System

Abstract

A system is provided.SOLUTION: A system, comprising: means for collecting advertisement information using a generative AI model; means for selecting and initializing the generative AI model based on the collected advertisement information; means for automatically generating a plurality of advertisement patterns using the generative AI model; means for providing the generated advertisement patterns to users and receiving reviews and feedback; and means for making final adjustments and outputs of the advertisement patterns based on user feedback.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] Recently, there has been an increase in cases where a company's image has been damaged due to scandals involving celebrities, which has increased the risk of commercial production. Furthermore, as commercial production costs and production time have increased, companies are seeking efficient, low-risk methods of commercial production. This invention aims to reduce costs, shorten production time, and eliminate talent risk in corporate commercial production by automatically generating commercials using a generative AI model. [Means for solving the problem]

[0005] To solve this problem, the present invention provides the following means: A generative AI model is used to collect advertising data as input data, and the generative AI model is selected and initialized based on the collected advertising data. Furthermore, the generative AI model is used to automatically generate multiple advertising patterns, and the generated advertising patterns are provided to users for review and feedback. Final adjustments and output of the advertising patterns are made based on user feedback, allowing for the efficient creation of high-quality commercials. This provides a system that allows companies to reduce production costs and shorten production times and avoid talent risks.

[0006] A "generative AI model" is an artificial intelligence model used to automatically generate advertisements and visual elements based on collected data.

[0007] "Advertising data" refers to information necessary for generating advertisements, such as product information provided by companies, target demographics, appealing points, and past advertising examples.

[0008] "Input data" refers to the data supplied to the generative AI model, and includes advertising data and information based on requests from companies to create commercials.

[0009] "Collection" refers to the gathering of information provided by companies or obtained from databases.

[0010] "Initialization" refers to setting up a generative AI model, loading the necessary training data, and getting it up and running.

[0011] "Advertising patterns" refer to multiple variations of advertising ideas or commercials automatically generated by a generative AI model.

[0012] "Review" refers to the process by which a user reviews and evaluates the generated advertising patterns.

[0013] "Feedback" refers to improvements or suggestions provided by users regarding advertising patterns.

[0014] "Adjustment" refers to modifying and optimizing advertising patterns based on user feedback.

[0015] "Export" means encoding the final advertising variations into high-resolution formats and optimizing and delivering them to various media.

[0016] "User" refers to a company representative or their agent who requests the creation of an advertisement.

[0017] "System" refers to a mechanism that integrates the above means and executes a series of processes that automatically generate and output advertisements. [Brief explanation of the drawings]

[0018] [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

[0019] 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.

[0020] First, the terms used in the following description will be explained.

[0021] 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).

[0022] 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.

[0023] 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.

[0024] 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.

[0025] 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."

[0026] [First embodiment]

[0027] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0028] 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.

[0029] 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).

[0030] 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.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] 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.

[0036] 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.

[0037] 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.

[0038] 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."

[0039] MODE FOR CARRYING OUT THE INVENTION

[0040] The following is a specific method for implementing a system for automatically generating advertisements using a generative AI model according to the present invention. The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting advertisements.

[0041] Data collection and preparation

[0042] 1. Data Collection

[0043] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[0044] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[0045] 2. Data Preprocessing

[0046] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model, which includes cleaning up text data, processing image data, and formatting audio files.

[0047] Preparing the generative AI model

[0048] 3. Model Selection and Initialization

[0049] The server analyzes the type and content of the provided data and selects the most appropriate generative AI model (text generation, image generation, speech generation, etc.). After selecting the model, it loads the necessary training data and initializes the model.

[0050] Ad generation

[0051] 4. Ad generation instructions

[0052] The server inputs the preprocessed data into the generative AI model, which then automatically generates multiple ad patterns, such as energetic ads, ads that emphasize technical aspects, and ads that show actual usage scenarios.

[0053] Reviews and Feedback

[0054] 5. Providing Reviews

[0055] The server generates a review interface for providing the generated plurality of advertisement patterns to a user, and the user reviews the advertisement patterns through the provided interface and inputs feedback for each advertisement.

[0056] 6. Gathering Feedback

[0057] The server collects and stores user feedback for use in the next ad generation task, providing specific suggestions for improvement, such as making the music faster or the message more concise.

[0058] Final ad adjustment and output

[0059] 7. Applying Feedback

[0060] The server selects the best ad pattern based on user feedback and makes necessary adjustments, such as changing the tempo of the voice announcement and modifying some of the text.

[0061] 8. Final encoding and output

[0062] The server encodes the tailored ad variations into high-definition formats and optimizes them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0063] Specific examples

[0064] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[0065] 1. Data Collection

[0066] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0067] 2. Data Preprocessing

[0068] The server processes the collected data and converts it into a format suitable for the generative AI model.

[0069] 3. Model Selection and Initialization

[0070] The server selects text generation and image generation models suitable for generating ads for smartwatches and loads the necessary training data.

[0071] 4. Ad generation instructions

[0072] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0073] 5. Providing Reviews

[0074] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[0075] 6. Gathering Feedback

[0076] The user requests that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster as feedback, and the server collects this.

[0077] 7. Applying Feedback

[0078] The server adjusts the "health monitoring emphasis advertisement" based on user feedback, speeding up the tempo of the voice announcements.

[0079] 8. Final encoding and output

[0080] The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[0081] Based on these steps, the system enables businesses to generate high-quality advertisements in an efficient and low-risk manner.

[0082] The processing flow will be explained below.

[0083] MODE FOR CARRYING OUT THE INVENTION

[0084] Program processing flow

[0085] Step 1:

[0086] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[0087] Step 2:

[0088] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[0089] Step 3:

[0090] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[0091] Step 4:

[0092] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[0093] Step 5:

[0094] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[0095] Step 6:

[0096] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[0097] Step 7:

[0098] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[0099] Step 8:

[0100] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[0101] Step 9:

[0102] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message shorter."

[0103] Step 10:

[0104] The server collects user feedback and adjusts the advertising pattern based on that feedback, for example speeding up the voice announcements or shortening the text.

[0105] Step 11:

[0106] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[0107] Step 12:

[0108] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[0109] In this way, generative AI can be used to generate high-quality ads in an efficient and low-risk manner.

[0110] Example 1

[0111] 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."

[0112] Ad production is a time-consuming and costly process, especially when multiple ad variations need to be generated for different target audiences. It's also challenging to quickly incorporate feedback and deliver optimal ads while maintaining ad quality, making it difficult to maximize the effectiveness of advertising campaigns.

[0113] 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.

[0114] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for analyzing and preprocessing the collected advertising data, means for selecting and initializing an optimal generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, and means for final adjustment and output of the advertising patterns based on user feedback. This reduces the time and cost of advertising production, makes it possible to quickly and effectively generate multiple advertising patterns, and provide optimal advertisements by reflecting feedback.

[0115] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to automatically generate advertising content from a variety of data, including text, images, and audio.

[0116] "Advertising data" is a general term for information necessary for generating advertisements, such as product information provided by companies, target demographics, appealing points, and past advertising examples.

[0117] "Input data" refers to advertising data that is input into the generative AI model, and is provided to the generative AI model after initial data collection and preprocessing.

[0118] "Preprocessing" refers to the process of data processing and cleaning to convert advertising data into a format suitable for generative AI models.

[0119] "Advertising patterns" is a collective term for multiple advertising formats automatically generated by a generative AI model.

[0120] "Review" is an operation in which a user checks and evaluates the generated advertisement pattern.

[0121] "Feedback" refers to opinions and suggestions for improvement that a user provides regarding the generated advertising pattern.

[0122] "Final adjustment" refers to the process of optimizing the content and format of the advertising pattern by reflecting user feedback.

[0123] "Output" refers to the process of encoding the final adjusted advertisement pattern to generate the final advertisement file.

[0124] A "promotion content creation request" is a request sent by a user to a server to request the creation of new advertising or marketing materials.

[0125] "Quality evaluation" is the process of evaluating the effectiveness and suitability of advertising patterns output by a generative AI model.

[0126] "High definition format" refers to a file format that allows the generated advertisement to be presented in high visual and audio resolution.

[0127] "Media optimization" is the process of converting the generated advertising patterns into formats suitable for different distribution media such as the Internet, television, and radio.

[0128] This invention is a system for automatically generating advertisements using a generative AI model. The main functions of the system consist of the following steps: collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting the advertisements.

[0129] Data collection and preparation

[0130] Data collection

[0131] The server collects advertising data provided by companies, such as product information, target demographics, key selling points, and past advertising examples. This data is structured and stored in an advertising database. Data is collected through APIs and database connections and converted into the format required to generate ads.

[0132] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[0133] Data Preprocessing

[0134] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model. This includes cleaning up text data, processing image data, and formatting audio files. For text data, unnecessary spaces and special characters are removed and natural language processing tools are used to segment the data. For image data, image resizing and formatting are performed, and for audio data, noise reduction and encoding are performed.

[0135] Preparing the generative AI model

[0136] Model Selection and Initialization

[0137] The server analyzes the provided data and selects the optimal generative AI model (text generation, image generation, speech generation, etc.). Specifically, it uses GPT-3 as the text generation model, DALL-E as the image generation model, and WaveNet as the speech generation model. After the selection, it loads the necessary training data and initializes the model.

[0138] Ad generation

[0139] Ad generation instructions

[0140] The server inputs the preprocessed data into the generative AI model to automatically generate multiple ad patterns, and the generated results are recorded in a log to generate multiple different ad patterns (such as energetic ads, technology-focused ads, and ads showing usage scenarios).

[0141] Reviews and Feedback

[0142] Providing a review

[0143] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface for displaying the generated advertisements in thumbnail and playback format and allowing the user to evaluate them.

[0144] Collecting feedback

[0145] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[0146] Final ad adjustment and output

[0147] Applying Feedback

[0148] The server adjusts the advertising patterns based on user feedback, specifically by readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[0149] Final encoding and output

[0150] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0151] Specific examples

[0152] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[0153] 1. Data Collection

[0154] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0155] 2. Data Preprocessing

[0156] The server preprocesses the collected data using natural language processing tools and image processing tools.

[0157] 3. Model Selection and Initialization

[0158] Based on the collected data, the server selects a generative AI model such as GPT-3, DALL-E, or WaveNet and loads the necessary training data.

[0159] 4. Ad generation instructions

[0160] The server inputs the preprocessed data into the model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0161] 5. Providing Reviews

[0162] A user uses a web interface to review and rate the generated ad variations.

[0163] 6. Gathering Feedback

[0164] The user may provide feedback requesting that the tempo of the audio for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback.

[0165] 7. Applying Feedback

[0166] The server adjusts the tempo of the audio and regenerates it based on the feedback.

[0167] 8. Final encoding and output

[0168] The server encodes the tailored advertisement into high definition format and provides the final version to the user via a secure download link.

[0169] Based on the above steps, the system enables companies to generate high-quality advertisements efficiently and with low risk.

[0170] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0171] Step 1: Data collection

[0172] The server collects data provided by companies, such as product information, target demographics, selling points, and past advertising examples. This data is retrieved through APIs and database connections, structured, and stored in an advertising database.

[0173] Input: Promotional content creation requests from companies, product information, target demographic data

[0174] Data processing: Acquired through APIs and database connections, stored in a structured database

[0175] Output: Structured data stored in the ad database

[0176] Step 2: Preprocessing the data

[0177] The server analyzes data collected from the advertising database and converts it into a format suitable for the generative AI model. Text data is stripped of unnecessary spaces and special characters and segmented using natural language processing tools. Image data is resized and formatted, and audio data is noise-reduced and encoded.

[0178] Input: Structured advertising data in an advertising database

[0179] Data processing: cleaning up text data, resizing and formatting images, noise reduction and encoding of audio data

[0180] Output: Preprocessed data (cleaned text, enhanced images, formatted audio)

[0181] Step 3: Model selection and initialization

[0182] The server selects the optimal generative AI model based on the preprocessed data. Specifically, it selects GPT-3 for text generation, DALL-E for image generation, and WaveNet for speech generation. After the selection, it loads the necessary training data and initializes the model.

[0183] Input: Preprocessed data

[0184] Data processing: Analyzing data content, selecting the optimal model, and loading training data

[0185] Output: Initialized generative AI model

[0186] Step 4: Ad generation instructions

[0187] The server inputs the preprocessed data into a generative AI model to automatically generate multiple ad patterns, which are then logged and used to generate different patterns, such as energetic ads, technology-focused ads, and ads showing usage scenarios.

[0188] Input: Initialized generative AI model and preprocessed data

[0189] Data processing: Generating advertising patterns using generative AI models

[0190] Output: Multiple ad variations generated

[0191] Step 5: Provide a review

[0192] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface is used to display the generated advertisements in thumbnail and playback format and allow the user to evaluate them.

[0193] Input: Generated ad patterns

[0194] Data processing: Generate review interface and display advertising patterns

[0195] Output: The review interface presented to the user

[0196] Step 6: Gather feedback

[0197] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[0198] Input: User feedback

[0199] Data processing: collecting and storing feedback

[0200] Output: Saved feedback data

[0201] Step 7: Applying feedback

[0202] The server adjusts the ad patterns based on user feedback, readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[0203] Input: Saved feedback data

[0204] Data processing: Readjust model parameters based on feedback and re-run ad generation

[0205] Output: Adjusted ad patterns

[0206] Step 8: Final encoding and output

[0207] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0208] Input: Adjusted ad pattern

[0209] Data processing: Encoding to high definition formats and optimizing for media

[0210] Output: Final ad file provided with a secure download link

[0211] (Application example 1)

[0212] 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."

[0213] Modern advertising production requires a significant amount of time and effort, making it difficult to generate effective ads, especially for targeted audiences. Furthermore, systems for quickly incorporating user feedback are inadequate, and improvements are needed to improve the quality of advertising. Furthermore, there is a lack of ad generation systems that utilize mobile devices such as smartphones, creating a demand for a more flexible and rapid advertising production process.

[0214] 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.

[0215] In this invention, the server includes: means for collecting advertising data as input data using a generative AI model; means for selecting and initializing a generative AI model based on the collected advertising data; means for automatically generating multiple advertising patterns using the generative AI model; means for providing the generated advertising patterns to users and receiving reviews and feedback; means for making final adjustments to and outputting the advertising patterns based on user feedback; means for providing an interface using a smartphone; means for users to input product information and send the data to the server; and means for analyzing the provided data and converting it into a format suitable for the generative AI model. This enables users to easily collect and input advertising data using a smartphone, quickly generate high-quality advertisements using an optimal generative AI model based on the data, and regenerate advertisements that reflect user feedback.

[0216] A "generative AI model" is an algorithm that uses machine learning to automatically analyze data and perform a specific task, such as generating an ad.

