a website.

[1081] "Search Keywords" are search terms entered by a user into a search engine.

[1082] "Behavioral data" refers to data related to a user's behavior on a website, such as their browsing history, click history, and search keywords.

[1083] "Facial expression data" is data collected based on a user's facial expressions.

[1084] "Emotional state" is information indicating the emotional state of the user, obtained by analyzing the user's facial expression data.

[1085] "Interest Information" is information about a user's interests and categories of interest that are identified through analysis of collected behavioral data.

[1086] "Generative AI" is an artificial intelligence technology that generates text and images based on a given prompt.

[1087] The "user's display terminal" refers to a display device such as a computer or smartphone used by the user.

[1088] "Advertising text" is the text displayed as an advertising message.

[1089] "Advertising image" is image data that is displayed as an advertising message.

[1090] "Feedback data" is data that records user responses to delivered advertisements and is used to optimize generative AI models.

[1091] "Delivery" refers to displaying the generated advertisement on the user's display device.

[1092] "Optimizing generative AI models" refers to making adjustments and improvements to improve the performance of generative AI based on collected feedback data.

[1093] MODE FOR CARRYING OUT THE INVENTION

[1094] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates primarily between a server and a user's device and is designed to provide a rich, personalized advertising experience.

[1095] Data collection and analysis

[1096] server:

[1097] 1. The server obtains a user identifier using a cookie each time a user visits the website. The user identifier is generated as a unique ID and stored in a database.

[1098] 2. The server then records users' browsing history, click history, search keywords, and other behavioral data in real time, which is then stored in a database such as MongoDB or MySQL along with a timestamp.

[1099] 3. The server reads the user's facial expressions through the camera installed on the user's device and analyzes the emotional data using an emotion engine (for example, Microsoft Azure's Emotion API). The resulting emotion score (for example, joy, sadness, surprise, etc.) is also stored in a database.

[1100] Ad generation

[1101] server:

[1102] 1. The server processes the collected behavioral and emotional data in batches, analyzes the data using a batch processing framework such as Apache Spark, and applies a machine learning model (e.g., a clustering algorithm using scikit-learn) to classify the user's interest categories.

[1103] 2. Send a request to a generative AI (e.g., GPT-3 and DALL-E using OpenAI APIs) to generate ad text and images based on the identified interests and real-time emotional state. Create a prompt like the following: "Generate ad copy with a light tone that corresponds to a relaxed emotional state for users interested in pet products."

[1104] 3. The emotion engine is used to appropriately adjust the tone of the generated ad: if the emotional state is positive, the ad content will be generated with bright images and positive language.

[1105] Ad serving

[1106] server:

[1107] 1. Every time a user visits a web page, the cookie information is analyzed to obtain the user identifier again, and an SQL query is issued to obtain the necessary interest information and emotional state from the user profile database.

[1108] 2. The retrieved ad creative is dynamically embedded into an HTML template and displayed in the user's browser using a web server framework such as Node.js. A template engine (e.g., EJS or Handlebars) is used to embed ad data into HTML.

[1109] Measurement and optimization

[1110] server:

[1111] 1. User responses to delivered ads (click-through rate, duration, etc.) are monitored in real time and the data is recorded. This is done using the Google Analytics API, which collects user behavior data and stores it in a database.

[1112] 2. Evaluate the performance of the generative AI model and emotion engine based on the recorded data. Use Python scripts to calculate the precision and recall of the model, and retrain or tune it as needed to improve the overall advertising effectiveness of the system.

[1113] Specific examples

[1114] For example, if User B reads many articles about pet supplies, the server will use the emotion engine to detect from User B's facial expression that he or she is relaxed while browsing the webpage. Based on this data, the server generates advertising text such as "Specialty Pet Food Discount Sale!" and an image of pet food. The generated advertisement is created using bright colors and positive wording to match User B's relaxed state. This advertisement is then embedded in real time into the webpages visited by User B, attracting User B's interest. When the advertisement is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future advertisement generation.

[1115] This system makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating user behavioral and emotional data.

[1116] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1117] Program processing flow

[1118] Step 1:

[1119] The server uses cookies to obtain a user identifier each time a user visits a website. It analyzes the information in the cookie (input) and extracts the user identifier (output). It stores this identifier in a database. Specifically, when an HTTP request is received, it reads the value of the HTTP cookie and records it in the database.

[1120] Step 2:

[1121] The server records user behavioral data such as browsing history, click history, and search keywords in real time. First, it monitors user behavior (input), activates an event logger accordingly, and saves the behavioral data (output) in a database. Specifically, it captures user events that occur within a web page (such as clicks and page transitions) and stores that information in MongoDB or MySQL.

[1122] Step 3:

[1123] The server reads the user's facial expression through the camera installed on the user's device and analyzes the emotional data using the emotion engine. The server obtains the user's facial expression image (input), analyzes it, and generates an emotion score (output). Specifically, it sends the image data to the Emotion API in real time and stores the results in a database.

[1124] Step 4:

[1125] The server processes the collected behavioral and emotional data in batches and analyzes the data. The behavioral and emotional data (input) is processed using a batch processing framework such as Apache Spark to identify the user's interest categories (output). Specifically, the server periodically runs a batch job and clusters the data using a machine learning model.

[1126] Step 5:

[1127] The server sends a request to the generation AI to generate ad text and images based on the identified interest information and real-time emotional state. The interest information and emotional state (input) are processed, and ad creative (output) is obtained by sending a prompt to the generation AI. Specifically, the server sends a prompt to the generation AI, such as "Please generate ad copy with a bright tone that corresponds to a relaxed emotional state, targeted at users interested in pet products."

[1128] Step 6:

[1129] The server uses the emotion engine to appropriately adjust the tone of the generated advertisement. It receives the ad creative (input) and modifies the tone depending on the emotional state. Specifically, if the emotional state is positive, it generates an advertisement with bright images and positive wording.

[1130] Step 7:

[1131] The server analyzes the cookie information again each time the user visits a web page to obtain the user identifier. It receives the HTTP request (input) when the web page is visited, analyzes the cookie information, and extracts the user identifier (output). Specifically, it reads the cookie value included in the HTTP request and uses that information to retrieve it from the user profile database.

[1132] Step 8:

[1133] The server dynamically embeds the acquired ad creative into an HTML template and displays it in the user's browser. By embedding the ad creative (input) into the HTML template, an ad-embedded page (output) is generated. Specifically, it uses a template engine (EJS or Handlebars) to insert ad data into the HTML code and sends the generated HTML to the client.

[1134] Step 9:

[1135] The server monitors user responses to delivered ads (click-through rate, duration, etc.) in real time and records the data. It analyzes user behavior data (input) and generates response data (output). Specifically, it uses the Google Analytics API to monitor user activity and stores that data in a database.

[1136] Step 10:

[1137] The server evaluates the performance of the generative AI model and emotion engine based on the recorded data. It analyzes the reaction data (input) and obtains the model evaluation results (output). Specifically, it uses a Python script to calculate the precision and recall of the model and retrain or tune the generative AI model and emotion engine as needed.

[1138] (Application example 2)

[1139] 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."

[1140] Conventional advertising systems typically generate personalized ads based on user behavior data, but they are unable to reflect the user's real-time emotional state. As a result, ads that do not match the user's emotions are sometimes displayed, making effective ad delivery difficult. Furthermore, there was no established method for collecting real-time user emotional data using mobile devices such as smart glasses and reflecting that data in ad generation. Therefore, a new system was needed to maximize the user experience and improve advertising effectiveness.

[1141] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords; means for analyzing the collected behavioral data to identify the user's interests; means for using a generation AI to generate advertising text and images based on the interest information and real-time emotional state; means for reading the user's facial expressions from a mobile device such as smart glasses and analyzing the emotional data; means for delivering the generated advertisement to the user's display device; and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to combine the user's behavioral data and real-time emotional data to generate and deliver personalized advertisements that are best suited to the user's situation.

[1142] "User browsing history" refers to historical data about the websites and pages a user visits on the Internet.

[1143] "Click history" is historical data about the links and buttons a user clicks on the Internet.

[1144] "Search keywords" are historical data of words and phrases entered by users into search engines, etc.

[1145] "Behavioral data" is data that records a user's activities on the Internet, such as their browsing history, click history, and search keywords.

[1146] "Collection methods" are the technical methods and systems used to collect user behavior data.

[1147] "Means for analyzing and identifying user interests" refers to technical methods or systems that analyze collected behavioral data to identify areas or products in which users are interested.

[1148] "Generative AI" is an artificial intelligence technology that uses machine learning models to generate new text or images based on specified conditions.

[1149] "Real-time emotional state" refers to data that instantly analyzes the user's current emotional state based on facial expressions, voice, etc.

[1150] "Smart glasses" are wearable devices equipped with cameras and sensors that can detect the user's gaze and facial expressions.

[1151] "Means for reading facial expressions and analyzing emotional data" refers to technical methods or systems that use a device to capture a user's facial expressions and use that data to analyze the user's emotions.

[1152] "Display terminal" refers to a device on which a generated advertisement is displayed, such as a smartphone or smart glasses.

[1153] "Delivery means" refers to the technical method or system by which the generated advertisement is transmitted to the user's display device via the Internet.

[1154] "Feedback data" refers to data that is collected from users' reactions and behavior to delivered advertisements and used to optimize the system.

[1155] "Means for measuring effectiveness and optimizing generative AI models" refers to technical methods and systems that analyze the performance of delivered advertisements and improve the performance of generative AI models based on the collected data.

[1156] System Overview

[1157] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral data and real-time emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[1158] Hardware and Software Configuration

[1159] The system mainly uses the following hardware and software:

[1160] Server: Hosts and processes databases, analytics engines, and ad generation AI.

[1161] Smart glasses: Captures the user's facial expressions and gaze and transmits the emotional data to a server.

[1162] Emotion Recognition Model: Recognizing emotions from facial expressions using Keras.

[1163] Face detector: Detects the user's face using dlib.

[1164] Generative AI: Generate ad text and images using GPT-3 and DALL-E.

[1165] User device: the device on which the generated advertisement is displayed (e.g. smartphone, PC, smart glasses).

[1166] Data collection and analysis

[1167] The servers collect the websites and pages you visit on the Internet (browsing history), the links and buttons you click (click history), and the words and phrases you enter into search engines (search keywords). This behavioral data is analyzed to identify your interests.

[1168] The smart glasses use built-in cameras and sensors to capture the user's facial expressions and send the data to a server, which then uses an emotion recognition model to analyze the user's emotional state in real time.

[1169] Ad generation

[1170] The server uses generative AI to generate personalized ad text and images based on the collected behavioral and emotional data. For example, if a user is interested in pet supplies and has a relaxed expression, the server generates ad text saying "Specialty pet food discount sale!" and an attractive image of pet food.

[1171] Ad serving

[1172] The generated ads are then delivered to the user's display device, such as the smart glasses display or smartphone screen, and personalized ads are presented in real time based on the user's gaze and emotional state.

[1173] Measurement and optimization

[1174] The server monitors user responses to the delivered ads and collects the data. The collected data is used to evaluate the performance of the generative AI model, and retraining and tuning are performed as necessary. This allows for optimization to improve the effectiveness of the ads.

[1175] Specific examples

[1176] For example, when a user visits a pet shop, the smart glasses identify the user's gaze. If the user shows interest in pet food and has a relaxed expression, the server uses generative AI to generate advertising text such as "Specialty Pet Food Discount Sale!" along with an image of attractive pet food. This advertisement is displayed in real time on the smart glasses' display, hoping to attract the user's attention.

[1177] Example prompt for a generative AI model:

[1178] User behavior data:

[1179] User ID: user123

[1180] Interests: Pet supplies

[1181] Recent activity: Reading a lot of articles about pet supplies

[1182] Real-time sentiment data:

[1183] Emotion: Relaxed

[1184] Generated ad text:

[1185] Text: Special discount sale on select pet foods!

[1186] Color: Bright

[1187] In this way, ads are dynamically generated that are tailored to the user's circumstances and interests, providing an optimal advertising experience.

[1188] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1189] Step 1:

[1190] When a user visits a website on the Internet, the server obtains a user identifier using cookies. The input is the website the user visits and their activities within it, and the output is the user identifier and behavioral data. This allows the user's browsing history, click history, search keywords, etc. to be collected and stored in a database.

[1191] Step 2:

[1192] When a user wears the smart glasses, the camera and sensors in the smart glasses capture the user's facial expressions. The input is the user's real-time facial expression data, and the output is the captured facial image, which is sent to the server.

[1193] Step 3:

[1194] The server runs an emotion recognition model using the received facial expression data. The input is the captured facial expression image, and the output is data indicating the user's emotional state. The emotion recognition model (Keras model) analyzes this facial expression data and identifies the user's real-time emotional state.

[1195] Step 4:

[1196] The server sends an ad generation request to the generative AI based on the collected behavioral and emotional data. The input is user interest data and real-time emotional data, and the output is ad text and images. The generative AI (GPT-3 and DALL-E) generates personalized ads based on the user's interests and emotions.

[1197] Step 5:

[1198] The generated advertisement is dynamically embedded into an HTML template by the server. The input is the advertisement text and images, and the output is an advertisement creative appropriate for the user's display device. This allows the advertisement to be displayed where the user sees it through smart glasses or a smartphone.

[1199] Step 6:

[1200] The server monitors how users respond to the displayed ads. The input is user reaction data to the ads (number of clicks, time spent, etc.), and the output is effectiveness measurement data. This allows the performance of the ads to be evaluated in real time.

[1201] Step 7:

[1202] The server retrains or tunes the generative AI model based on the collected effectiveness measurement data. The input is the advertising effectiveness measurement data, and the output is an optimized generative AI model. This allows for optimization that can be expected to produce even greater effectiveness in the next ad generation.

[1203] 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.

[1204] 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.

[1205] 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.

[1206] [Fourth embodiment]

[1207] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1208] 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.

[1209] 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).

[1210] 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.

[1211] 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.

[1212] 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).

[1213] 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.

[1214] 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.

[1215] 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.

[1216] 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.

[1217] 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.

[1218] 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.

[1219] 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."