[0217] "Advertising data" refers to information such as product information, target demographics, and appealing points that are necessary when creating an advertisement.

[0218] A "server" is a computing system that is responsible for collecting, analyzing, generating, storing, and transmitting data.

[0219] "Input data" refers to data such as product information, target demographics, and past advertising examples provided to generate advertisements.

[0220] "Ad Variants" refers to multiple different variations of an advertisement automatically generated by a generative AI model.

[0221] "User" refers to any person or entity that utilizes the Ad Generation System to generate, review, and provide feedback on Ads.

[0222] "Review interface" refers to a user interface that allows a user to review the generated advertising patterns and provide feedback.

[0223] "Feedback" refers to improvements and evaluations provided by users regarding the generated advertising patterns.

[0224] "Smartphone" means a mobile device that allows access to the ad generation system and enables data entry, review and feedback.

[0225] "Interface" refers to the means by which a user interacts with a system.

[0226] "Data analysis" refers to the process of converting collected data into useful information and forms.

[0227] "High-resolution format" refers to a format for exporting high-quality advertisements to a variety of media formats.

[0228] "Encoding" refers to the process of converting a generated advertisement into a particular format for storage or delivery.

[0229] MODE FOR CARRYING OUT THE INVENTION

[0230] This invention relates to a system for automatically generating advertisements by collecting advertising data using a generative AI model. This system provides a smartphone interface, and when a user inputs product information, it generates multiple advertising patterns and adjusts them based on the user's feedback to output the optimal advertisement.

[0231] System configuration

[0232] The system mainly consists of the following components:

[0233] 1. Server: Responsible for collecting, analyzing, generating, storing, and transmitting data.

[0234] 2. Generative AI model: An algorithm that automatically generates ads based on collected data.

[0235] 3. Smartphone: A device that allows users to enter advertising data and provide reviews and feedback.

[0236] Hardware and software used

[0237] Hardware: Smartphones, cloud servers (e.g., AWS)

[0238] software:

[0239] Mobile frontend (e.g., Swift, Kotlin)

[0240] Backend (e.g. Node.js, Django)

[0241] Data analysis tools (Python, Pandas)

[0242] Image processing library (OpenCV)

[0243] Generative AI models (OpenAI GPT-4, TensorFlow, PyTorch)

[0244] API frameworks (e.g. FastAPI)

[0245] Cloud storage (e.g. AWS S3)

[0246] Streaming encoding tool (e.g. FFmpeg)

[0247] System Operation

[0248] 1. Data collection: Users use a smartphone application to enter advertising data such as product information, target demographics, and selling points, and send it to the server.

[0249] 2. Data analysis: The server analyzes the received data and converts it into a format suitable for the generative AI model, such as cleaning up text data and resizing image data.

[0250] 3. Ad generation: The server inputs the analyzed data into a generative AI model to automatically generate multiple ad patterns.

[0251] 4. Review and Feedback: The generated ad variants are provided to a review interface on the smartphone, where users can review them and provide feedback.

[0252] 5. Applying feedback: The server regenerates and adjusts the advertising pattern based on user feedback and outputs the optimal advertisement.

[0253] 6. Encoding into high-definition formats: The adjusted advertisements are encoded into high-definition formats and optimized for various media.

[0254] Specific examples

[0255] For example, if a company requests an ad for a new smartwatch, the process might go something like this:

[0256] 1. Data collection: The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0257] 2. Data pre-processing: The server processes the collected data and converts it into a format suitable for the generative AI model.

[0258] 3. Model Selection and Initialization: The server selects suitable text generation and image generation models for smartwatch ad generation and loads the necessary training data.

[0259] 4. Instructions for generating advertisements: The server inputs the preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0260] 5. Providing reviews: The server provides the generated advertising patterns to the user through a review interface, where the user can check and rate them.

[0261] 6. Feedback collection: The user requests the server to speed up the voice announcement tempo of the “health monitoring-emphasizing advertisement” as feedback, and the server collects this.

[0262] 7. Applying feedback: The server adjusts the "health monitoring emphasis advertisement" based on the user's feedback and speeds up the tempo of the voice announcement.

[0263] 8. Final encoding and output: The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[0264] Examples of prompt statements

[0265] Below are some examples of prompts to input to the generative AI model.

[0266] Prompt statement:

[0267] text

[0268] Product name: XYZ Smart Watch

[0269] Target demographic: Technology enthusiasts in their 20s and 30s

[0270] Features:

[0271] Health Monitoring Features

[0272] Stylish design

[0273] Long battery life

[0274] Selling points:

[0275] Ideal for health management

[0276] Stylish design using the latest technology

[0277] Lasts for a week on a single charge

[0278] Based on this information, generate three ad variations (emphasizing health monitoring, design, and usage scenarios, respectively).

[0279] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0280] Step 1:

[0281] User enters product information

[0282] The user uses a smartphone application to input product information (e.g., product name, features, target demographic, selling points). The input data is sent to the server. Input: Product information data. Output: Data sent to the server.

[0283] Step 2:

[0284] The server collects the data

[0285] The server receives product information data sent by the user and stores it in a database. It prepares the data for analysis. Input: Product information data sent by the user. Output: Product information data stored in the database.

[0286] Step 3:

[0287] The server analyzes the data

[0288] The server analyzes the collected product information data, cleans up the text data (e.g., removes unnecessary spaces and special characters), resizes the image data, and converts it into a format suitable for the generative AI model. Input: Product information data stored in the database. Output: Data in a format suitable for the generative AI model.

[0289] Step 4:

[0290] Selecting and initializing a generative AI model

[0291] The server selects the optimal generative AI model depending on the type of data, loads the necessary training data, and initializes the model. Input: Data in a format suitable for the generative AI model. Output: The initialized generative AI model.

[0292] Step 5:

[0293] Generating advertising variations

[0294] The server supplies input data to the generative AI model to generate multiple ad patterns. For example, ads that emphasize health monitoring, stylish design, or usage scenarios can be generated. Input: Initialized generative AI model and input data. Output: Multiple ad patterns.

[0295] Step 6:

[0296] User reviews ad variations

[0297] The server provides the generated ad patterns to a review interface on the smartphone, where the user can review them. Input: Multiple ad patterns. Output: User reviews and feedback.

[0298] Step 7:

[0299] Users provide feedback

[0300] The user inputs feedback for the advertisement pattern through the review interface, for example, requesting that the tempo of the voice announcement for the "advertisement emphasizing health monitoring" be made faster. Input: Feedback for the advertisement pattern. Output: Feedback sent to the server.

[0301] Step 8:

[0302] The server reflects the feedback

[0303] The server analyzes the feedback from the user and regenerates / adjusts the ad pattern based on it, for example by speeding up the tempo of the voice announcement. Input: Feedback sent to the server. Output: Adjusted ad pattern.

[0304] Step 9:

[0305] Encoding and Output

[0306] The server encodes the adjusted ad variants into high-resolution formats, optimizing them for various media. It generates the final ad file and provides a secure download link to the user. Input: Adjusted ad variant. Output: Final ad file and download link.

[0307] 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.

[0308] MODE FOR CARRYING OUT THE INVENTION

[0309] In accordance with this invention, a specific implementation method of a system for automatically generating advertisements by combining a generative AI model and an emotion engine is as follows: The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, adjusting and outputting the final advertisements, and recognizing and reflecting user emotions.

[0310] Data collection and preparation

[0311] 1. Data Collection

[0312] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[0313] For example, if a user requests an advertisement for a new smartwatch, information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness will be provided and collected by the server.

[0314] 2. Data Preprocessing

[0315] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model and emotion engine, which includes cleaning up text data, processing image data, and formatting audio data.

[0316] Preparing the generative AI model

[0317] 3. Model Selection and Initialization

[0318] The server analyzes the provided advertising data and its contents and selects an appropriate generative AI model and emotion engine. The generative AI model includes text generation, image generation, and voice generation, and the emotion engine is used to recognize the user's emotional state.

[0319] The server initializes these models and loads the necessary training data.

[0320] Ad generation

[0321] 4. Ad generation instructions

[0322] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, multiple advertisement patterns (such as advertisements emphasizing health monitoring functions or stylish designs) are generated.

[0323] Reviews and Feedback

[0324] 5. Providing Reviews

[0325] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each of them.

[0326] 6. Emotion recognition

[0327] The server uses an emotion engine to detect the user's emotional state, for example by analyzing facial expressions and voice tone to recognize emotions while the user is reviewing an advertisement.

[0328] 7. Emotion-based evaluation

[0329] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[0330] 8. Feedback and Emotion Data Storage

[0331] The feedback provided by the user and the emotional data at that time are stored on the server. For example, if a user likes an advertisement that emphasizes health monitoring, the positive emotional feedback is recorded.

[0332] Final ad adjustment and output

[0333] 9. Applying Feedback

[0334] The server then optimally adjusts the advertising pattern based on user feedback and emotional data, for example by changing the tempo of the voice announcement, modifying the text, or changing some of the visuals.

[0335] 10. Encoding and Serving Ads

[0336] The server then encodes the tailored ad variations into high-definition formats, optimizing them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0337] Specific examples

[0338] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[0339] 1. Data Collection

[0340] The server collects data on the smartwatch's functions, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0341] 2. Data Preprocessing

[0342] The server processes the collected data and converts it into a format suitable for generative AI models and emotion engines.

[0343] 3. Model Selection and Initialization

[0344] The server selects text generation and image generation models and emotion engines suitable for generating ads for smartwatches and loads the necessary training data.

[0345] 4. Ad generation instructions

[0346] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0347] 5. Providing Reviews

[0348] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[0349] 6. Emotion recognition

[0350] The server analyzes the user's emotions during the review using an emotion engine and collects the user's reactions for each advertising pattern.

[0351] 7. Gathering Feedback

[0352] As feedback, users request that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback while also recording the user's positive emotional data.

[0353] 8. Applying Feedback

[0354] The server adjusts the advertising pattern and speeds up the tempo of the voice announcements based on the user's feedback and emotional data.

[0355] 9. Encoding and Serving Ads

[0356] The server encodes the tailored advertisement into a high-resolution format and provides the final version to the user via a secure download link.

[0357] Based on the above steps, this system enables companies to generate high-quality advertisements in an efficient and low-risk manner, while also enabling advertisements that reflect user emotions.

[0358] The processing flow will be explained below.

[0359] MODE FOR CARRYING OUT THE INVENTION

[0360] Program processing flow

[0361] Step 1:

[0362] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[0363] Step 2:

[0364] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[0365] Step 3:

[0366] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[0367] Step 4:

[0368] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[0369] Step 5:

[0370] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[0371] Step 6:

[0372] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[0373] Step 7:

[0374] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[0375] Step 8:

[0376] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[0377] Step 9:

[0378] The server uses an emotion engine to analyze the user's emotional state during the review, for example, by evaluating the user's facial expressions and tone of voice in real time to recognize the user's emotions.

[0379] Step 10:

[0380] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[0381] Step 11:

[0382] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message more concise."

[0383] Step 12:

[0384] The server adjusts the advertising pattern based on user feedback and analyzed emotion data, for example, by changing the tempo of the voice announcement and modifying some of the text.

[0385] Step 13:

[0386] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[0387] Step 14:

[0388] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[0389] In this way, generative AI can generate high-quality ads in an efficient and low-risk manner. In addition, by combining it with an emotion engine, it becomes possible to create ads that reflect user emotions, resulting in highly appealing ads.

[0390] Example 2

[0391] 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."

[0392] Conventional ad generation systems have difficulty creating ads that are optimized for the target audience because they do not adequately adjust ads to reflect specific user feedback and emotions. Furthermore, there is no established method for integrating emotion recognition technology into the ad generation process to effectively reflect users' positive reactions.

[0393] 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 advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for using an emotion engine to recognize the user's emotional state, and means for final adjustment and output of the advertising patterns based on the user's feedback and emotion data. This enables the generation of optimized advertisements that reflect user emotions in real time.

[0394] A "generative AI model" is an algorithm that uses artificial intelligence technology to automatically generate advertising content such as text, images, and audio.

[0395] "Advertising data" refers to input data necessary to generate advertisements, such as product information, target demographics, appealing points, and past advertising examples.

[0396] An "emotion engine" is a software component for detecting and analyzing a user's emotional state, specifically analyzing data such as facial expressions and vocal tone.

[0397] "Feedback" refers to specific evaluations and opinions provided by users regarding the generated advertising patterns, as well as instructions for improvement.

[0398] "Ad Variants" are different variations or formats of an advertisement automatically generated by a generative AI model.

[0399] The "review interface" is an interface that allows a user to check the generated advertisement pattern and input feedback.

[0400] "Initialization" is the process of loading the settings and training data required to run a generative AI model.

[0401] "Encoding" is the process of outputting the generated advertising content in a high-resolution format or in a format optimized for a particular medium.

[0402] "Data collection" is the process of obtaining, structuring, and storing advertising data needed to generate ads from companies and other sources.

[0403] "Adjustment" refers to the process of optimizing advertising patterns based on user feedback and sentiment data, including text correction, audio tempo adjustment, and visual changes.

[0404] This invention is a system that automatically generates advertisements by combining a generative AI model and an emotion engine. Detailed embodiments are described below. The invention is composed of various means and processes centered on a server, a terminal, and a user.

[0405] Data collection and preparation

[0406] The server collects advertising data provided by companies, such as product information, target demographics, selling points, and past advertising examples. The collected data is stored in an advertising database and classified as text data, image data, and audio data. This data is then converted into the appropriate format for input into the generative AI model and emotion engine. Unnecessary information is removed from the text data and the formatting is adjusted. Image data is resized and its resolution is adjusted, and audio data is subjected to noise removal and sound quality adjustment.

[0407] For example, if a user requests an advertisement for a new smartwatch, the server collects information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness, and stores it in a database.

[0408] Preparing the generative AI model

[0409] The server selects the appropriate generative AI model (e.g., GPT-3 for text generation, DALL-E for image generation) and emotion engine based on the collected ad data. These models are initialized and loaded with the necessary training data to apply them to the specific ad generation task.

[0410] Ad generation

[0411] The server then inputs the pre-processed data into the selected generative AI model and sends a prompt to generate an ad, such as:

[0412] "Generate an ad for a smartwatch targeted at men in their 20s. Features include health monitoring, stylish design, and fitness features."

[0413] As a result, the generative AI model automatically generates multiple advertising patterns, such as ads that emphasize health monitoring, ads that emphasize design, and ads that show usage scenarios.

[0414] Reviews and Feedback

[0415] The server provides the generated ad template to the user through a review interface that can be viewed on the device. The user reviews the ad template and enters feedback. For example, the user can review the text, images, and audio of the ad through a smartphone or PC interface and enter specific feedback.

[0416] The server uses an emotion engine to analyze the user's facial expressions and tone of voice while reviewing the ad, for example, by using a webcam and microphone to collect the user's real-time reactions and recognize the user's emotional state based on the data obtained.

[0417] Final ad adjustment and output

[0418] The server then adjusts the ad template based on the collected feedback and sentiment data, modifying the text, adjusting the tempo of the voice announcement, and changing the visuals. The final ad template is then encoded in high-resolution format and output in a format optimized for each medium. The final ad file is then provided to the user via a secure download link.

[0419] For example, if an "advertisement emphasizing health monitoring" is adjusted to have a faster audio tempo, the adjusted advertisement will be output as the final version.

[0420] The above is a specific embodiment for carrying out the present invention, which makes it possible to generate an optimized advertisement that reflects the user's emotions in real time.

[0421] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0422] Step 1:

[0423] Data collection

[0424] The server collects advertising data such as product information, target demographics, appealing points, and past advertising examples provided by companies.

[0425] The server receives the data through an API or a management portal.

[0426] The server stores the received data in an advertisement database.

[0427] Input: Provided product information, target demographic, selling points, past advertising examples

[0428] Output: Structured data in an advertising database

[0429] Step 2:

[0430] Data preprocessing

[0431] The server analyzes and cleans the raw data collected.

[0432] The server cleans the text data, removing unnecessary information and formatting.

[0433] The server resizes and adjusts the resolution of the image data.

[0434] The server performs noise removal and sound quality adjustment on the audio data.

[0435] Input: Raw data stored in the advertising database

[0436] Output: Data converted into a format suitable for generative AI models and emotion engines

[0437] Step 3:

[0438] Model Selection and Initialization

[0439] The server selects the appropriate generative AI model and emotion engine based on the collected advertising data. Specifically, it uses GPT-3 for text generation and DALL-E for image generation.

[0440] The server initializes the selected generative AI model and emotion engine and loads the training data.

[0441] Input: Data converted into a format suitable for generative AI models and emotion engines

[0442] Output: Initialized generative AI model and emotion engine

[0443] Step 4:

[0444] Ad generation instructions

[0445] The server inputs the preprocessed data into the generative AI model and sends prompt sentences.

[0446] For example, enter the prompt text, "Generate a smartwatch ad for men in their 20s. Features include health monitoring, stylish design, and fitness functions."

[0447] The server receives the generated advertisement pattern.

[0448] Input: prompts, data input by generative AI models

[0449] Output: Multiple ad variations

[0450] Step 5:

[0451] Providing a review

[0452] The server provides the generated advertisement patterns to the user through a review interface that can be viewed on a terminal.

[0453] The user reviews each ad variant and enters specific feedback.

[0454] Input: Generated ad patterns

[0455] Output: User feedback

[0456] Step 6:

[0457] emotion recognition

[0458] The server uses an emotion engine to analyze the user's facial expressions and tone of voice during the advertisement review.

[0459] The server uses a webcam and microphone to collect the user's real-time responses.

[0460] The server recognizes the user's emotional state based on the obtained data.

[0461] Input: Real-time data of the user while reviewing the ad (facial expressions, tone of voice)

[0462] Output: User's emotional state data

[0463] Step 7:

[0464] Emotion-based ratings

[0465] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[0466] The server sets a higher priority to the advertisement pattern that receives the most positive responses.

[0467] Input: User's emotional state data

[0468] Output: Priority of the evaluated ad variants

[0469] Step 8:

[0470] Feedback and sentiment data storage

[0471] The server stores the user feedback and emotion data in a database.

[0472] The server saves it for future use in the ad generation task.

[0473] Input: User feedback and sentiment data

[0474] Output: Stored feedback and emotion data

[0475] Step 9:

[0476] Applying Feedback

[0477] The server adjusts advertising patterns based on collected feedback and sentiment data.

[0478] The server modifies the text of the advertisement, adjusts the tempo of the voice announcement, changes the visuals, etc.

[0479] Input: User feedback and sentiment data

[0480] Output: Adjusted ad patterns

[0481] Step 10:

[0482] Encoding and Serving Ads

[0483] The server encodes the finalized ad variations in high-definition format.

[0484] The server outputs the advertisement in a format optimized for each medium.

[0485] The server provides the final ad file to the user via a secure download link.

[0486] Input: Adjusted ad pattern

[0487] Output: Final ad file in high resolution format

[0488] The above is the specific processing flow of the program for this system.

[0489] (Application example 2)

[0490] 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."

[0491] Conventional ad generation systems have difficulty taking into account user emotional data, making it difficult for the ads they provide to reflect the user's psychological state and preferences. Furthermore, there are limited means for effectively utilizing real-time user feedback, making it difficult to improve the quality of ads. There is a need for a system that can analyze user emotions in real time and reflect that data in ad generation to provide more effective and attractive ads.

[0492] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0493] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for collecting and analyzing user emotional data in real time, and means for making final adjustments to and outputting advertising patterns based on the user emotional data and feedback, thereby making it possible to instantly reflect the user's emotions and provide advertisements optimized for the user's emotional state.

[0494] A "generative AI model" is a model that uses artificial intelligence to generate advertising data such as text, images, and audio.

[0495] "Advertising data" refers to information for creating advertisements, including product information, target demographics, selling points, and past advertising examples.

[0496] "Emotion data" refers to data relating to the user's emotional state obtained by analyzing the user's facial expressions, voice tone, and other sensor information.

[0497] "Advertising variations" are multiple variations of advertisements automatically generated by a generative AI model.

[0498] "Reviews and feedback" refers to the evaluation and opinions provided by users regarding the provided advertising patterns.

[0499] "Final adjustment and output" refers to modifying the advertising pattern based on user feedback and emotional data and generating the final version.

[0500] A "virtual store" is a virtual shop that exists on the Internet and is a place where users can browse and purchase products.

[0501] "Means for collecting and analyzing in real time" refers to methods and technologies for instantly collecting user emotional data and analyzing it on the spot.

[0502] The "means for prioritizing" is a means for selecting the most suitable advertising pattern from among a plurality of advertising patterns based on the user's emotional data.

[0503] In this invention, we use a generative AI model and an emotion engine to build a system that provides advertisements that reflect customer emotions in a virtual store. The specific system configuration is as follows.

[0504] The server first collects advertising data provided by each company, such as product information, target demographics, selling points, and past advertising examples. This data is structured and stored in an advertising database. Devices such as smartphones and head-mounted displays collect users' purchasing history and preference tag information through apps. Based on this collected data, it is converted into a format suitable for the generative AI model. During this process, text data is cleaned up, image data is processed, and audio data is shaped.

[0505] The server uses this processed data to select and initialize a generative AI model suitable for the product. The generative AI models include a text generation model, an image generation model, and a voice generation model. The server uses these models to automatically generate multiple ad variations. The user then reviews these ads on a smartphone app and provides feedback.

[0506] During this review, the device collects the user's emotional data in real time and analyzes it using an emotion engine. For example, it uses a camera and microphone to analyze the user's facial expressions and tone of voice to obtain emotional data. The server then uses this emotional data and user feedback to make final adjustments to the advertising patterns. For example, if the user expresses positive emotions toward an advertisement emphasizing health monitoring, that advertisement will be displayed preferentially. The finalized advertisements are then encoded in high-resolution formats and optimized for various media.

[0507] As a concrete example, if a company requests the creation of an advertisement for a "new smartwatch," the following process takes place: The server collects and preprocesses data on the smartwatch's functional information, its target demographic of men in their 20s, and its appealing health monitoring function and stylish design. A generative AI model automatically generates patterns, such as advertisements that emphasize health monitoring, advertisements that emphasize design, and advertisements that show usage scenarios. Users view these advertisements on their smartphone app, and emotional data analyzed by an emotion engine is sent to the server. The server prioritizes and provides advertisement patterns that elicit a positive response to the user. Finally, these advertisements are encoded into high-resolution format and displayed in a virtual store.

[0508] An example of a prompt is:

[0509] "We collect data for product ID 12345, perform text and image preprocessing, and then use a generative AI model to generate product ads. Based on user sentiment data, we select the most appropriate ad pattern."

[0510] As described above, this system can instantly reflect the user's emotions and provide personalized advertisements.

[0511] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0512] Step 1:

[0513] The server collects advertising data such as product information, target demographics, selling points, and past advertising examples provided by each company. This collected data is structured and stored in an advertising database. The input is various advertising data provided by companies, and the output is a structured advertising database. Specifically, the server obtains data through APIs and data upload functions and stores it in the database.

[0514] Step 2:

[0515] The server performs preprocessing of the collected advertising data, such as cleaning up text data, processing image data, and formatting audio data. The input is data from the advertising database, and the output is preprocessed data. Specifically, the server converts the data into an appropriate format using natural language processing technology and image processing software (e.g., OpenCV).

[0516] Step 3:

[0517] The server selects and initializes a generative AI model using the preprocessed data. The input is the preprocessed data, and the output is the initialized generative AI model. Specifically, the server loads the model using a generative AI model library (e.g., TensorFlow or PyTorch) and feeds it training data.

[0518] Step 4:

[0519] The server inputs preprocessed data into the generative AI model to automatically generate multiple ad patterns. The input is the initialized generative AI model and preprocessed data, and the output is multiple ad patterns. Specifically, it uses the model's inference function to generate ad text, images, and audio.

[0520] Step 5:

[0521] The terminal provides the generated advertisement pattern to the user and receives reviews and feedback. The input is the advertisement pattern from the server, and the output is the user's review and feedback. Specific operations include displaying the generated advertisement through a user interface and collecting ratings and comments.

[0522] Step 6:

[0523] The device collects the user's emotional data in real time and analyzes it using an emotion engine. The input is the user's facial expressions and vocal tone, and the output is the analyzed emotional data. Specifically, it uses a camera and microphone and analyzes the data using emotion recognition software (e.g., Affectiva SDK).

[0524] Step 7:

[0525] The server then makes final adjustments to the ad patterns based on the collected user emotion data and feedback. The input is the analyzed emotion data and feedback, and the output is the adjusted ad pattern. Specific operations include correcting text and images to improve the quality of the ad.

[0526] Step 8:

[0527] The server then encodes the final adjusted ad template in a high-resolution format, optimizing it for various media. The input is the adjusted ad template, and the output is a high-resolution ad file. Specifically, the server uses video encoding software (e.g., FFmpeg) to convert the data into the appropriate format.

[0528] Step 9:

[0529] Finally, users can view the optimized advertisements in the virtual store, develop interest in the products, and make a purchase. The input is a high-resolution advertisement file, and the output is the user's purchasing behavior. Specifically, users view the advertisements using a smartphone or head-mounted display and click the purchase button.

[0530] 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.

[0531] 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.

[0532] 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.

[0533] [Second embodiment]

[0534] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0535] 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.

[0536] 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).

[0537] 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.

[0538] 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.

[0539] 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).

[0540] 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.

[0541] 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.

[0542] 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.

[0543] 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.

[0544] 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.

[0545] 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."

[0546] MODE FOR CARRYING OUT THE INVENTION

[0547] The following is a specific method for implementing a system for automatically generating advertisements using a generative AI model according to the present invention. The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting advertisements.

[0548] Data collection and preparation

[0549] 1. Data Collection

[0550] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[0551] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[0552] 2. Data Preprocessing

[0553] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model, which includes cleaning up text data, processing image data, and formatting audio files.

[0554] Preparing the generative AI model

[0555] 3. Model Selection and Initialization

[0556] The server analyzes the type and content of the provided data and selects the most appropriate generative AI model (text generation, image generation, speech generation, etc.). After selecting the model, it loads the necessary training data and initializes the model.

[0557] Ad generation

[0558] 4. Ad generation instructions

[0559] The server inputs the preprocessed data into the generative AI model, which then automatically generates multiple ad patterns, such as energetic ads, ads that emphasize technical aspects, and ads that show actual usage scenarios.

[0560] Reviews and Feedback

[0561] 5. Providing Reviews

[0562] The server generates a review interface for providing the generated plurality of advertisement patterns to a user, and the user reviews the advertisement patterns through the provided interface and inputs feedback for each advertisement.

[0563] 6. Gathering Feedback

[0564] The server collects and stores user feedback for use in the next ad generation task, providing specific suggestions for improvement, such as making the music faster or the message more concise.

[0565] Final ad adjustment and output

[0566] 7. Applying Feedback

[0567] The server selects the best ad pattern based on user feedback and makes necessary adjustments, such as changing the tempo of the voice announcement and modifying some of the text.

[0568] 8. Final encoding and output

[0569] The server encodes the tailored ad variations into high-definition formats and optimizes them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0570] Specific examples

[0571] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[0572] 1. Data Collection

[0573] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0574] 2. Data Preprocessing

[0575] The server processes the collected data and converts it into a format suitable for the generative AI model.

[0576] 3. Model Selection and Initialization

[0577] The server selects text generation and image generation models suitable for generating ads for smartwatches and loads the necessary training data.

[0578] 4. Ad generation instructions

[0579] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0580] 5. Providing Reviews

[0581] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[0582] 6. Gathering Feedback

[0583] The user requests that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster as feedback, and the server collects this.

[0584] 7. Applying Feedback

[0585] The server adjusts the "health monitoring emphasis advertisement" based on user feedback, speeding up the tempo of the voice announcements.

[0586] 8. Final encoding and output

[0587] The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[0588] Based on these steps, the system enables businesses to generate high-quality advertisements in an efficient and low-risk manner.

[0589] The processing flow will be explained below.

[0590] MODE FOR CARRYING OUT THE INVENTION

[0591] Program processing flow

[0592] Step 1:

[0593] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[0594] Step 2:

[0595] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[0596] Step 3:

[0597] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[0598] Step 4:

[0599] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[0600] Step 5:

[0601] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[0602] Step 6:

[0603] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[0604] Step 7:

[0605] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[0606] Step 8:

[0607] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[0608] Step 9:

[0609] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message shorter."

[0610] Step 10:

[0611] The server collects user feedback and adjusts the advertising pattern based on that feedback, for example speeding up the voice announcements or shortening the text.

[0612] Step 11:

[0613] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[0614] Step 12:

[0615] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[0616] In this way, generative AI can be used to generate high-quality ads in an efficient and low-risk manner.

[0617] Example 1

[0618] 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."

[0619] Ad production is a time-consuming and costly process, especially when multiple ad variations need to be generated for different target audiences. It's also challenging to quickly incorporate feedback and deliver optimal ads while maintaining ad quality, making it difficult to maximize the effectiveness of advertising campaigns.

[0620] 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.

[0621] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for analyzing and preprocessing the collected advertising data, means for selecting and initializing an optimal generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, and means for final adjustment and output of the advertising patterns based on user feedback. This reduces the time and cost of advertising production, makes it possible to quickly and effectively generate multiple advertising patterns, and provide optimal advertisements by reflecting feedback.