[1220] System Overview

[1221] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. This system operates between a server and user devices, collecting and analyzing user behavior data to generate and deliver appropriate advertisements.

[1222] Program processing flow and operation

[1223] Data collection and analysis

[1224] server:

[1225] 1. When a user visits a website, the server obtains a user identifier using cookies, and also collects behavioral data such as browsing history, click history, and search keywords.

[1226] 2. The collected data is stored in a database.

[1227] 3. The server periodically analyzes the data in batches and uses machine learning algorithms and natural language processing techniques to identify user interests. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[1228] Ad generation

[1229] server:

[1230] 1. Based on the identified interest information, the server requests the ad text and images from the generating AI.

[1231] 2. For example, a generative AI model (e.g., GPT-3) generates advertising text like, "Special Offer on the Latest Fitness Equipment!" An image-generating AI like DALL-E generates images of fitness equipment.

[1232] 3. The generated ad text and images are stored in the ad content database.

[1233] Ad serving

[1234] server:

[1235] 1. When a user visits a web page, the server retrieves the user identifier from the cookie.

[1236] 2. The server reads the user's interest information from the user profile database and retrieves the relevant ad creative from the ad content database.

[1237] 3. The acquired ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[1238] Measurement and optimization

[1239] server:

[1240] 1. The server monitors user responses to the delivered advertisements (e.g., click-through rate and duration of visit) and records the data in real time.

[1241] 2. The recorded data is used to evaluate and optimize the generative AI model. The model's performance is evaluated, and retraining and tuning are carried out as necessary to further improve advertising effectiveness.

[1242] Specific examples

[1243] If User A reads many articles about pet supplies, their behavioral data is collected and analyzed. The server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of adorable pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[1244] In this way, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests, maximizing the effectiveness of the advertisements.

[1245] The processing flow will be explained below.

[1246] Step 1:

[1247] server:

[1248] Every time a user visits a website, the server obtains a user identifier using a cookie, and also records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[1249] Step 2:

[1250] server:

[1251] Periodically, collected data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify user interest categories (e.g., pets, travel, technology) and update the user profile database.

[1252] Step 3:

[1253] server:

[1254] Based on the interest information, a request is sent to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty pet food discount sale!". Furthermore, an image-generating AI (e.g., DALL-E) is used to generate attractive images of pet food.

[1255] Step 4:

[1256] server:

[1257] The generated advertisement text and images are stored in an advertisement content database.

[1258] Step 5:

[1259] server:

[1260] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from a user profile database, and retrieves relevant ad creative for that user from an ad content database.

[1261] Step 6:

[1262] server:

[1263] Ad creatives are dynamically embedded into HTML templates and displayed in the user's browser.

[1264] Step 7:

[1265] server:

[1266] It monitors user responses to delivered ads (click-through rate, duration of visit, etc.) in real time and records the data.

[1267] Step 8:

[1268] server:

[1269] The performance of the generative AI model is evaluated based on the collected effectiveness measurement data, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[1270] Step 9:

[1271] User device:

[1272] When an ad is displayed, the user's browser sends user actions such as clicks and skips to the server, which collects this data as feedback and uses it for performance measurement and optimization.

[1273] This series of steps allows personalized advertisements to be generated and delivered to users, maximizing advertising effectiveness.

[1274] Example 1

[1275] 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."

[1276] Conventional advertising systems struggled to generate and deliver personalized ads that accurately reflected user interests, resulting in a tendency for advertising effectiveness to decline. Furthermore, they lacked the functionality to measure the effectiveness of delivered ads in real time and optimize and retrain the AI ​​model, making it difficult to improve advertising accuracy. This could lead to dissatisfaction for both advertisers and users.

[1277] 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.

[1278] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generation AI to generate advertising text and images based on the interest information, means for recording the generated advertising text and images in a database, means for acquiring a user identifier each time the user visits a web page and delivering advertisements based on the user's interest information, and means for measuring the effectiveness of the delivered advertisements and collecting feedback data for optimizing the generation AI model. This enables the generation and delivery of highly accurate personalized advertisements that accurately reflect user interests and concerns, maximizing advertising effectiveness and improving user satisfaction.

[1279] "User browsing history" refers to the history of pages a user has viewed on a website, and is information about which pages a user has visited in the past.

[1280] "Click history" refers to the history of links and buttons that a user clicks on a web page, and is a record of the user's operational behavior.

[1281] "Search keywords" refer to keywords entered by users using search engines, and are data that indicate the user's interests and the information they are looking for.

[1282] "Behavioral data" refers to data about a series of actions a user takes on a website, such as the user's browsing history, click history, and search keywords.

[1283] "Means of collection" refers to the technologies and methods used to obtain and store user behavioral data.

[1284] "Means for analyzing and identifying user interests" refers to technologies and methods for analyzing collected behavioral data and identifying the interests of users.

[1285] "Interest information" refers to information about a user's interests and subjects of interest, and is data that falls into a specific category.

[1286] "Generative AI" refers to algorithms and models that use artificial intelligence technology to create advertising text and images.

[1287] "Dynamic embedding" refers to the real-time insertion of advertisements into web pages accessed by users.

[1288] "Means for measuring effectiveness" refers to technologies and methods for evaluating how users respond to the delivered advertisements, such as recording click-through rates and duration of visits.

[1289] "Feedback data" refers to data collected to evaluate the effectiveness of advertising and is information used to optimize and retrain generative AI models.

[1290] "Optimization means" refers to techniques and methods for improving the performance of generative AI models based on feedback data.

[1291] A "database" refers to a system for systematically managing stored data and retrieving data as needed.

[1292] The present invention provides a system for generating personalized advertisements based on user interests and improving advertising effectiveness. This system operates mainly between a server and a user terminal, and is realized using the following specific hardware and software.

[1293] Data collection and analysis

[1294] server:

[1295] When a user accesses a website, the server obtains a user identifier using cookies. It also collects behavioral data such as the user's browsing history, click history, and search keywords in real time. This data is initially stored in memory and periodically written to a database. Data analysis is performed using machine learning libraries such as Python's Scikit-learn library and TensorFlow. Based on the analysis results, the user's interest categories are identified and added to the user profile.

[1296] Ad generation

[1297] server:

[1298] Based on the identified user interest information, the server requests the generation AI to generate the text and images of the advertisement. The generation AI model uses a natural language processing algorithm (e.g., GPT-3) and an image generation algorithm (e.g., image generation AI). The server sends the following prompt to the generation AI:

[1299] Generate ad text with a "Fitness Equipment" theme. Include a short catchphrase to grab users' attention.

[1300] Example: Special deals on the latest fitness equipment!

[1301] The generated advertisement text and images are recorded in an advertisement content database.

[1302] Ad serving

[1303] server:

[1304] When a user visits a web page, the server retrieves the user's identifier using a cookie. The server reads the user's interests from the user profile database and retrieves the relevant ad creative from the ad content database. The retrieved ad creative is dynamically embedded into an HTML template and sent to the user's browser.

[1305] Measurement and optimization

[1306] server:

[1307] The server monitors user responses to the delivered ads and records click information and ad viewing time in real time. The recorded data is used to evaluate and optimize the generative AI model. The server periodically uses this data to retrain the generative AI model and improve the accuracy of ad generation.

[1308] Specific examples

[1309] If User A reads many articles about pet supplies, that behavioral data is collected in real time. Based on this data, the server identifies User A's interest category as "pets" and requests the generation AI to generate an ad. For example, a text ad saying "Specially Selected Pet Food Discount Sale!" and an image of cute pet food are generated. These ad creatives are incorporated in real time into the web pages visited by User A, creating ads that catch User A's interest. When an ad is clicked, the server records the click information and uses it as optimization data for the generation AI model. The next time the user visits, a more accurate ad is delivered, improving the effectiveness of the ad.

[1310] As described above, this system generates and delivers personalized advertisements based on user interests, maximizing advertising effectiveness and improving user satisfaction.

[1311] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1312] Step 1: Data collection

[1313] When a user accesses a website, the server obtains a user identifier using a cookie. Specifically, the server reads the cookie information included in the HTTP request and extracts the user identifier. The server then collects behavioral data such as the user's browsing history, click history, and search keywords. This data is temporarily stored in the server's memory and eventually written to a database.

[1314] Input: User access information (cookies, browsing history, click history, search keywords)

[1315] Output: Behavioral data record (database)

[1316] Step 2: Data analysis

[1317] The server periodically analyzes the collected behavioral data in batches. The server uses machine learning algorithms, such as Python's Scikit-learn library and TensorFlow, to identify user interest categories. This analysis allows the server to add interest categories based on the content the user frequently accesses.

[1318] Input: Behavioral data (database)

[1319] Output: Interest categories (user profile database)

[1320] Step 3: Request Ad Generation

[1321] The server sends a prompt to the AI ​​based on the identified interests. For example, if the user is interested in "health and fitness," the server sends the following prompt to the AI:

[1322] Generate ad text for "Fitness Equipment." Include a short catchphrase to grab users' attention.

[1323] Example: Special deals on the latest fitness equipment!

[1324] The AI ​​generator generates ad text based on this prompt, while the image generator algorithm simultaneously generates related images.

[1325] Input: Interest information (user profile database)

[1326] Output: Ad text and images (ad content database)

[1327] Step 4: Save your ad

[1328] The generated ad text and images are stored in an ad content database by the server, which does this by performing write operations to the database. Each ad is given a unique identifier and linked to user interest information.

[1329] Input: Ad text and image

[1330] Output: Record of advertising content (advertising content database)

[1331] Step 5: Serving Ads

[1332] Each time the user visits a new web page, the server retrieves the user identifier from the cookie and looks up the user's interests in a user profile database. The server then retrieves relevant advertisements from an advertising content database and dynamically inserts them into an HTML template that is sent to the browser, where the advertisements are displayed on the user's web page.

[1333] Input: User access information (cookies), interest information (user profile database), advertising content (advertising content database)

[1334] Output: A web page with dynamically generated ads

[1335] Step 6: Measure and optimize

[1336] The server monitors user responses to the delivered ads. Specifically, it records the ad click-through rate and duration in real time. This data is stored in a database and used to evaluate and optimize the generative AI model. The server uses this data to periodically retrain the generative AI model and improve the accuracy of ad generation.

[1337] Input: Ad response data (click-through rate, duration)

[1338] Output: Feedback data (database), optimized generative AI model

[1339] This is the flow of processing in the program for this system, which enables the generation and delivery of highly accurate personalized advertisements based on user interests.

[1340] (Application example 1)

[1341] 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."

[1342] Conventional ad delivery systems have difficulty generating personalized ads that accurately capture user interests, resulting in reduced advertising effectiveness. Additionally, it is difficult to measure the effectiveness of generated ads or optimize the AI ​​model in real time, making it difficult to maximize the effectiveness of ads.

[1343] 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.

[1344] In this invention, the server includes means for collecting behavioral data such as user browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify user interests, means for using a generative AI to generate advertising text and images based on the interest information, means for delivering the generated advertisement to the user's display device, means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generative AI model, and means for optimizing the generative AI model based on user response data. This makes it possible to generate and deliver advertisements individually personalized for each user and optimize the effectiveness of the advertisements in real time.

[1345] "User browsing history" refers to the history of web pages a user has accessed on the Internet.

[1346] "Click history" refers to the history of which links and ads a user clicked on a website.

[1347] "Search Keywords" refers to words or phrases entered by a user using a search engine.

[1348] "Behavioral data" is a collective term for data related to a series of operations or activities a user performs on the Internet.

[1349] "Generative AI" refers to systems or models that use artificial intelligence techniques to generate content for specific purposes.

[1350] "Advertising text" refers to the text and copy used in an advertisement.

[1351] "Image" means any illustration, photograph or graphic visually displayed in an Advertisement.

[1352] "User display device" refers to the device a user uses to view advertisements or web pages, including smartphones, tablets, and PCs.

[1353] "Delivery" refers to the process of displaying advertising content on a designated user's device.

[1354] "Measuring effectiveness" refers to analyzing users' responses and behavior to the ads delivered and evaluating their performance.

[1355] "Feedback Data" refers to user response data collected to evaluate the performance of an advertisement.

[1356] "User Response Data" means data regarding user clicks, view times, and other interactions with advertisements.

[1357] A "generative AI model" refers to a concrete implementation of an algorithm or system for generating advertising content using artificial intelligence technology.

[1358] "Optimization" refers to using collected data to adjust and refine generative AI models to improve their performance.

[1359] System Overview

[1360] This invention is a system that uses generative AI to generate personalized advertisements based on user interests and improve advertising effectiveness. The system operates between a server and user terminals, collecting and analyzing user behavior data to generate and deliver appropriate advertisements. Specifically, it includes the following means and processes:

[1361] Data collection and analysis

[1362] The server obtains a user identifier using cookies when a user visits a website. It also collects behavioral data, such as browsing history, click history, and search keywords. This behavioral data is stored in a database, and the server periodically analyzes the data in batches to identify user interests using machine learning algorithms and natural language processing techniques. For example, if a user frequently reads articles related to "health and fitness," this interest category will be added to the user profile.

[1363] Ad generation

[1364] The server requests ad text and images from the generation AI based on the identified interest information. For example, the generative AI model generates ad text such as "Special Offer on the Latest Fitness Equipment!", and the image generation AI generates images of fitness equipment. The generated ad text and images are stored in the ad content database.

[1365] Ad serving

[1366] When a user visits a web page, the server retrieves the user identifier from the cookie. The server reads the user's interest information from the user profile database and retrieves relevant advertising content from the advertising content database. The retrieved advertising content is dynamically embedded into an HTML template and sent to the user's browser. This allows the user to see personalized ads in real time.

[1367] Measurement and optimization

[1368] The server monitors user responses to the delivered ads (e.g., click-through rate and dwell time) and records the data in real time. This collected data is used to evaluate and optimize the generative AI model. By evaluating the model's performance and re-training or tuning as necessary, the effectiveness of the ads can be further improved.

[1369] Specific examples of technology

[1370] For example, if a user reads many articles about "pet supplies," that behavioral data is collected and analyzed. The server identifies the user's interest category as "pets" and requests the generation AI to generate an ad. The generation AI model generates ad text such as "Specialty pet food discount sale!", and the image generation AI generates images of cute pet food. These ad creatives are embedded in real time into the web pages the user visits. When an ad is clicked, the server collects click information and uses it as effectiveness measurement data. The generation AI model is optimized based on this data and reflected in future ad generation.