[0622] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to automatically generate advertising content from a variety of data, including text, images, and audio.

[0623] "Advertising data" is a general term for information necessary for generating advertisements, such as product information provided by companies, target demographics, appealing points, and past advertising examples.

[0624] "Input data" refers to advertising data that is input into the generative AI model, and is provided to the generative AI model after initial data collection and preprocessing.

[0625] "Preprocessing" refers to the process of data processing and cleaning to convert advertising data into a format suitable for generative AI models.

[0626] "Advertising patterns" is a collective term for multiple advertising formats automatically generated by a generative AI model.

[0627] "Review" is an operation in which a user checks and evaluates the generated advertisement pattern.

[0628] "Feedback" refers to opinions and suggestions for improvement that a user provides regarding the generated advertising pattern.

[0629] "Final adjustment" refers to the process of optimizing the content and format of the advertising pattern by reflecting user feedback.

[0630] "Output" refers to the process of encoding the final adjusted advertisement pattern to generate the final advertisement file.

[0631] A "promotion content creation request" is a request sent by a user to a server to request the creation of new advertising or marketing materials.

[0632] "Quality evaluation" is the process of evaluating the effectiveness and suitability of advertising patterns output by a generative AI model.

[0633] "High definition format" refers to a file format that allows the generated advertisement to be presented in high visual and audio resolution.

[0634] "Media optimization" is the process of converting the generated advertising patterns into formats suitable for different distribution media such as the Internet, television, and radio.

[0635] This invention is a system for automatically generating advertisements using a generative AI model. The main functions of the system consist of the following steps: collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting the advertisements.

[0636] Data collection and preparation

[0637] Data collection

[0638] The server collects advertising data provided by companies, such as product information, target demographics, key selling points, and past advertising examples. This data is structured and stored in an advertising database. Data is collected through APIs and database connections and converted into the format required to generate ads.

[0639] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[0640] Data Preprocessing

[0641] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model. This includes cleaning up text data, processing image data, and formatting audio files. For text data, unnecessary spaces and special characters are removed and natural language processing tools are used to segment the data. For image data, image resizing and formatting are performed, and for audio data, noise reduction and encoding are performed.

[0642] Preparing the generative AI model

[0643] Model Selection and Initialization

[0644] The server analyzes the provided data and selects the optimal generative AI model (text generation, image generation, speech generation, etc.). Specifically, it uses GPT-3 as the text generation model, DALL-E as the image generation model, and WaveNet as the speech generation model. After the selection, it loads the necessary training data and initializes the model.

[0645] Ad generation

[0646] Ad generation instructions

[0647] The server inputs the preprocessed data into the generative AI model to automatically generate multiple ad patterns, and the generated results are recorded in a log to generate multiple different ad patterns (such as energetic ads, technology-focused ads, and ads showing usage scenarios).

[0648] Reviews and Feedback

[0649] Providing a review

[0650] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface for displaying the generated advertisements in thumbnail and playback format and allowing the user to evaluate them.

[0651] Collecting feedback

[0652] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[0653] Final ad adjustment and output

[0654] Applying Feedback

[0655] The server adjusts the advertising patterns based on user feedback, specifically by readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[0656] Final encoding and output

[0657] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0658] Specific examples

[0659] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[0660] 1. Data Collection

[0661] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0662] 2. Data Preprocessing

[0663] The server preprocesses the collected data using natural language processing tools and image processing tools.

[0664] 3. Model Selection and Initialization

[0665] Based on the collected data, the server selects a generative AI model such as GPT-3, DALL-E, or WaveNet and loads the necessary training data.

[0666] 4. Ad generation instructions

[0667] The server inputs the preprocessed data into the model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0668] 5. Providing Reviews

[0669] A user uses a web interface to review and rate the generated ad variations.

[0670] 6. Gathering Feedback

[0671] The user may provide feedback requesting that the tempo of the audio for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback.

[0672] 7. Applying Feedback

[0673] The server adjusts the tempo of the audio and regenerates it based on the feedback.

[0674] 8. Final encoding and output

[0675] The server encodes the tailored advertisement into high definition format and provides the final version to the user via a secure download link.

[0676] Based on the above steps, the system enables companies to generate high-quality advertisements efficiently and with low risk.

[0677] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0678] Step 1: Data collection

[0679] The server collects data provided by companies, such as product information, target demographics, selling points, and past advertising examples. This data is retrieved through APIs and database connections, structured, and stored in an advertising database.

[0680] Input: Promotional content creation requests from companies, product information, target demographic data

[0681] Data processing: Acquired through APIs and database connections, stored in a structured database

[0682] Output: Structured data stored in the ad database

[0683] Step 2: Preprocessing the data

[0684] The server analyzes data collected from the advertising database and converts it into a format suitable for the generative AI model. Text data is stripped of unnecessary spaces and special characters and segmented using natural language processing tools. Image data is resized and formatted, and audio data is noise-reduced and encoded.

[0685] Input: Structured advertising data in an advertising database

[0686] Data processing: cleaning up text data, resizing and formatting images, noise reduction and encoding of audio data

[0687] Output: Preprocessed data (cleaned text, enhanced images, formatted audio)

[0688] Step 3: Model selection and initialization

[0689] The server selects the optimal generative AI model based on the preprocessed data. Specifically, it selects GPT-3 for text generation, DALL-E for image generation, and WaveNet for speech generation. After the selection, it loads the necessary training data and initializes the model.

[0690] Input: Preprocessed data

[0691] Data processing: Analyzing data content, selecting the optimal model, and loading training data

[0692] Output: Initialized generative AI model

[0693] Step 4: Ad generation instructions

[0694] The server inputs the preprocessed data into a generative AI model to automatically generate multiple ad patterns, which are then logged and used to generate different patterns, such as energetic ads, technology-focused ads, and ads showing usage scenarios.

[0695] Input: Initialized generative AI model and preprocessed data

[0696] Data processing: Generating advertising patterns using generative AI models

[0697] Output: Multiple ad variations generated

[0698] Step 5: Provide a review

[0699] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface is used to display the generated advertisements in thumbnail and playback format and allow the user to evaluate them.

[0700] Input: Generated ad patterns

[0701] Data processing: Generate review interface and display advertising patterns

[0702] Output: The review interface presented to the user

[0703] Step 6: Gather feedback

[0704] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[0705] Input: User feedback

[0706] Data processing: collecting and storing feedback

[0707] Output: Saved feedback data

[0708] Step 7: Applying feedback

[0709] The server adjusts the ad patterns based on user feedback, readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[0710] Input: Saved feedback data

[0711] Data processing: Readjust model parameters based on feedback and re-run ad generation

[0712] Output: Adjusted ad patterns

[0713] Step 8: Final encoding and output

[0714] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0715] Input: Adjusted ad pattern

[0716] Data processing: Encoding to high definition formats and optimizing for media

[0717] Output: Final ad file provided with a secure download link

[0718] (Application example 1)

[0719] 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."

[0720] Modern advertising production requires a significant amount of time and effort, making it difficult to generate effective ads, especially for targeted audiences. Furthermore, systems for quickly incorporating user feedback are inadequate, and improvements are needed to improve the quality of advertising. Furthermore, there is a lack of ad generation systems that utilize mobile devices such as smartphones, creating a demand for a more flexible and rapid advertising production process.

[0721] 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.

[0722] In this invention, the server includes: means for collecting advertising data as input data using a generative AI model; means for selecting and initializing a generative AI model based on the collected advertising data; means for automatically generating multiple advertising patterns using the generative AI model; means for providing the generated advertising patterns to users and receiving reviews and feedback; means for making final adjustments to and outputting the advertising patterns based on user feedback; means for providing an interface using a smartphone; means for users to input product information and send the data to the server; and means for analyzing the provided data and converting it into a format suitable for the generative AI model. This enables users to easily collect and input advertising data using a smartphone, quickly generate high-quality advertisements using an optimal generative AI model based on the data, and regenerate advertisements that reflect user feedback.

[0723] A "generative AI model" is an algorithm that uses machine learning to automatically analyze data and perform a specific task, such as generating an ad.

[0724] "Advertising data" refers to information such as product information, target demographics, and appealing points that are necessary when creating an advertisement.

[0725] A "server" is a computing system that is responsible for collecting, analyzing, generating, storing, and transmitting data.

[0726] "Input data" refers to data such as product information, target demographics, and past advertising examples provided to generate advertisements.

[0727] "Ad Variants" refers to multiple different variations of an advertisement automatically generated by a generative AI model.

[0728] "User" refers to any person or entity that utilizes the Ad Generation System to generate, review, and provide feedback on Ads.

[0729] "Review interface" refers to a user interface that allows a user to review the generated advertising patterns and provide feedback.

[0730] "Feedback" refers to improvements and evaluations provided by users regarding the generated advertising patterns.

[0731] "Smartphone" means a mobile device that allows access to the ad generation system and enables data entry, review and feedback.

[0732] "Interface" refers to the means by which a user interacts with a system.

[0733] "Data analysis" refers to the process of converting collected data into useful information and forms.

[0734] "High-resolution format" refers to a format for exporting high-quality advertisements to a variety of media formats.

[0735] "Encoding" refers to the process of converting a generated advertisement into a particular format for storage or delivery.

[0736] MODE FOR CARRYING OUT THE INVENTION

[0737] This invention relates to a system for automatically generating advertisements by collecting advertising data using a generative AI model. This system provides a smartphone interface, and when a user inputs product information, it generates multiple advertising patterns and adjusts them based on the user's feedback to output the optimal advertisement.

[0738] System configuration

[0739] The system mainly consists of the following components:

[0740] 1. Server: Responsible for collecting, analyzing, generating, storing, and transmitting data.

[0741] 2. Generative AI model: An algorithm that automatically generates ads based on collected data.

[0742] 3. Smartphone: A device that allows users to enter advertising data and provide reviews and feedback.

[0743] Hardware and software used

[0744] Hardware: Smartphones, cloud servers (e.g., AWS)

[0745] software:

[0746] Mobile frontend (e.g., Swift, Kotlin)

[0747] Backend (e.g. Node.js, Django)

[0748] Data analysis tools (Python, Pandas)

[0749] Image processing library (OpenCV)

[0750] Generative AI models (OpenAI GPT-4, TensorFlow, PyTorch)

[0751] API frameworks (e.g. FastAPI)

[0752] Cloud storage (e.g. AWS S3)

[0753] Streaming encoding tool (e.g. FFmpeg)

[0754] System Operation

[0755] 1. Data collection: Users use a smartphone application to enter advertising data such as product information, target demographics, and selling points, and send it to the server.

[0756] 2. Data analysis: The server analyzes the received data and converts it into a format suitable for the generative AI model, such as cleaning up text data and resizing image data.

[0757] 3. Ad generation: The server inputs the analyzed data into a generative AI model to automatically generate multiple ad patterns.

[0758] 4. Review and Feedback: The generated ad variants are provided to a review interface on the smartphone, where users can review them and provide feedback.

[0759] 5. Applying feedback: The server regenerates and adjusts the advertising pattern based on user feedback and outputs the optimal advertisement.

[0760] 6. Encoding into high-definition formats: The adjusted advertisements are encoded into high-definition formats and optimized for various media.

[0761] Specific examples

[0762] For example, if a company requests an ad for a new smartwatch, the process might go something like this:

[0763] 1. Data collection: The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0764] 2. Data pre-processing: The server processes the collected data and converts it into a format suitable for the generative AI model.

[0765] 3. Model Selection and Initialization: The server selects suitable text generation and image generation models for smartwatch ad generation and loads the necessary training data.

[0766] 4. Instructions for generating advertisements: The server inputs the preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0767] 5. Providing reviews: The server provides the generated advertising patterns to the user through a review interface, where the user can check and rate them.

[0768] 6. Feedback collection: The user requests the server to speed up the voice announcement tempo of the “health monitoring-emphasizing advertisement” as feedback, and the server collects this.

[0769] 7. Applying feedback: The server adjusts the "health monitoring emphasis advertisement" based on the user's feedback and speeds up the tempo of the voice announcement.

[0770] 8. Final encoding and output: The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[0771] Examples of prompt statements

[0772] Below are some examples of prompts to input to the generative AI model.

[0773] Prompt statement:

[0774] text

[0775] Product name: XYZ Smart Watch

[0776] Target demographic: Technology enthusiasts in their 20s and 30s

[0777] Features:

[0778] Health Monitoring Features

[0779] Stylish design

[0780] Long battery life

[0781] Selling points:

[0782] Ideal for health management

[0783] Stylish design using the latest technology

[0784] Lasts for a week on a single charge

[0785] Based on this information, generate three ad variations (emphasizing health monitoring, design, and usage scenarios, respectively).

[0786] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0787] Step 1:

[0788] User enters product information

[0789] The user uses a smartphone application to input product information (e.g., product name, features, target demographic, selling points). The input data is sent to the server. Input: Product information data. Output: Data sent to the server.

[0790] Step 2:

[0791] The server collects the data

[0792] The server receives product information data sent by the user and stores it in a database. It prepares the data for analysis. Input: Product information data sent by the user. Output: Product information data stored in the database.

[0793] Step 3:

[0794] The server analyzes the data

[0795] The server analyzes the collected product information data, cleans up the text data (e.g., removes unnecessary spaces and special characters), resizes the image data, and converts it into a format suitable for the generative AI model. Input: Product information data stored in the database. Output: Data in a format suitable for the generative AI model.

[0796] Step 4:

[0797] Selecting and initializing a generative AI model

[0798] The server selects the optimal generative AI model depending on the type of data, loads the necessary training data, and initializes the model. Input: Data in a format suitable for the generative AI model. Output: The initialized generative AI model.

[0799] Step 5:

[0800] Generating advertising variations

[0801] The server supplies input data to the generative AI model to generate multiple ad patterns. For example, ads that emphasize health monitoring, stylish design, or usage scenarios can be generated. Input: Initialized generative AI model and input data. Output: Multiple ad patterns.

[0802] Step 6:

[0803] User reviews ad variations

[0804] The server provides the generated ad patterns to a review interface on the smartphone, where the user can review them. Input: Multiple ad patterns. Output: User reviews and feedback.

[0805] Step 7:

[0806] Users provide feedback

[0807] The user inputs feedback for the advertisement pattern through the review interface, for example, requesting that the tempo of the voice announcement for the "advertisement emphasizing health monitoring" be made faster. Input: Feedback for the advertisement pattern. Output: Feedback sent to the server.

[0808] Step 8:

[0809] The server reflects the feedback

[0810] The server analyzes the feedback from the user and regenerates / adjusts the ad pattern based on it, for example by speeding up the tempo of the voice announcement. Input: Feedback sent to the server. Output: Adjusted ad pattern.