[1371] Prompt Sentence Examples

[1372] 1. "Create an ad for the latest pet products."

[1373] 2. "Pet supplies images"

[1374] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1375] Step 1:

[1376] Collects behavioral data such as user browsing history, click history, and search keywords.

[1377] Specific behavior:

[1378] Cookies are used in the user's browser to obtain a user identifier. Data on the user's browsing history when visiting websites, the links and ads clicked, and search keywords entered into search engines are collected. This behavioral data is stored in a database on the server.

[1379] Input: User behavior data on the website

[1380] Data processing: Use of cookies to identify user identifiers and collect behavioral data

[1381] Output: Behavioral data is saved to a database

[1382] Step 2:

[1383] Analyze collected behavioral data to identify user interests.

[1384] Specific behavior:

[1385] The server periodically extracts batches of behavioral data from the database and analyzes them using machine learning algorithms and natural language processing techniques. It extracts categories that users frequently access and adds interest categories, such as "health and fitness," to the profile database.

[1386] Input: Behavioral data stored in a database

[1387] Data processing: Analyzing data using machine learning algorithms and natural language processing techniques

[1388] Output: User interest categories are identified and stored in a profile database

[1389] Step 3:

[1390] Uses generative AI to generate ad text and images based on interest information.

[1391] Specific behavior:

[1392] The server sends the specified interest information as an input prompt to a generation AI (e.g., GPT-3 and DALL-E). The generation AI generates ad text such as "Special Offer on the Latest Fitness Equipment!" and a matching image. The generated ad content is stored in an ad content database.

[1393] Input: Interest information

[1394] Data processing: Send prompts to the generative AI to generate text and images

[1395] Output: The generated ad text and images are stored in the ad content database.

[1396] Step 4:

[1397] The generated advertisement is delivered to the user's display device.

[1398] Specific behavior:

[1399] When a user visits a web page, the server retrieves the user identifier from the cookie, reads interest information from the profile database, retrieves relevant advertising content from the advertising content database, dynamically embeds the retrieved advertisement into an HTML template, and sends it to the user's browser.

[1400] Input: User identifier, advertising content

[1401] Data processing: Dynamically embedding advertising content into HTML templates

[1402] Output: A personalized ad is displayed in the user's browser

[1403] Step 5:

[1404] Measure the effectiveness of delivered ads and collect feedback data to optimize generative AI models.

[1405] Specific behavior:

[1406] The server monitors user responses to the delivered ads, such as click-through rates and browser dwell times. The collected data is stored as feedback data and used to evaluate and optimize the generative AI model. The model's performance is analyzed and retrained or adjusted as necessary.

[1407] Input: User response data

[1408] Data processing: Collect and analyze reaction data, and retrain and adjust the model

[1409] Output: An optimized generative AI model

[1410] 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.

[1411] System Overview

[1412] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral and emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[1413] Program processing flow and operation

[1414] Data collection and analysis

[1415] server:

[1416] 1. The server obtains a user identifier using cookies each time a user visits a website, and records the user's browsing history, click history, search keywords, and other behavioral data in real time. This data is then stored in a database.

[1417] 2. The server reads the user's facial expressions through the webcam on the web page the user is viewing, and analyzes the emotional data using the emotion engine. This emotional data is also stored in the database.

[1418] Ad generation

[1419] server:

[1420] 1. Periodically collected behavioral and emotional data is batch processed and analyzed using machine learning algorithms and natural language processing techniques to identify the user's interest categories and emotional state.

[1421] 2. Based on the identified interests and real-time emotional state, the system sends a request to a generative AI to generate ad text and images. For example, GPT-3 is used to generate ad text such as "Specialty Pet Food Sale!", and an image-generating AI such as DALL-E is used to generate attractive images of the pet food.

[1422] 3. The emotion engine generates ads with bright colors and positive words when the user is in a positive emotional state, and conversely, generates ads with calming tones and soothing words when the user is in a negative emotional state.

[1423] Ad serving

[1424] server:

[1425] 1. When a user visits a web page, the server retrieves the user identifier from the cookie, reads the user's interests and emotional state from the user profile database, and retrieves the relevant ad creative for that user from the ad content database.

[1426] 2. The acquired ad creative is dynamically embedded into an HTML template and displayed in the user's browser.

[1427] Measurement and optimization

[1428] server:

[1429] 1. The server monitors user responses to delivered ads (click-through rate, length of stay, etc.) in real time and records the data.

[1430] 2. Based on the recorded data, evaluate the performance of the generative AI model and emotion engine, and retrain or tune it as needed to further improve the effectiveness of advertising.

[1431] Specific examples

[1432] For example, suppose User B reads many articles about pet supplies. While User B is browsing a web page, the emotion engine detects that the user is relaxed from their facial expression. Based on this data, the server generates ad text such as "Specialty Pet Food Discount Sale!" along with an image of pet food. The generated ad is created using bright colors and positive wording to match User B's relaxed state. This ad is then embedded in real time into the web pages visited by User B, attracting User B's interest. When the ad is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future ad generation.

[1433] As described above, the present invention is a system that generates and delivers personalized advertisements that accurately capture the user's interests and real-time emotional state, thereby maximizing advertising effectiveness.

[1434] The processing flow will be explained below.

[1435] Step 1:

[1436] server:

[1437] When a user visits a website, the server uses cookies to obtain a user identifier, and also obtains behavioral data such as the user's browsing history, click history, and search keywords in real time, which are then stored in a database.

[1438] Step 2:

[1439] User device:

[1440] While a user is browsing a website, their facial expressions are captured by a webcam, and this image data is instantly sent to a server.

[1441] Step 3:

[1442] server:

[1443] Using the transmitted image data, the emotion engine analyzes the user's facial expressions to identify their real-time emotional state (e.g., happy, sad, surprised), which is also recorded in a database.

[1444] Step 4:

[1445] server:

[1446] It periodically analyzes behavioral and emotional data through batch processing, and uses machine learning algorithms and natural language processing techniques to identify users' interest categories and emotional states, thereby revealing which categories users are interested in.

[1447] Step 5:

[1448] server:

[1449] Based on the identified interests and real-time emotional state, the system requests a generative AI (e.g., GPT-3, DALL-E) to generate ad text and images. For example, if the emotion engine determines that the user is relaxed, the generative AI will generate ad text that matches the relaxed state, such as "Specialty pet food discount sale!". DALL-E will then generate a bright, relaxing image of the pet food.

[1450] Step 6:

[1451] server:

[1452] The generated advertisement text and images are stored in an advertisement content database.

[1453] Step 7:

[1454] server:

[1455] Each time a user visits a web page, the system retrieves the user identifier from the cookie, reads their interests and emotional state from the user profile database, and retrieves ad creative from the ad content database based on this information.

[1456] Step 8:

[1457] server:

[1458] The resulting ad creative is dynamically embedded into an HTML template and displayed in the user's browser, with the tone and color of the ad adjusted based on the user's emotional state.