[0811] Step 9:

[0812] Encoding and Output

[0813] The server encodes the adjusted ad variants into high-resolution formats, optimizing them for various media. It generates the final ad file and provides a secure download link to the user. Input: Adjusted ad variant. Output: Final ad file and download link.

[0814] 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.

[0815] MODE FOR CARRYING OUT THE INVENTION

[0816] In accordance with this invention, a specific implementation method of a system for automatically generating advertisements by combining a generative AI model and an emotion engine is as follows: The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, adjusting and outputting the final advertisements, and recognizing and reflecting user emotions.

[0817] Data collection and preparation

[0818] 1. Data Collection

[0819] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[0820] For example, if a user requests an advertisement for a new smartwatch, information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness will be provided and collected by the server.

[0821] 2. Data Preprocessing

[0822] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model and emotion engine, which includes cleaning up text data, processing image data, and formatting audio data.

[0823] Preparing the generative AI model

[0824] 3. Model Selection and Initialization

[0825] The server analyzes the provided advertising data and its contents and selects an appropriate generative AI model and emotion engine. The generative AI model includes text generation, image generation, and voice generation, and the emotion engine is used to recognize the user's emotional state.

[0826] The server initializes these models and loads the necessary training data.

[0827] Ad generation

[0828] 4. Ad generation instructions

[0829] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, multiple advertisement patterns (such as advertisements emphasizing health monitoring functions or stylish designs) are generated.

[0830] Reviews and Feedback

[0831] 5. Providing Reviews

[0832] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each of them.

[0833] 6. Emotion recognition

[0834] The server uses an emotion engine to detect the user's emotional state, for example by analyzing facial expressions and voice tone to recognize emotions while the user is reviewing an advertisement.

[0835] 7. Emotion-based evaluation

[0836] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[0837] 8. Feedback and Emotion Data Storage

[0838] The feedback provided by the user and the emotional data at that time are stored on the server. For example, if a user likes an advertisement that emphasizes health monitoring, the positive emotional feedback is recorded.

[0839] Final ad adjustment and output

[0840] 9. Applying Feedback

[0841] The server then optimally adjusts the advertising pattern based on user feedback and emotional data, for example by changing the tempo of the voice announcement, modifying the text, or changing some of the visuals.

[0842] 10. Encoding and Serving Ads

[0843] The server then encodes the tailored ad variations into high-definition formats, optimizing them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[0844] Specific examples

[0845] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[0846] 1. Data Collection

[0847] The server collects data on the smartwatch's functions, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[0848] 2. Data Preprocessing

[0849] The server processes the collected data and converts it into a format suitable for generative AI models and emotion engines.

[0850] 3. Model Selection and Initialization

[0851] The server selects text generation and image generation models and emotion engines suitable for generating ads for smartwatches and loads the necessary training data.

[0852] 4. Ad generation instructions

[0853] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[0854] 5. Providing Reviews

[0855] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[0856] 6. Emotion recognition

[0857] The server analyzes the user's emotions during the review using an emotion engine and collects the user's reactions for each advertising pattern.

[0858] 7. Gathering Feedback

[0859] As feedback, users request that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback while also recording the user's positive emotional data.

[0860] 8. Applying Feedback

[0861] The server adjusts the advertising pattern and speeds up the tempo of the voice announcements based on the user's feedback and emotional data.

[0862] 9. Encoding and Serving Ads

[0863] The server encodes the tailored advertisement into a high-resolution format and provides the final version to the user via a secure download link.

[0864] Based on the above steps, this system enables companies to generate high-quality advertisements in an efficient and low-risk manner, while also enabling advertisements that reflect user emotions.

[0865] The processing flow will be explained below.

[0866] MODE FOR CARRYING OUT THE INVENTION

[0867] Program processing flow

[0868] Step 1:

[0869] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[0870] Step 2:

[0871] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[0872] Step 3:

[0873] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[0874] Step 4:

[0875] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[0876] Step 5:

[0877] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[0878] Step 6:

[0879] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[0880] Step 7:

[0881] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[0882] Step 8:

[0883] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[0884] Step 9:

[0885] The server uses an emotion engine to analyze the user's emotional state during the review, for example, by evaluating the user's facial expressions and tone of voice in real time to recognize the user's emotions.

[0886] Step 10:

[0887] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[0888] Step 11:

[0889] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message more concise."

[0890] Step 12:

[0891] The server adjusts the advertising pattern based on user feedback and analyzed emotion data, for example, by changing the tempo of the voice announcement and modifying some of the text.

[0892] Step 13:

[0893] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[0894] Step 14:

[0895] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[0896] In this way, generative AI can generate high-quality ads in an efficient and low-risk manner. In addition, by combining it with an emotion engine, it becomes possible to create ads that reflect user emotions, resulting in highly appealing ads.

[0897] Example 2

[0898] 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."

[0899] Conventional ad generation systems have difficulty creating ads that are optimized for the target audience because they do not adequately adjust ads to reflect specific user feedback and emotions. Furthermore, there is no established method for integrating emotion recognition technology into the ad generation process to effectively reflect users' positive reactions.

[0900] 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 advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for using an emotion engine to recognize the user's emotional state, and means for final adjustment and output of the advertising patterns based on the user's feedback and emotion data. This enables the generation of optimized advertisements that reflect user emotions in real time.

[0901] A "generative AI model" is an algorithm that uses artificial intelligence technology to automatically generate advertising content such as text, images, and audio.

[0902] "Advertising data" refers to input data necessary to generate advertisements, such as product information, target demographics, appealing points, and past advertising examples.

[0903] An "emotion engine" is a software component for detecting and analyzing a user's emotional state, specifically analyzing data such as facial expressions and vocal tone.

[0904] "Feedback" refers to specific evaluations and opinions provided by users regarding the generated advertising patterns, as well as instructions for improvement.

[0905] "Ad Variants" are different variations or formats of an advertisement automatically generated by a generative AI model.

[0906] The "review interface" is an interface that allows a user to check the generated advertisement pattern and input feedback.

[0907] "Initialization" is the process of loading the settings and training data required to run a generative AI model.

[0908] "Encoding" is the process of outputting the generated advertising content in a high-resolution format or in a format optimized for a particular medium.

[0909] "Data collection" is the process of obtaining, structuring, and storing advertising data needed to generate ads from companies and other sources.

[0910] "Adjustment" refers to the process of optimizing advertising patterns based on user feedback and sentiment data, including text correction, audio tempo adjustment, and visual changes.

[0911] This invention is a system that automatically generates advertisements by combining a generative AI model and an emotion engine. Detailed embodiments are described below. The invention is composed of various means and processes centered on a server, a terminal, and a user.

[0912] Data collection and preparation

[0913] The server collects advertising data provided by companies, such as product information, target demographics, selling points, and past advertising examples. The collected data is stored in an advertising database and classified as text data, image data, and audio data. This data is then converted into the appropriate format for input into the generative AI model and emotion engine. Unnecessary information is removed from the text data and the formatting is adjusted. Image data is resized and its resolution is adjusted, and audio data is subjected to noise removal and sound quality adjustment.

[0914] For example, if a user requests an advertisement for a new smartwatch, the server collects information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness, and stores it in a database.

[0915] Preparing the generative AI model

[0916] The server selects the appropriate generative AI model (e.g., GPT-3 for text generation, DALL-E for image generation) and emotion engine based on the collected ad data. These models are initialized and loaded with the necessary training data to apply them to the specific ad generation task.

[0917] Ad generation

[0918] The server then inputs the pre-processed data into the selected generative AI model and sends a prompt to generate an ad, such as:

[0919] "Generate an ad for a smartwatch targeted at men in their 20s. Features include health monitoring, stylish design, and fitness features."

[0920] As a result, the generative AI model automatically generates multiple advertising patterns, such as ads that emphasize health monitoring, ads that emphasize design, and ads that show usage scenarios.

[0921] Reviews and Feedback

[0922] The server provides the generated ad template to the user through a review interface that can be viewed on the device. The user reviews the ad template and enters feedback. For example, the user can review the text, images, and audio of the ad through a smartphone or PC interface and enter specific feedback.

[0923] The server uses an emotion engine to analyze the user's facial expressions and tone of voice while reviewing the ad, for example, by using a webcam and microphone to collect the user's real-time reactions and recognize the user's emotional state based on the data obtained.

[0924] Final ad adjustment and output

[0925] The server then adjusts the ad template based on the collected feedback and sentiment data, modifying the text, adjusting the tempo of the voice announcement, and changing the visuals. The final ad template is then encoded in high-resolution format and output in a format optimized for each medium. The final ad file is then provided to the user via a secure download link.

[0926] For example, if an "advertisement emphasizing health monitoring" is adjusted to have a faster audio tempo, the adjusted advertisement will be output as the final version.

[0927] The above is a specific embodiment for carrying out the present invention, which makes it possible to generate an optimized advertisement that reflects the user's emotions in real time.

[0928] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0929] Step 1:

[0930] Data collection

[0931] The server collects advertising data such as product information, target demographics, appealing points, and past advertising examples provided by companies.

[0932] The server receives the data through an API or a management portal.

[0933] The server stores the received data in an advertisement database.

[0934] Input: Provided product information, target demographic, selling points, past advertising examples

[0935] Output: Structured data in an advertising database

[0936] Step 2:

[0937] Data preprocessing

[0938] The server analyzes and cleans the raw data collected.

[0939] The server cleans the text data, removing unnecessary information and formatting.

[0940] The server resizes and adjusts the resolution of the image data.

[0941] The server performs noise removal and sound quality adjustment on the audio data.

[0942] Input: Raw data stored in the advertising database

[0943] Output: Data converted into a format suitable for generative AI models and emotion engines

[0944] Step 3:

[0945] Model Selection and Initialization

[0946] The server selects the appropriate generative AI model and emotion engine based on the collected advertising data. Specifically, it uses GPT-3 for text generation and DALL-E for image generation.

[0947] The server initializes the selected generative AI model and emotion engine and loads the training data.

[0948] Input: Data converted into a format suitable for generative AI models and emotion engines

[0949] Output: Initialized generative AI model and emotion engine

[0950] Step 4:

[0951] Ad generation instructions

[0952] The server inputs the preprocessed data into the generative AI model and sends prompt sentences.

[0953] For example, enter the prompt text, "Generate a smartwatch ad for men in their 20s. Features include health monitoring, stylish design, and fitness functions."

[0954] The server receives the generated advertisement pattern.

[0955] Input: prompts, data input by generative AI models

[0956] Output: Multiple ad variations

[0957] Step 5:

[0958] Providing a review

[0959] The server provides the generated advertisement patterns to the user through a review interface that can be viewed on a terminal.

[0960] The user reviews each ad variant and enters specific feedback.

[0961] Input: Generated ad patterns

[0962] Output: User feedback

[0963] Step 6:

[0964] emotion recognition

[0965] The server uses an emotion engine to analyze the user's facial expressions and tone of voice during the advertisement review.

[0966] The server uses a webcam and microphone to collect the user's real-time responses.

[0967] The server recognizes the user's emotional state based on the obtained data.

[0968] Input: Real-time data of the user while reviewing the ad (facial expressions, tone of voice)

[0969] Output: User's emotional state data

[0970] Step 7:

[0971] Emotion-based ratings

[0972] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[0973] The server sets a higher priority to the advertisement pattern that receives the most positive responses.

[0974] Input: User's emotional state data

[0975] Output: Priority of the evaluated ad variants

[0976] Step 8:

[0977] Feedback and sentiment data storage

[0978] The server stores the user feedback and emotion data in a database.

[0979] The server saves it for future use in the ad generation task.

[0980] Input: User feedback and sentiment data

[0981] Output: Stored feedback and emotion data

[0982] Step 9:

[0983] Applying Feedback

[0984] The server adjusts advertising patterns based on collected feedback and sentiment data.

[0985] The server modifies the text of the advertisement, adjusts the tempo of the voice announcement, changes the visuals, etc.

[0986] Input: User feedback and sentiment data

[0987] Output: Adjusted ad patterns

[0988] Step 10:

[0989] Encoding and Serving Ads

[0990] The server encodes the finalized ad variations in high-definition format.

[0991] The server outputs the advertisement in a format optimized for each medium.

[0992] The server provides the final ad file to the user via a secure download link.

[0993] Input: Adjusted ad pattern

[0994] Output: Final ad file in high resolution format

[0995] The above is the specific processing flow of the program for this system.

[0996] (Application example 2)

[0997] 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."

[0998] Conventional ad generation systems have difficulty taking into account user emotional data, making it difficult for the ads they provide to reflect the user's psychological state and preferences. Furthermore, there are limited means for effectively utilizing real-time user feedback, making it difficult to improve the quality of ads. There is a need for a system that can analyze user emotions in real time and reflect that data in ad generation to provide more effective and attractive ads.

[0999] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1000] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for collecting and analyzing user emotional data in real time, and means for making final adjustments to and outputting advertising patterns based on the user emotional data and feedback, thereby making it possible to instantly reflect the user's emotions and provide advertisements optimized for the user's emotional state.

[1001] A "generative AI model" is a model that uses artificial intelligence to generate advertising data such as text, images, and audio.

[1002] "Advertising data" refers to information for creating advertisements, including product information, target demographics, selling points, and past advertising examples.

[1003] "Emotion data" refers to data relating to the user's emotional state obtained by analyzing the user's facial expressions, voice tone, and other sensor information.

[1004] "Advertising variations" are multiple variations of advertisements automatically generated by a generative AI model.

[1005] "Reviews and feedback" refers to the evaluation and opinions provided by users regarding the provided advertising patterns.

[1006] "Final adjustment and output" refers to modifying the advertising pattern based on user feedback and emotional data and generating the final version.

[1007] A "virtual store" is a virtual shop that exists on the Internet and is a place where users can browse and purchase products.

[1008] "Means for collecting and analyzing in real time" refers to methods and technologies for instantly collecting user emotional data and analyzing it on the spot.

[1009] The "means for prioritizing" is a means for selecting the most suitable advertising pattern from among a plurality of advertising patterns based on the user's emotional data.

[1010] In this invention, we use a generative AI model and an emotion engine to build a system that provides advertisements that reflect customer emotions in a virtual store. The specific system configuration is as follows.

[1011] The server first collects advertising data provided by each company, such as product information, target demographics, selling points, and past advertising examples. This data is structured and stored in an advertising database. Devices such as smartphones and head-mounted displays collect users' purchasing history and preference tag information through apps. Based on this collected data, it is converted into a format suitable for the generative AI model. During this process, text data is cleaned up, image data is processed, and audio data is shaped.

[1012] The server uses this processed data to select and initialize a generative AI model suitable for the product. The generative AI models include a text generation model, an image generation model, and a voice generation model. The server uses these models to automatically generate multiple ad variations. The user then reviews these ads on a smartphone app and provides feedback.