[1459] Step 9:

[1460] server:

[1461] It monitors user responses to delivered ads (whether the ad was clicked or skipped, etc.) in real time and records the data.

[1462] Step 10:

[1463] server:

[1464] Based on the collected effectiveness measurement data, the performance of the generative AI model and emotion engine is evaluated, and if necessary, retraining and tuning are carried out to further improve the effectiveness of advertising.

[1465] Step 11:

[1466] User device:

[1467] When an ad is displayed, the user's browser sends actions such as clicks or skips to the server, and this feedback data is used in the measurement and optimization process.

[1468] This series of steps allows for the generation and delivery of personalized ads that are optimized for the user's interests and emotional state, maximizing advertising effectiveness.

[1469] Example 2

[1470] 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."

[1471] Conventional personalized advertising systems generate ads based solely on user behavioral data and lack the ability to generate ads that take into account the user's real-time emotional state. This makes it difficult to optimize ads based on the user's emotional state and maximize advertising effectiveness. Furthermore, the AI ​​model for generating ads is not adequately optimized based on the effectiveness measurement data of the generated ads, making it difficult to expect continuous improvement in advertising effectiveness.

[1472] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords, means for analyzing the collected behavioral data to identify the user's interests, means for analyzing the collected facial expression data to identify the user's emotional state, means for using a generation AI to generate advertisement text and images based on the interest information and emotional state, means for delivering the generated advertisement to the user's display terminal, and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating the user's behavioral data and emotional data.

[1473] "User browsing history" refers to historical data about the pages a user visits on a website.

[1474] "Click history" is historical data about the links and buttons a user clicks on a website.

[1475] "Search Keywords" are search terms entered by a user into a search engine.

[1476] "Behavioral data" refers to data related to a user's behavior on a website, such as their browsing history, click history, and search keywords.

[1477] "Facial expression data" is data collected based on a user's facial expressions.

[1478] "Emotional state" is information indicating the emotional state of the user, obtained by analyzing the user's facial expression data.

[1479] "Interest Information" is information about a user's interests and categories of interest that are identified through analysis of collected behavioral data.

[1480] "Generative AI" is an artificial intelligence technology that generates text and images based on a given prompt.

[1481] The "user's display terminal" refers to a display device such as a computer or smartphone used by the user.

[1482] "Advertising text" is the text displayed as an advertising message.

[1483] "Advertising image" is image data that is displayed as an advertising message.

[1484] "Feedback data" is data that records user responses to delivered advertisements and is used to optimize generative AI models.

[1485] "Delivery" refers to displaying the generated advertisement on the user's display device.

[1486] "Optimizing generative AI models" refers to making adjustments and improvements to improve the performance of generative AI based on collected feedback data.

[1487] MODE FOR CARRYING OUT THE INVENTION

[1488] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates primarily between a server and a user's device and is designed to provide a rich, personalized advertising experience.

[1489] Data collection and analysis

[1490] server:

[1491] 1. The server obtains a user identifier using a cookie each time a user visits the website. The user identifier is generated as a unique ID and stored in a database.

[1492] 2. The server then records users' browsing history, click history, search keywords, and other behavioral data in real time, which is then stored in a database such as MongoDB or MySQL along with a timestamp.

[1493] 3. The server reads the user's facial expressions through the camera installed on the user's device and analyzes the emotional data using an emotion engine (for example, Microsoft Azure's Emotion API). The resulting emotion score (for example, joy, sadness, surprise, etc.) is also stored in a database.

[1494] Ad generation

[1495] server:

[1496] 1. The server processes the collected behavioral and emotional data in batches, analyzes the data using a batch processing framework such as Apache Spark, and applies a machine learning model (e.g., a clustering algorithm using scikit-learn) to classify the user's interest categories.

[1497] 2. Send a request to a generative AI (e.g., GPT-3 and DALL-E using OpenAI APIs) to generate ad text and images based on the identified interests and real-time emotional state. Create a prompt like the following: "Generate ad copy with a light tone that corresponds to a relaxed emotional state for users interested in pet products."

[1498] 3. The emotion engine is used to appropriately adjust the tone of the generated ad: if the emotional state is positive, the ad content will be generated with bright images and positive language.

[1499] Ad serving

[1500] server:

[1501] 1. Every time a user visits a web page, the cookie information is analyzed to obtain the user identifier again, and an SQL query is issued to obtain the necessary interest information and emotional state from the user profile database.

[1502] 2. The retrieved ad creative is dynamically embedded into an HTML template and displayed in the user's browser using a web server framework such as Node.js. A template engine (e.g., EJS or Handlebars) is used to embed ad data into HTML.

[1503] Measurement and optimization

[1504] server:

[1505] 1. User responses to delivered ads (click-through rate, duration, etc.) are monitored in real time and the data is recorded. This is done using the Google Analytics API, which collects user behavior data and stores it in a database.

[1506] 2. Evaluate the performance of the generative AI model and emotion engine based on the recorded data. Use Python scripts to calculate the precision and recall of the model, and retrain or tune it as needed to improve the overall advertising effectiveness of the system.

[1507] Specific examples

[1508] For example, if User B reads many articles about pet supplies, the server will use the emotion engine to detect from User B's facial expression that he or she is relaxed while browsing the webpage. Based on this data, the server generates advertising text such as "Specialty Pet Food Discount Sale!" and an image of pet food. The generated advertisement is created using bright colors and positive wording to match User B's relaxed state. This advertisement is then embedded in real time into the webpages visited by User B, attracting User B's interest. When the advertisement is clicked, the server collects click information and uses it as feedback data. Based on this data, the generative AI model and emotion engine are optimized and reflected in future advertisement generation.

[1509] This system makes it possible to generate and deliver personalized advertisements that are optimized in real time by integrating user behavioral and emotional data.

[1510] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1511] Program processing flow

[1512] Step 1:

[1513] The server uses cookies to obtain a user identifier each time a user visits a website. It analyzes the information in the cookie (input) and extracts the user identifier (output). It stores this identifier in a database. Specifically, when an HTTP request is received, it reads the value of the HTTP cookie and records it in the database.

[1514] Step 2:

[1515] The server records user behavioral data such as browsing history, click history, and search keywords in real time. First, it monitors user behavior (input), activates an event logger accordingly, and saves the behavioral data (output) in a database. Specifically, it captures user events that occur within a web page (such as clicks and page transitions) and stores that information in MongoDB or MySQL.

[1516] Step 3:

[1517] The server reads the user's facial expression through the camera installed on the user's device and analyzes the emotional data using the emotion engine. The server obtains the user's facial expression image (input), analyzes it, and generates an emotion score (output). Specifically, it sends the image data to the Emotion API in real time and stores the results in a database.

[1518] Step 4:

[1519] The server processes the collected behavioral and emotional data in batches and analyzes the data. The behavioral and emotional data (input) is processed using a batch processing framework such as Apache Spark to identify the user's interest categories (output). Specifically, the server periodically runs a batch job and clusters the data using a machine learning model.

[1520] Step 5:

[1521] The server sends a request to the generation AI to generate ad text and images based on the identified interest information and real-time emotional state. The interest information and emotional state (input) are processed, and ad creative (output) is obtained by sending a prompt to the generation AI. Specifically, the server sends a prompt to the generation AI, such as "Please generate ad copy with a bright tone that corresponds to a relaxed emotional state, targeted at users interested in pet products."

[1522] Step 6:

[1523] The server uses the emotion engine to appropriately adjust the tone of the generated advertisement. It receives the ad creative (input) and modifies the tone depending on the emotional state. Specifically, if the emotional state is positive, it generates an advertisement with bright images and positive wording.

[1524] Step 7:

[1525] The server analyzes the cookie information again each time the user visits a web page to obtain the user identifier. It receives the HTTP request (input) when the web page is visited, analyzes the cookie information, and extracts the user identifier (output). Specifically, it reads the cookie value included in the HTTP request and uses that information to retrieve it from the user profile database.

[1526] Step 8:

[1527] The server dynamically embeds the acquired ad creative into an HTML template and displays it in the user's browser. By embedding the ad creative (input) into the HTML template, an ad-embedded page (output) is generated. Specifically, it uses a template engine (EJS or Handlebars) to insert ad data into the HTML code and sends the generated HTML to the client.

[1528] Step 9:

[1529] The server monitors user responses to delivered ads (click-through rate, duration, etc.) in real time and records the data. It analyzes user behavior data (input) and generates response data (output). Specifically, it uses the Google Analytics API to monitor user activity and stores that data in a database.

[1530] Step 10:

[1531] The server evaluates the performance of the generative AI model and emotion engine based on the recorded data. It analyzes the reaction data (input) and obtains the model evaluation results (output). Specifically, it uses a Python script to calculate the precision and recall of the model and retrain or tune the generative AI model and emotion engine as needed.

[1532] (Application example 2)

[1533] 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."

[1534] Conventional advertising systems typically generate personalized ads based on user behavior data, but they are unable to reflect the user's real-time emotional state. As a result, ads that do not match the user's emotions are sometimes displayed, making effective ad delivery difficult. Furthermore, there was no established method for collecting real-time user emotional data using mobile devices such as smart glasses and reflecting that data in ad generation. Therefore, a new system was needed to maximize the user experience and improve advertising effectiveness.

[1535] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting behavioral data such as a user's browsing history, click history, and search keywords; means for analyzing the collected behavioral data to identify the user's interests; means for using a generation AI to generate advertising text and images based on the interest information and real-time emotional state; means for reading the user's facial expressions from a mobile device such as smart glasses and analyzing the emotional data; means for delivering the generated advertisement to the user's display device; and means for measuring the effectiveness of the delivered advertisement and collecting feedback data for optimizing the generation AI model. This makes it possible to combine the user's behavioral data and real-time emotional data to generate and deliver personalized advertisements that are best suited to the user's situation.

[1536] "User browsing history" refers to historical data about the websites and pages a user visits on the Internet.

[1537] "Click history" is historical data about the links and buttons a user clicks on the Internet.

[1538] "Search keywords" are historical data of words and phrases entered by users into search engines, etc.

[1539] "Behavioral data" is data that records a user's activities on the Internet, such as their browsing history, click history, and search keywords.

[1540] "Collection methods" are the technical methods and systems used to collect user behavior data.

[1541] "Means for analyzing and identifying user interests" refers to technical methods or systems that analyze collected behavioral data to identify areas or products in which users are interested.

[1542] "Generative AI" is an artificial intelligence technology that uses machine learning models to generate new text or images based on specified conditions.

[1543] "Real-time emotional state" refers to data that instantly analyzes the user's current emotional state based on facial expressions, voice, etc.

[1544] "Smart glasses" are wearable devices equipped with cameras and sensors that can detect the user's gaze and facial expressions.

[1545] "Means for reading facial expressions and analyzing emotional data" refers to technical methods or systems that use a device to capture a user's facial expressions and use that data to analyze the user's emotions.

[1546] "Display terminal" refers to a device on which a generated advertisement is displayed, such as a smartphone or smart glasses.

[1547] "Delivery means" refers to the technical method or system by which the generated advertisement is transmitted to the user's display device via the Internet.

[1548] "Feedback data" refers to data that is collected from users' reactions and behavior to delivered advertisements and used to optimize the system.

[1549] "Means for measuring effectiveness and optimizing generative AI models" refers to technical methods and systems that analyze the performance of delivered advertisements and improve the performance of generative AI models based on the collected data.

[1550] System Overview

[1551] This invention is a system that uses generative AI to generate and deliver personalized advertisements based on user behavioral and emotional data. This system operates between a server and user devices, collecting and analyzing user behavioral data and real-time emotional data to generate and deliver appropriate advertisements. In addition, the effectiveness of the generated advertisements is collected as feedback, and the system is optimized.

[1552] Hardware and Software Configuration

[1553] The system mainly uses the following hardware and software:

[1554] Server: Hosts and processes databases, analytics engines, and ad generation AI.

[1555] Smart glasses: Captures the user's facial expressions and gaze and transmits the emotional data to a server.

[1556] Emotion Recognition Model: Recognizing emotions from facial expressions using Keras.

[1557] Face detector: Detects the user's face using dlib.

[1558] Generative AI: Generate ad text and images using GPT-3 and DALL-E.

[1559] User device: the device on which the generated advertisement is displayed (e.g. smartphone, PC, smart glasses).

[1560] Data collection and analysis

[1561] The servers collect the websites and pages you visit on the Internet (browsing history), the links and buttons you click (click history), and the words and phrases you enter into search engines (search keywords). This behavioral data is analyzed to identify your interests.

[1562] The smart glasses use built-in cameras and sensors to capture the user's facial expressions and send the data to a server, which then uses an emotion recognition model to analyze the user's emotional state in real time.

[1563] Ad generation

[1564] The server uses generative AI to generate personalized ad text and images based on the collected behavioral and emotional data. For example, if a user is interested in pet supplies and has a relaxed expression, the server generates ad text saying "Specialty pet food discount sale!" and an attractive image of pet food.

[1565] Ad serving

[1566] The generated ads are then delivered to the user's display device, such as the smart glasses display or smartphone screen, and personalized ads are presented in real time based on the user's gaze and emotional state.

[1567] Measurement and optimization

[1568] The server monitors user responses to the delivered ads and collects the data. The collected data is used to evaluate the performance of the generative AI model, and retraining and tuning are performed as necessary. This allows for optimization to improve the effectiveness of the ads.

[1569] Specific examples

[1570] For example, when a user visits a pet shop, the smart glasses identify the user's gaze. If the user shows interest in pet food and has a relaxed expression, the server uses generative AI to generate advertising text such as "Specialty Pet Food Discount Sale!" along with an image of attractive pet food. This advertisement is displayed in real time on the smart glasses' display, hoping to attract the user's attention.

[1571] Example prompt for a generative AI model:

[1572] User behavior data:

[1573] User ID: user123

[1574] Interests: Pet supplies

[1575] Recent activity: Reading a lot of articles about pet supplies

[1576] Real-time sentiment data:

[1577] Emotion: Relaxed

[1578] Generated ad text:

[1579] Text: Special discount sale on select pet foods!

[1580] Color: Bright

[1581] In this way, ads are dynamically generated that are tailored to the user's circumstances and interests, providing an optimal advertising experience.