[1013] During this review, the device collects the user's emotional data in real time and analyzes it using an emotion engine. For example, it uses a camera and microphone to analyze the user's facial expressions and tone of voice to obtain emotional data. The server then uses this emotional data and user feedback to make final adjustments to the advertising patterns. For example, if the user expresses positive emotions toward an advertisement emphasizing health monitoring, that advertisement will be displayed preferentially. The finalized advertisements are then encoded in high-resolution formats and optimized for various media.

[1014] As a concrete example, if a company requests the creation of an advertisement for a "new smartwatch," the following process takes place: The server collects and preprocesses data on the smartwatch's functional information, its target demographic of men in their 20s, and its appealing health monitoring function and stylish design. A generative AI model automatically generates patterns, such as advertisements that emphasize health monitoring, advertisements that emphasize design, and advertisements that show usage scenarios. Users view these advertisements on their smartphone app, and emotional data analyzed by an emotion engine is sent to the server. The server prioritizes and provides advertisement patterns that elicit a positive response to the user. Finally, these advertisements are encoded into high-resolution format and displayed in a virtual store.

[1015] An example of a prompt is:

[1016] "We collect data for product ID 12345, perform text and image preprocessing, and then use a generative AI model to generate product ads. Based on user sentiment data, we select the most appropriate ad pattern."

[1017] As described above, this system can instantly reflect the user's emotions and provide personalized advertisements.

[1018] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1019] Step 1:

[1020] The server collects advertising data such as product information, target demographics, selling points, and past advertising examples provided by each company. This collected data is structured and stored in an advertising database. The input is various advertising data provided by companies, and the output is a structured advertising database. Specifically, the server obtains data through APIs and data upload functions and stores it in the database.

[1021] Step 2:

[1022] The server performs preprocessing of the collected advertising data, such as cleaning up text data, processing image data, and formatting audio data. The input is data from the advertising database, and the output is preprocessed data. Specifically, the server converts the data into an appropriate format using natural language processing technology and image processing software (e.g., OpenCV).

[1023] Step 3:

[1024] The server selects and initializes a generative AI model using the preprocessed data. The input is the preprocessed data, and the output is the initialized generative AI model. Specifically, the server loads the model using a generative AI model library (e.g., TensorFlow or PyTorch) and feeds it training data.

[1025] Step 4:

[1026] The server inputs preprocessed data into the generative AI model to automatically generate multiple ad patterns. The input is the initialized generative AI model and preprocessed data, and the output is multiple ad patterns. Specifically, it uses the model's inference function to generate ad text, images, and audio.

[1027] Step 5:

[1028] The terminal provides the generated advertisement pattern to the user and receives reviews and feedback. The input is the advertisement pattern from the server, and the output is the user's review and feedback. Specific operations include displaying the generated advertisement through a user interface and collecting ratings and comments.

[1029] Step 6:

[1030] The device collects the user's emotional data in real time and analyzes it using an emotion engine. The input is the user's facial expressions and vocal tone, and the output is the analyzed emotional data. Specifically, it uses a camera and microphone and analyzes the data using emotion recognition software (e.g., Affectiva SDK).

[1031] Step 7:

[1032] The server then makes final adjustments to the ad patterns based on the collected user emotion data and feedback. The input is the analyzed emotion data and feedback, and the output is the adjusted ad pattern. Specific operations include correcting text and images to improve the quality of the ad.

[1033] Step 8:

[1034] The server then encodes the final adjusted ad template in a high-resolution format, optimizing it for various media. The input is the adjusted ad template, and the output is a high-resolution ad file. Specifically, the server uses video encoding software (e.g., FFmpeg) to convert the data into the appropriate format.

[1035] Step 9:

[1036] Finally, users can view the optimized advertisements in the virtual store, develop interest in the products, and make a purchase. The input is a high-resolution advertisement file, and the output is the user's purchasing behavior. Specifically, users view the advertisements using a smartphone or head-mounted display and click the purchase button.

[1037] 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.

[1038] 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.

[1039] 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.

[1040] [Third embodiment]

[1041] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1042] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[1043] 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).

[1044] 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.

[1045] 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.

[1046] 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).

[1047] 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.

[1048] 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.

[1049] 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.

[1050] 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.

[1051] 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.

[1052] 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."

[1053] MODE FOR CARRYING OUT THE INVENTION

[1054] The following is a specific method for implementing a system for automatically generating advertisements using a generative AI model according to the present invention. The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting advertisements.

[1055] Data collection and preparation

[1056] 1. Data Collection

[1057] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[1058] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[1059] 2. Data Preprocessing

[1060] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model, which includes cleaning up text data, processing image data, and formatting audio files.

[1061] Preparing the generative AI model

[1062] 3. Model Selection and Initialization

[1063] The server analyzes the type and content of the provided data and selects the most appropriate generative AI model (text generation, image generation, speech generation, etc.). After selecting the model, it loads the necessary training data and initializes the model.

[1064] Ad generation

[1065] 4. Ad generation instructions

[1066] The server inputs the preprocessed data into the generative AI model, which then automatically generates multiple ad patterns, such as energetic ads, ads that emphasize technical aspects, and ads that show actual usage scenarios.

[1067] Reviews and Feedback

[1068] 5. Providing Reviews

[1069] The server generates a review interface for providing the generated plurality of advertisement patterns to a user, and the user reviews the advertisement patterns through the provided interface and inputs feedback for each advertisement.

[1070] 6. Gathering Feedback

[1071] The server collects and stores user feedback for use in the next ad generation task, providing specific suggestions for improvement, such as making the music faster or the message more concise.

[1072] Final ad adjustment and output

[1073] 7. Applying Feedback

[1074] The server selects the best ad pattern based on user feedback and makes necessary adjustments, such as changing the tempo of the voice announcement and modifying some of the text.

[1075] 8. Final encoding and output

[1076] The server encodes the tailored ad variations into high-definition formats and optimizes them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1077] Specific examples

[1078] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[1079] 1. Data Collection

[1080] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1081] 2. Data Preprocessing

[1082] The server processes the collected data and converts it into a format suitable for the generative AI model.

[1083] 3. Model Selection and Initialization

[1084] The server selects text generation and image generation models suitable for generating ads for smartwatches and loads the necessary training data.

[1085] 4. Ad generation instructions

[1086] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1087] 5. Providing Reviews

[1088] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[1089] 6. Gathering Feedback

[1090] The user requests that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster as feedback, and the server collects this.

[1091] 7. Applying Feedback

[1092] The server adjusts the "health monitoring emphasis advertisement" based on user feedback, speeding up the tempo of the voice announcements.

[1093] 8. Final encoding and output

[1094] The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[1095] Based on these steps, the system enables businesses to generate high-quality advertisements in an efficient and low-risk manner.

[1096] The processing flow will be explained below.

[1097] MODE FOR CARRYING OUT THE INVENTION

[1098] Program processing flow

[1099] Step 1:

[1100] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[1101] Step 2:

[1102] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[1103] Step 3:

[1104] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[1105] Step 4:

[1106] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[1107] Step 5:

[1108] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[1109] Step 6:

[1110] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[1111] Step 7:

[1112] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[1113] Step 8:

[1114] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[1115] Step 9:

[1116] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message shorter."

[1117] Step 10:

[1118] The server collects user feedback and adjusts the advertising pattern based on that feedback, for example speeding up the voice announcements or shortening the text.

[1119] Step 11:

[1120] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[1121] Step 12:

[1122] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[1123] In this way, generative AI can be used to generate high-quality ads in an efficient and low-risk manner.

[1124] Example 1

[1125] 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."

[1126] Ad production is a time-consuming and costly process, especially when multiple ad variations need to be generated for different target audiences. It's also challenging to quickly incorporate feedback and deliver optimal ads while maintaining ad quality, making it difficult to maximize the effectiveness of advertising campaigns.

[1127] 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.

[1128] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for analyzing and preprocessing the collected advertising data, means for selecting and initializing an optimal generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, and means for final adjustment and output of the advertising patterns based on user feedback. This reduces the time and cost of advertising production, makes it possible to quickly and effectively generate multiple advertising patterns, and provide optimal advertisements by reflecting feedback.

[1129] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to automatically generate advertising content from a variety of data, including text, images, and audio.

[1130] "Advertising data" is a general term for information necessary for generating advertisements, such as product information provided by companies, target demographics, appealing points, and past advertising examples.

[1131] "Input data" refers to advertising data that is input into the generative AI model, and is provided to the generative AI model after initial data collection and preprocessing.

[1132] "Preprocessing" refers to the process of data processing and cleaning to convert advertising data into a format suitable for generative AI models.

[1133] "Advertising patterns" is a collective term for multiple advertising formats automatically generated by a generative AI model.

[1134] "Review" is an operation in which a user checks and evaluates the generated advertisement pattern.

[1135] "Feedback" refers to opinions and suggestions for improvement that a user provides regarding the generated advertising pattern.

[1136] "Final adjustment" refers to the process of optimizing the content and format of the advertising pattern by reflecting user feedback.

[1137] "Output" refers to the process of encoding the final adjusted advertisement pattern to generate the final advertisement file.

[1138] A "promotion content creation request" is a request sent by a user to a server to request the creation of new advertising or marketing materials.

[1139] "Quality evaluation" is the process of evaluating the effectiveness and suitability of advertising patterns output by a generative AI model.

[1140] "High definition format" refers to a file format that allows the generated advertisement to be presented in high visual and audio resolution.

[1141] "Media optimization" is the process of converting the generated advertising patterns into formats suitable for different distribution media such as the Internet, television, and radio.

[1142] This invention is a system for automatically generating advertisements using a generative AI model. The main functions of the system consist of the following steps: collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting the advertisements.

[1143] Data collection and preparation

[1144] Data collection

[1145] The server collects advertising data provided by companies, such as product information, target demographics, key selling points, and past advertising examples. This data is structured and stored in an advertising database. Data is collected through APIs and database connections and converted into the format required to generate ads.

[1146] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[1147] Data Preprocessing

[1148] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model. This includes cleaning up text data, processing image data, and formatting audio files. For text data, unnecessary spaces and special characters are removed and natural language processing tools are used to segment the data. For image data, image resizing and formatting are performed, and for audio data, noise reduction and encoding are performed.

[1149] Preparing the generative AI model

[1150] Model Selection and Initialization

[1151] The server analyzes the provided data and selects the optimal generative AI model (text generation, image generation, speech generation, etc.). Specifically, it uses GPT-3 as the text generation model, DALL-E as the image generation model, and WaveNet as the speech generation model. After the selection, it loads the necessary training data and initializes the model.

[1152] Ad generation

[1153] Ad generation instructions

[1154] The server inputs the preprocessed data into the generative AI model to automatically generate multiple ad patterns, and the generated results are recorded in a log to generate multiple different ad patterns (such as energetic ads, technology-focused ads, and ads showing usage scenarios).

[1155] Reviews and Feedback

[1156] Providing a review

[1157] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface for displaying the generated advertisements in thumbnail and playback format and allowing the user to evaluate them.

[1158] Collecting feedback

[1159] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[1160] Final ad adjustment and output

[1161] Applying Feedback

[1162] The server adjusts the advertising patterns based on user feedback, specifically by readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[1163] Final encoding and output

[1164] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1165] Specific examples

[1166] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[1167] 1. Data Collection

[1168] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1169] 2. Data Preprocessing

[1170] The server preprocesses the collected data using natural language processing tools and image processing tools.

[1171] 3. Model Selection and Initialization

[1172] Based on the collected data, the server selects a generative AI model such as GPT-3, DALL-E, or WaveNet and loads the necessary training data.

[1173] 4. Ad generation instructions

[1174] The server inputs the preprocessed data into the model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1175] 5. Providing Reviews

[1176] A user uses a web interface to review and rate the generated ad variations.

[1177] 6. Gathering Feedback

[1178] The user may provide feedback requesting that the tempo of the audio for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback.

[1179] 7. Applying Feedback

[1180] The server adjusts the tempo of the audio and regenerates it based on the feedback.

[1181] 8. Final encoding and output

[1182] The server encodes the tailored advertisement into high definition format and provides the final version to the user via a secure download link.

[1183] Based on the above steps, the system enables companies to generate high-quality advertisements efficiently and with low risk.

[1184] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1185] Step 1: Data collection

[1186] The server collects data provided by companies, such as product information, target demographics, selling points, and past advertising examples. This data is retrieved through APIs and database connections, structured, and stored in an advertising database.

[1187] Input: Promotional content creation requests from companies, product information, target demographic data

[1188] Data processing: Acquired through APIs and database connections, stored in a structured database

[1189] Output: Structured data stored in the ad database

[1190] Step 2: Preprocessing the data

[1191] The server analyzes data collected from the advertising database and converts it into a format suitable for the generative AI model. Text data is stripped of unnecessary spaces and special characters and segmented using natural language processing tools. Image data is resized and formatted, and audio data is noise-reduced and encoded.

[1192] Input: Structured advertising data in an advertising database

[1193] Data processing: cleaning up text data, resizing and formatting images, noise reduction and encoding of audio data

[1194] Output: Preprocessed data (cleaned text, enhanced images, formatted audio)

[1195] Step 3: Model selection and initialization

[1196] The server selects the optimal generative AI model based on the preprocessed data. Specifically, it selects GPT-3 for text generation, DALL-E for image generation, and WaveNet for speech generation. After the selection, it loads the necessary training data and initializes the model.

[1197] Input: Preprocessed data

[1198] Data processing: Analyzing data content, selecting the optimal model, and loading training data

[1199] Output: Initialized generative AI model

[1200] Step 4: Ad generation instructions

[1201] The server inputs the preprocessed data into a generative AI model to automatically generate multiple ad patterns, which are then logged and used to generate different patterns, such as energetic ads, technology-focused ads, and ads showing usage scenarios.

[1202] Input: Initialized generative AI model and preprocessed data

[1203] Data processing: Generating advertising patterns using generative AI models

[1204] Output: Multiple ad variations generated

[1205] Step 5: Provide a review

[1206] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface is used to display the generated advertisements in thumbnail and playback format and allow the user to evaluate them.

[1207] Input: Generated ad patterns

[1208] Data processing: Generate review interface and display advertising patterns

[1209] Output: The review interface presented to the user

[1210] Step 6: Gather feedback

[1211] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[1212] Input: User feedback

[1213] Data processing: collecting and storing feedback

[1214] Output: Saved feedback data

[1215] Step 7: Applying feedback

[1216] The server adjusts the ad patterns based on user feedback, readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[1217] Input: Saved feedback data

[1218] Data processing: Readjust model parameters based on feedback and re-run ad generation

[1219] Output: Adjusted ad patterns

[1220] Step 8: Final encoding and output

[1221] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1222] Input: Adjusted ad pattern

[1223] Data processing: Encoding to high definition formats and optimizing for media

[1224] Output: Final ad file provided with a secure download link

[1225] (Application example 1)

[1226] 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."

[1227] Modern advertising production requires a significant amount of time and effort, making it difficult to generate effective ads, especially for targeted audiences. Furthermore, systems for quickly incorporating user feedback are inadequate, and improvements are needed to improve the quality of advertising. Furthermore, there is a lack of ad generation systems that utilize mobile devices such as smartphones, creating a demand for a more flexible and rapid advertising production process.

[1228] 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.

[1229] In this invention, the server includes: means for collecting advertising data as input data using a generative AI model; means for selecting and initializing a generative AI model based on the collected advertising data; means for automatically generating multiple advertising patterns using the generative AI model; means for providing the generated advertising patterns to users and receiving reviews and feedback; means for making final adjustments to and outputting the advertising patterns based on user feedback; means for providing an interface using a smartphone; means for users to input product information and send the data to the server; and means for analyzing the provided data and converting it into a format suitable for the generative AI model. This enables users to easily collect and input advertising data using a smartphone, quickly generate high-quality advertisements using an optimal generative AI model based on the data, and regenerate advertisements that reflect user feedback.

[1230] A "generative AI model" is an algorithm that uses machine learning to automatically analyze data and perform a specific task, such as generating an ad.

[1231] "Advertising data" refers to information such as product information, target demographics, and appealing points that are necessary when creating an advertisement.

[1232] A "server" is a computing system that is responsible for collecting, analyzing, generating, storing, and transmitting data.

[1233] "Input data" refers to data such as product information, target demographics, and past advertising examples provided to generate advertisements.

[1234] "Ad Variants" refers to multiple different variations of an advertisement automatically generated by a generative AI model.

[1235] "User" refers to any person or entity that utilizes the Ad Generation System to generate, review, and provide feedback on Ads.

[1236] "Review interface" refers to a user interface that allows a user to review the generated advertising patterns and provide feedback.

[1237] "Feedback" refers to improvements and evaluations provided by users regarding the generated advertising patterns.

[1238] "Smartphone" means a mobile device that allows access to the ad generation system and enables data entry, review and feedback.

[1239] "Interface" refers to the means by which a user interacts with a system.

[1240] "Data analysis" refers to the process of converting collected data into useful information and forms.

[1241] "High-resolution format" refers to a format for exporting high-quality advertisements to a variety of media formats.

[1242] "Encoding" refers to the process of converting a generated advertisement into a particular format for storage or delivery.

[1243] MODE FOR CARRYING OUT THE INVENTION

[1244] This invention relates to a system for automatically generating advertisements by collecting advertising data using a generative AI model. This system provides a smartphone interface, and when a user inputs product information, it generates multiple advertising patterns and adjusts them based on the user's feedback to output the optimal advertisement.

[1245] System configuration

[1246] The system mainly consists of the following components:

[1247] 1. Server: Responsible for collecting, analyzing, generating, storing, and transmitting data.

[1248] 2. Generative AI model: An algorithm that automatically generates ads based on collected data.

[1249] 3. Smartphone: A device that allows users to enter advertising data and provide reviews and feedback.

[1250] Hardware and software used

[1251] Hardware: Smartphones, cloud servers (e.g., AWS)

[1252] software:

[1253] Mobile frontend (e.g., Swift, Kotlin)

[1254] Backend (e.g. Node.js, Django)

[1255] Data analysis tools (Python, Pandas)

[1256] Image processing library (OpenCV)

[1257] Generative AI models (OpenAI GPT-4, TensorFlow, PyTorch)

[1258] API frameworks (e.g. FastAPI)

[1259] Cloud storage (e.g. AWS S3)

[1260] Streaming encoding tool (e.g. FFmpeg)

[1261] System Operation

[1262] 1. Data collection: Users use a smartphone application to enter advertising data such as product information, target demographics, and selling points, and send it to the server.

[1263] 2. Data analysis: The server analyzes the received data and converts it into a format suitable for the generative AI model, such as cleaning up text data and resizing image data.

[1264] 3. Ad generation: The server inputs the analyzed data into a generative AI model to automatically generate multiple ad patterns.

[1265] 4. Review and Feedback: The generated ad variants are provided to a review interface on the smartphone, where users can review them and provide feedback.

[1266] 5. Applying feedback: The server regenerates and adjusts the advertising pattern based on user feedback and outputs the optimal advertisement.

[1267] 6. Encoding into high-definition formats: The adjusted advertisements are encoded into high-definition formats and optimized for various media.

[1268] Specific examples

[1269] For example, if a company requests an ad for a new smartwatch, the process might go something like this:

[1270] 1. Data collection: The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1271] 2. Data pre-processing: The server processes the collected data and converts it into a format suitable for the generative AI model.

[1272] 3. Model Selection and Initialization: The server selects suitable text generation and image generation models for smartwatch ad generation and loads the necessary training data.

[1273] 4. Instructions for generating advertisements: The server inputs the preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1274] 5. Providing reviews: The server provides the generated advertising patterns to the user through a review interface, where the user can check and rate them.

[1275] 6. Feedback collection: The user requests the server to speed up the voice announcement tempo of the “health monitoring-emphasizing advertisement” as feedback, and the server collects this.

[1276] 7. Applying feedback: The server adjusts the "health monitoring emphasis advertisement" based on the user's feedback and speeds up the tempo of the voice announcement.

[1277] 8. Final encoding and output: The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[1278] Examples of prompt statements

[1279] Below are some examples of prompts to input to the generative AI model.

[1280] Prompt statement:

[1281] text

[1282] Product name: XYZ Smart Watch

[1283] Target demographic: Technology enthusiasts in their 20s and 30s

[1284] Features:

[1285] Health Monitoring Features

[1286] Stylish design

[1287] Long battery life

[1288] Selling points:

[1289] Ideal for health management

[1290] Stylish design using the latest technology

[1291] Lasts for a week on a single charge

[1292] Based on this information, generate three ad variations (emphasizing health monitoring, design, and usage scenarios, respectively).

[1293] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1294] Step 1:

[1295] User enters product information

[1296] The user uses a smartphone application to input product information (e.g., product name, features, target demographic, selling points). The input data is sent to the server. Input: Product information data. Output: Data sent to the server.

[1297] Step 2:

[1298] The server collects the data

[1299] The server receives product information data sent by the user and stores it in a database. It prepares the data for analysis. Input: Product information data sent by the user. Output: Product information data stored in the database.

[1300] Step 3:

[1301] The server analyzes the data

[1302] The server analyzes the collected product information data, cleans up the text data (e.g., removes unnecessary spaces and special characters), resizes the image data, and converts it into a format suitable for the generative AI model. Input: Product information data stored in the database. Output: Data in a format suitable for the generative AI model.

[1303] Step 4:

[1304] Selecting and initializing a generative AI model

[1305] The server selects the optimal generative AI model depending on the type of data, loads the necessary training data, and initializes the model. Input: Data in a format suitable for the generative AI model. Output: The initialized generative AI model.

[1306] Step 5:

[1307] Generating advertising variations

[1308] The server supplies input data to the generative AI model to generate multiple ad patterns. For example, ads that emphasize health monitoring, stylish design, or usage scenarios can be generated. Input: Initialized generative AI model and input data. Output: Multiple ad patterns.

[1309] Step 6:

[1310] User reviews ad variations

[1311] The server provides the generated ad patterns to a review interface on the smartphone, where the user can review them. Input: Multiple ad patterns. Output: User reviews and feedback.

[1312] Step 7:

[1313] Users provide feedback

[1314] The user inputs feedback for the advertisement pattern through the review interface, for example, requesting that the tempo of the voice announcement for the "advertisement emphasizing health monitoring" be made faster. Input: Feedback for the advertisement pattern. Output: Feedback sent to the server.

[1315] Step 8:

[1316] The server reflects the feedback

[1317] The server analyzes the feedback from the user and regenerates / adjusts the ad pattern based on it, for example by speeding up the tempo of the voice announcement. Input: Feedback sent to the server. Output: Adjusted ad pattern.

[1318] Step 9:

[1319] Encoding and Output

[1320] The server encodes the adjusted ad variants into high-resolution formats, optimizing them for various media. It generates the final ad file and provides a secure download link to the user. Input: Adjusted ad variant. Output: Final ad file and download link.

[1321] 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.

[1322] MODE FOR CARRYING OUT THE INVENTION

[1323] In accordance with this invention, a specific implementation method of a system for automatically generating advertisements by combining a generative AI model and an emotion engine is as follows: The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, adjusting and outputting the final advertisements, and recognizing and reflecting user emotions.

[1324] Data collection and preparation

[1325] 1. Data Collection

[1326] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[1327] For example, if a user requests an advertisement for a new smartwatch, information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness will be provided and collected by the server.

[1328] 2. Data Preprocessing

[1329] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model and emotion engine, which includes cleaning up text data, processing image data, and formatting audio data.

[1330] Preparing the generative AI model

[1331] 3. Model Selection and Initialization

[1332] The server analyzes the provided advertising data and its contents and selects an appropriate generative AI model and emotion engine. The generative AI model includes text generation, image generation, and voice generation, and the emotion engine is used to recognize the user's emotional state.

[1333] The server initializes these models and loads the necessary training data.

[1334] Ad generation

[1335] 4. Ad generation instructions

[1336] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, multiple advertisement patterns (such as advertisements emphasizing health monitoring functions or stylish designs) are generated.

[1337] Reviews and Feedback

[1338] 5. Providing Reviews

[1339] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each of them.

[1340] 6. Emotion recognition

[1341] The server uses an emotion engine to detect the user's emotional state, for example by analyzing facial expressions and voice tone to recognize emotions while the user is reviewing an advertisement.

[1342] 7. Emotion-based evaluation

[1343] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[1344] 8. Feedback and Emotion Data Storage

[1345] The feedback provided by the user and the emotional data at that time are stored on the server. For example, if a user likes an advertisement that emphasizes health monitoring, the positive emotional feedback is recorded.

[1346] Final ad adjustment and output

[1347] 9. Applying Feedback

[1348] The server then optimally adjusts the advertising pattern based on user feedback and emotional data, for example by changing the tempo of the voice announcement, modifying the text, or changing some of the visuals.

[1349] 10. Encoding and Serving Ads

[1350] The server then encodes the tailored ad variations into high-definition formats, optimizing them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1351] Specific examples

[1352] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[1353] 1. Data Collection

[1354] The server collects data on the smartwatch's functions, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1355] 2. Data Preprocessing

[1356] The server processes the collected data and converts it into a format suitable for generative AI models and emotion engines.

[1357] 3. Model Selection and Initialization

[1358] The server selects text generation and image generation models and emotion engines suitable for generating ads for smartwatches and loads the necessary training data.

[1359] 4. Ad generation instructions

[1360] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1361] 5. Providing Reviews

[1362] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[1363] 6. Emotion recognition

[1364] The server analyzes the user's emotions during the review using an emotion engine and collects the user's reactions for each advertising pattern.

[1365] 7. Gathering Feedback

[1366] As feedback, users request that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback while also recording the user's positive emotional data.

[1367] 8. Applying Feedback

[1368] The server adjusts the advertising pattern and speeds up the tempo of the voice announcements based on the user's feedback and emotional data.

[1369] 9. Encoding and Serving Ads

[1370] The server encodes the tailored advertisement into a high-resolution format and provides the final version to the user via a secure download link.

[1371] Based on the above steps, this system enables companies to generate high-quality advertisements in an efficient and low-risk manner, while also enabling advertisements that reflect user emotions.

[1372] The processing flow will be explained below.

[1373] MODE FOR CARRYING OUT THE INVENTION

[1374] Program processing flow

[1375] Step 1:

[1376] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[1377] Step 2:

[1378] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[1379] Step 3:

[1380] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[1381] Step 4:

[1382] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[1383] Step 5:

[1384] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[1385] Step 6:

[1386] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[1387] Step 7:

[1388] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[1389] Step 8:

[1390] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[1391] Step 9:

[1392] The server uses an emotion engine to analyze the user's emotional state during the review, for example, by evaluating the user's facial expressions and tone of voice in real time to recognize the user's emotions.

[1393] Step 10:

[1394] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[1395] Step 11:

[1396] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message more concise."

[1397] Step 12:

[1398] The server adjusts the advertising pattern based on user feedback and analyzed emotion data, for example, by changing the tempo of the voice announcement and modifying some of the text.

[1399] Step 13:

[1400] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[1401] Step 14:

[1402] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[1403] In this way, generative AI can generate high-quality ads in an efficient and low-risk manner. In addition, by combining it with an emotion engine, it becomes possible to create ads that reflect user emotions, resulting in highly appealing ads.

[1404] Example 2

[1405] 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."

[1406] Conventional ad generation systems have difficulty creating ads that are optimized for the target audience because they do not adequately adjust ads to reflect specific user feedback and emotions. Furthermore, there is no established method for integrating emotion recognition technology into the ad generation process to effectively reflect users' positive reactions.

[1407] 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 advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for using an emotion engine to recognize the user's emotional state, and means for final adjustment and output of the advertising patterns based on the user's feedback and emotion data. This enables the generation of optimized advertisements that reflect user emotions in real time.

[1408] A "generative AI model" is an algorithm that uses artificial intelligence technology to automatically generate advertising content such as text, images, and audio.

[1409] "Advertising data" refers to input data necessary to generate advertisements, such as product information, target demographics, appealing points, and past advertising examples.

[1410] An "emotion engine" is a software component for detecting and analyzing a user's emotional state, specifically analyzing data such as facial expressions and vocal tone.

[1411] "Feedback" refers to specific evaluations and opinions provided by users regarding the generated advertising patterns, as well as instructions for improvement.

[1412] "Ad Variants" are different variations or formats of an advertisement automatically generated by a generative AI model.

[1413] The "review interface" is an interface that allows a user to check the generated advertisement pattern and input feedback.

[1414] "Initialization" is the process of loading the settings and training data required to run a generative AI model.

[1415] "Encoding" is the process of outputting the generated advertising content in a high-resolution format or in a format optimized for a particular medium.

[1416] "Data collection" is the process of obtaining, structuring, and storing advertising data needed to generate ads from companies and other sources.

[1417] "Adjustment" refers to the process of optimizing advertising patterns based on user feedback and sentiment data, including text correction, audio tempo adjustment, and visual changes.

[1418] This invention is a system that automatically generates advertisements by combining a generative AI model and an emotion engine. Detailed embodiments are described below. The invention is composed of various means and processes centered on a server, a terminal, and a user.

[1419] Data collection and preparation

[1420] The server collects advertising data provided by companies, such as product information, target demographics, selling points, and past advertising examples. The collected data is stored in an advertising database and classified as text data, image data, and audio data. This data is then converted into the appropriate format for input into the generative AI model and emotion engine. Unnecessary information is removed from the text data and the formatting is adjusted. Image data is resized and its resolution is adjusted, and audio data is subjected to noise removal and sound quality adjustment.