[1582] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1583] Step 1:

[1584] When a user visits a website on the Internet, the server obtains a user identifier using cookies. The input is the website the user visits and their activities within it, and the output is the user identifier and behavioral data. This allows the user's browsing history, click history, search keywords, etc. to be collected and stored in a database.

[1585] Step 2:

[1586] When a user wears the smart glasses, the camera and sensors in the smart glasses capture the user's facial expressions. The input is the user's real-time facial expression data, and the output is the captured facial image, which is sent to the server.

[1587] Step 3:

[1588] The server runs an emotion recognition model using the received facial expression data. The input is the captured facial expression image, and the output is data indicating the user's emotional state. The emotion recognition model (Keras model) analyzes this facial expression data and identifies the user's real-time emotional state.

[1589] Step 4:

[1590] The server sends an ad generation request to the generative AI based on the collected behavioral and emotional data. The input is user interest data and real-time emotional data, and the output is ad text and images. The generative AI (GPT-3 and DALL-E) generates personalized ads based on the user's interests and emotions.

[1591] Step 5:

[1592] The generated advertisement is dynamically embedded into an HTML template by the server. The input is the advertisement text and images, and the output is an advertisement creative appropriate for the user's display device. This allows the advertisement to be displayed where the user sees it through smart glasses or a smartphone.

[1593] Step 6:

[1594] The server monitors how users respond to the displayed ads. The input is user reaction data to the ads (number of clicks, time spent, etc.), and the output is effectiveness measurement data. This allows the performance of the ads to be evaluated in real time.

[1595] Step 7:

[1596] The server retrains or tunes the generative AI model based on the collected effectiveness measurement data. The input is the advertising effectiveness measurement data, and the output is an optimized generative AI model. This allows for optimization that can be expected to produce even greater effectiveness in the next ad generation.

[1597] 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.

[1598] 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.

[1599] 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.

[1600] 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.

[1601] 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.

[1602] 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.

[1603] 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).

[1604] 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.

[1605] 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."

[1606] 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.

[1607] 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).

[1608] 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.

[1609] 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.

[1610] 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.

[1611] 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.

[1612] 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.

[1613] 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.

[1614] 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.

[1615] 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.

[1616] 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.

[1617] 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.

[1618] The following is further disclosed regarding the above embodiment.

[1619] (Claim 1)

[1620] A means of collecting user behavioral data such as browsing history, click history, and search keywords,

[1621] A means of analyzing collected behavioral data to identify user interests;

[1622] a means for using generative AI to generate text and images for advertisements based on the interest information;

[1623] means for delivering the generated advertisement to a user's display device;

[1624] A means of measuring the effectiveness of delivered ads and collecting feedback data to optimize generative AI models;

[1625] A system including:

[1626] (Claim 2)

[1627] 10. The system of claim 1, wherein the generated advertisement text and images are dynamically embedded into web pages accessed by the user.

[1628] (Claim 3)

[1629] The system of claim 1, further comprising means for learning and tuning the generation AI model based on the effectiveness measurement data of the generated advertisement.

[1630] "Example 1"

[1631] (Claim 1)

[1632] A means of collecting user behavioral data such as browsing history, click history, and search keywords,

[1633] A means of analyzing collected behavioral data to identify user interests;

[1634] a means for using generative AI to generate text and images for advertisements based on the interest information;

[1635] means for recording the generated advertisement text and images in a storage database;

[1636] A means for acquiring a user identifier each time a user visits a web page and delivering advertisements based on the user's interest information;

[1637] A means of measuring the effectiveness of delivered ads and collecting feedback data to optimize generative AI models;

[1638] A system including:

[1639] (Claim 2)

[1640] 10. The system of claim 1, wherein the generated advertisement text and images are dynamically embedded into web pages accessed by the user.

[1641] (Claim 3)

[1642] The system of claim 1, further comprising means for learning and tuning the generation AI model based on the effectiveness measurement data of the generated advertisement.

[1643] "Application Example 1"

[1644] (Claim 1)

[1645] A means of collecting user behavioral data such as browsing history, click history, and search keywords,

[1646] A means of analyzing collected behavioral data to identify user interests;

[1647] a means for using generative AI to generate text and images for advertisements based on the interest information;

[1648] means for delivering the generated advertisement to a user's display device;

[1649] A means of measuring the effectiveness of delivered ads and collecting feedback data to optimize generative AI models;

[1650] A means of optimizing the generative AI model based on user response data; and

[1651] A system including:

[1652] (Claim 2)

[1653] 10. The system of claim 1, wherein the generated advertisement text and images are dynamically embedded into web pages accessed by the user.

[1654] (Claim 3)

[1655] The system of claim 1, further comprising means for learning and tuning the generation AI model based on the effectiveness measurement data of the generated advertisement.

[1656] "Example 2: Combining Emotion Engines"

[1657] (Claim 1)

[1658] A means of collecting user behavioral data such as browsing history, click history, and search keywords,

[1659] A means of analyzing collected behavioral data to identify user interests;

[1660] a means for analyzing the collected facial expression data to identify the user's emotional state;

[1661] a means for using generative AI to generate text and images for advertisements based on interest information and emotional state;

[1662] means for delivering the generated advertisement to a user's display device;

[1663] A means of measuring the effectiveness of delivered ads and collecting feedback data to optimize generative AI models;

[1664] A system including:

[1665] (Claim 2)

[1666] 10. The system of claim 1, wherein the generated advertisement text and images are dynamically embedded into web pages accessed by the user.

[1667] (Claim 3)

[1668] The system of claim 1, further comprising means for learning and tuning the generation AI model based on the effectiveness measurement data of the generated advertisement.

[1669] "Application example 2 when combining emotion engines"

[1670] (Claim 1)

[1671] A means of collecting user behavioral data such as browsing history, click history, and search keywords,

[1672] A means of analyzing collected behavioral data to identify user interests;

[1673] a means for using generative AI to generate text and images for advertisements based on interest information and real-time emotional state;

[1674] A means of reading the user's facial expressions from a mobile device such as smart glasses and analyzing emotional data;

[1675] means for delivering the generated advertisement to a user's display device;

[1676] A means of measuring the effectiveness of delivered ads and collecting feedback data to optimize generative AI models;

[1677] A system including:

[1678] (Claim 2)

[1679] 10. The system of claim 1, wherein advertising text and images generated using data obtained from the smart glasses' cameras and sensors are dynamically embedded in a display area viewed by a user.

[1680] (Claim 3)

[1681] The system of claim 1, further comprising means for learning and tuning the generation AI model based on the effectiveness measurement data of the generated advertisement. [Explanation of symbols]

[1682] 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 means of collecting user behavioral data such as browsing history, click history, and search keywords, A means of analyzing collected behavioral data to identify user interests; a means for using generative AI to generate text and images for advertisements based on the interest information; means for delivering the generated advertisement to a user's display device; A means of measuring the effectiveness of delivered ads and collecting feedback data to optimize generative AI models; A system including:

2. 10. The system of claim 1, wherein the generated advertisement text and images are dynamically embedded into web pages accessed by users.

3. The system according to claim 1, further comprising means for learning and tuning the generation AI model based on the effectiveness measurement data of the generated advertisement.

Images