[1421] For example, if a user requests an advertisement for a new smartwatch, the server collects information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness, and stores it in a database.

[1422] Preparing the generative AI model

[1423] The server selects the appropriate generative AI model (e.g., GPT-3 for text generation, DALL-E for image generation) and emotion engine based on the collected ad data. These models are initialized and loaded with the necessary training data to apply them to the specific ad generation task.

[1424] Ad generation

[1425] The server then inputs the pre-processed data into the selected generative AI model and sends a prompt to generate an ad, such as:

[1426] "Generate an ad for a smartwatch targeted at men in their 20s. Features include health monitoring, stylish design, and fitness features."

[1427] As a result, the generative AI model automatically generates multiple advertising patterns, such as ads that emphasize health monitoring, ads that emphasize design, and ads that show usage scenarios.

[1428] Reviews and Feedback

[1429] The server provides the generated ad template to the user through a review interface that can be viewed on the device. The user reviews the ad template and enters feedback. For example, the user can review the text, images, and audio of the ad through a smartphone or PC interface and enter specific feedback.

[1430] The server uses an emotion engine to analyze the user's facial expressions and tone of voice while reviewing the ad, for example, by using a webcam and microphone to collect the user's real-time reactions and recognize the user's emotional state based on the data obtained.

[1431] Final ad adjustment and output

[1432] The server then adjusts the ad template based on the collected feedback and sentiment data, modifying the text, adjusting the tempo of the voice announcement, and changing the visuals. The final ad template is then encoded in high-resolution format and output in a format optimized for each medium. The final ad file is then provided to the user via a secure download link.

[1433] For example, if an "advertisement emphasizing health monitoring" is adjusted to have a faster audio tempo, the adjusted advertisement will be output as the final version.

[1434] The above is a specific embodiment for carrying out the present invention, which makes it possible to generate an optimized advertisement that reflects the user's emotions in real time.

[1435] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1436] Step 1:

[1437] Data collection

[1438] The server collects advertising data such as product information, target demographics, appealing points, and past advertising examples provided by companies.

[1439] The server receives the data through an API or a management portal.

[1440] The server stores the received data in an advertisement database.

[1441] Input: Provided product information, target demographic, selling points, past advertising examples

[1442] Output: Structured data in an advertising database

[1443] Step 2:

[1444] Data preprocessing

[1445] The server analyzes and cleans the raw data collected.

[1446] The server cleans the text data, removing unnecessary information and formatting.

[1447] The server resizes and adjusts the resolution of the image data.

[1448] The server performs noise removal and sound quality adjustment on the audio data.

[1449] Input: Raw data stored in the advertising database

[1450] Output: Data converted into a format suitable for generative AI models and emotion engines

[1451] Step 3:

[1452] Model Selection and Initialization

[1453] The server selects the appropriate generative AI model and emotion engine based on the collected advertising data. Specifically, it uses GPT-3 for text generation and DALL-E for image generation.

[1454] The server initializes the selected generative AI model and emotion engine and loads the training data.

[1455] Input: Data converted into a format suitable for generative AI models and emotion engines

[1456] Output: Initialized generative AI model and emotion engine

[1457] Step 4:

[1458] Ad generation instructions

[1459] The server inputs the preprocessed data into the generative AI model and sends prompt sentences.

[1460] For example, enter the prompt text, "Generate a smartwatch ad for men in their 20s. Features include health monitoring, stylish design, and fitness functions."

[1461] The server receives the generated advertisement pattern.

[1462] Input: prompts, data input by generative AI models

[1463] Output: Multiple ad variations

[1464] Step 5:

[1465] Providing a review

[1466] The server provides the generated advertisement patterns to the user through a review interface that can be viewed on a terminal.

[1467] The user reviews each ad variant and enters specific feedback.

[1468] Input: Generated ad patterns

[1469] Output: User feedback

[1470] Step 6:

[1471] emotion recognition

[1472] The server uses an emotion engine to analyze the user's facial expressions and tone of voice during the advertisement review.

[1473] The server uses a webcam and microphone to collect the user's real-time responses.

[1474] The server recognizes the user's emotional state based on the obtained data.

[1475] Input: Real-time data of the user while reviewing the ad (facial expressions, tone of voice)

[1476] Output: User's emotional state data

[1477] Step 7:

[1478] Emotion-based ratings

[1479] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[1480] The server sets a higher priority to the advertisement pattern that receives the most positive responses.

[1481] Input: User's emotional state data

[1482] Output: Priority of the evaluated ad variants

[1483] Step 8:

[1484] Feedback and sentiment data storage

[1485] The server stores the user feedback and emotion data in a database.

[1486] The server saves it for future use in the ad generation task.

[1487] Input: User feedback and sentiment data

[1488] Output: Stored feedback and emotion data

[1489] Step 9:

[1490] Applying Feedback

[1491] The server adjusts advertising patterns based on collected feedback and sentiment data.

[1492] The server modifies the text of the advertisement, adjusts the tempo of the voice announcement, changes the visuals, etc.

[1493] Input: User feedback and sentiment data

[1494] Output: Adjusted ad patterns

[1495] Step 10:

[1496] Encoding and Serving Ads

[1497] The server encodes the finalized ad variations in high-definition format.

[1498] The server outputs the advertisement in a format optimized for each medium.

[1499] The server provides the final ad file to the user via a secure download link.

[1500] Input: Adjusted ad pattern

[1501] Output: Final ad file in high resolution format

[1502] The above is the specific processing flow of the program for this system.

[1503] (Application example 2)

[1504] 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."

[1505] Conventional ad generation systems have difficulty taking into account user emotional data, making it difficult for the ads they provide to reflect the user's psychological state and preferences. Furthermore, there are limited means for effectively utilizing real-time user feedback, making it difficult to improve the quality of ads. There is a need for a system that can analyze user emotions in real time and reflect that data in ad generation to provide more effective and attractive ads.

[1506] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1507] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for collecting and analyzing user emotional data in real time, and means for making final adjustments to and outputting advertising patterns based on the user emotional data and feedback, thereby making it possible to instantly reflect the user's emotions and provide advertisements optimized for the user's emotional state.

[1508] A "generative AI model" is a model that uses artificial intelligence to generate advertising data such as text, images, and audio.

[1509] "Advertising data" refers to information for creating advertisements, including product information, target demographics, selling points, and past advertising examples.

[1510] "Emotion data" refers to data relating to the user's emotional state obtained by analyzing the user's facial expressions, voice tone, and other sensor information.

[1511] "Advertising variations" are multiple variations of advertisements automatically generated by a generative AI model.

[1512] "Reviews and feedback" refers to the evaluation and opinions provided by users regarding the provided advertising patterns.

[1513] "Final adjustment and output" refers to modifying the advertising pattern based on user feedback and emotional data and generating the final version.

[1514] A "virtual store" is a virtual shop that exists on the Internet and is a place where users can browse and purchase products.

[1515] "Means for collecting and analyzing in real time" refers to methods and technologies for instantly collecting user emotional data and analyzing it on the spot.

[1516] The "means for prioritizing" is a means for selecting the most suitable advertising pattern from among a plurality of advertising patterns based on the user's emotional data.

[1517] In this invention, we use a generative AI model and an emotion engine to build a system that provides advertisements that reflect customer emotions in a virtual store. The specific system configuration is as follows.

[1518] The server first collects advertising data provided by each company, such as product information, target demographics, selling points, and past advertising examples. This data is structured and stored in an advertising database. Devices such as smartphones and head-mounted displays collect users' purchasing history and preference tag information through apps. Based on this collected data, it is converted into a format suitable for the generative AI model. During this process, text data is cleaned up, image data is processed, and audio data is shaped.

[1519] The server uses this processed data to select and initialize a generative AI model suitable for the product. The generative AI models include a text generation model, an image generation model, and a voice generation model. The server uses these models to automatically generate multiple ad variations. The user then reviews these ads on a smartphone app and provides feedback.

[1520] During this review, the device collects the user's emotional data in real time and analyzes it using an emotion engine. For example, it uses a camera and microphone to analyze the user's facial expressions and tone of voice to obtain emotional data. The server then uses this emotional data and user feedback to make final adjustments to the advertising patterns. For example, if the user expresses positive emotions toward an advertisement emphasizing health monitoring, that advertisement will be displayed preferentially. The finalized advertisements are then encoded in high-resolution formats and optimized for various media.

[1521] As a concrete example, if a company requests the creation of an advertisement for a "new smartwatch," the following process takes place: The server collects and preprocesses data on the smartwatch's functional information, its target demographic of men in their 20s, and its appealing health monitoring function and stylish design. A generative AI model automatically generates patterns, such as advertisements that emphasize health monitoring, advertisements that emphasize design, and advertisements that show usage scenarios. Users view these advertisements on their smartphone app, and emotional data analyzed by an emotion engine is sent to the server. The server prioritizes and provides advertisement patterns that elicit a positive response to the user. Finally, these advertisements are encoded into high-resolution format and displayed in a virtual store.

[1522] An example of a prompt is:

[1523] "We collect data for product ID 12345, perform text and image preprocessing, and then use a generative AI model to generate product ads. Based on user sentiment data, we select the most appropriate ad pattern."

[1524] As described above, this system can instantly reflect the user's emotions and provide personalized advertisements.

[1525] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1526] Step 1:

[1527] The server collects advertising data such as product information, target demographics, selling points, and past advertising examples provided by each company. This collected data is structured and stored in an advertising database. The input is various advertising data provided by companies, and the output is a structured advertising database. Specifically, the server obtains data through APIs and data upload functions and stores it in the database.

[1528] Step 2:

[1529] The server performs preprocessing of the collected advertising data, such as cleaning up text data, processing image data, and formatting audio data. The input is data from the advertising database, and the output is preprocessed data. Specifically, the server converts the data into an appropriate format using natural language processing technology and image processing software (e.g., OpenCV).

[1530] Step 3:

[1531] The server selects and initializes a generative AI model using the preprocessed data. The input is the preprocessed data, and the output is the initialized generative AI model. Specifically, the server loads the model using a generative AI model library (e.g., TensorFlow or PyTorch) and feeds it training data.

[1532] Step 4:

[1533] The server inputs preprocessed data into the generative AI model to automatically generate multiple ad patterns. The input is the initialized generative AI model and preprocessed data, and the output is multiple ad patterns. Specifically, it uses the model's inference function to generate ad text, images, and audio.

[1534] Step 5:

[1535] The terminal provides the generated advertisement pattern to the user and receives reviews and feedback. The input is the advertisement pattern from the server, and the output is the user's review and feedback. Specific operations include displaying the generated advertisement through a user interface and collecting ratings and comments.

[1536] Step 6:

[1537] The device collects the user's emotional data in real time and analyzes it using an emotion engine. The input is the user's facial expressions and vocal tone, and the output is the analyzed emotional data. Specifically, it uses a camera and microphone and analyzes the data using emotion recognition software (e.g., Affectiva SDK).

[1538] Step 7:

[1539] The server then makes final adjustments to the ad patterns based on the collected user emotion data and feedback. The input is the analyzed emotion data and feedback, and the output is the adjusted ad pattern. Specific operations include correcting text and images to improve the quality of the ad.

[1540] Step 8:

[1541] The server then encodes the final adjusted ad template in a high-resolution format, optimizing it for various media. The input is the adjusted ad template, and the output is a high-resolution ad file. Specifically, the server uses video encoding software (e.g., FFmpeg) to convert the data into the appropriate format.

[1542] Step 9:

[1543] Finally, users can view the optimized advertisements in the virtual store, develop interest in the products, and make a purchase. The input is a high-resolution advertisement file, and the output is the user's purchasing behavior. Specifically, users view the advertisements using a smartphone or head-mounted display and click the purchase button.

[1544] 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.

[1545] 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.

[1546] 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.

[1547] [Fourth embodiment]

[1548] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1549] 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.

[1550] 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).

[1551] 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.

[1552] 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.

[1553] 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).

[1554] 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.

[1555] 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.

[1556] 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.

[1557] 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.

[1558] 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.

[1559] 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.

[1560] 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."

[1561] MODE FOR CARRYING OUT THE INVENTION

[1562] The following is a specific method for implementing a system for automatically generating advertisements using a generative AI model according to the present invention. The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting advertisements.

[1563] Data collection and preparation

[1564] 1. Data Collection

[1565] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[1566] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[1567] 2. Data Preprocessing

[1568] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model, which includes cleaning up text data, processing image data, and formatting audio files.

[1569] Preparing the generative AI model

[1570] 3. Model Selection and Initialization

[1571] The server analyzes the type and content of the provided data and selects the most appropriate generative AI model (text generation, image generation, speech generation, etc.). After selecting the model, it loads the necessary training data and initializes the model.

[1572] Ad generation

[1573] 4. Ad generation instructions

[1574] The server inputs the preprocessed data into the generative AI model, which then automatically generates multiple ad patterns, such as energetic ads, ads that emphasize technical aspects, and ads that show actual usage scenarios.

[1575] Reviews and Feedback

[1576] 5. Providing Reviews

[1577] The server generates a review interface for providing the generated plurality of advertisement patterns to a user, and the user reviews the advertisement patterns through the provided interface and inputs feedback for each advertisement.

[1578] 6. Gathering Feedback

[1579] The server collects and stores user feedback for use in the next ad generation task, providing specific suggestions for improvement, such as making the music faster or the message more concise.

[1580] Final ad adjustment and output

[1581] 7. Applying Feedback

[1582] The server selects the best ad pattern based on user feedback and makes necessary adjustments, such as changing the tempo of the voice announcement and modifying some of the text.

[1583] 8. Final encoding and output

[1584] The server encodes the tailored ad variations into high-definition formats and optimizes them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1585] Specific examples

[1586] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[1587] 1. Data Collection

[1588] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1589] 2. Data Preprocessing

[1590] The server processes the collected data and converts it into a format suitable for the generative AI model.

[1591] 3. Model Selection and Initialization

[1592] The server selects text generation and image generation models suitable for generating ads for smartwatches and loads the necessary training data.

[1593] 4. Ad generation instructions

[1594] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1595] 5. Providing Reviews

[1596] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[1597] 6. Gathering Feedback

[1598] The user requests that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster as feedback, and the server collects this.

[1599] 7. Applying Feedback

[1600] The server adjusts the "health monitoring emphasis advertisement" based on user feedback, speeding up the tempo of the voice announcements.

[1601] 8. Final encoding and output

[1602] The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[1603] Based on these steps, the system enables businesses to generate high-quality advertisements in an efficient and low-risk manner.

[1604] The processing flow will be explained below.

[1605] MODE FOR CARRYING OUT THE INVENTION

[1606] Program processing flow

[1607] Step 1:

[1608] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[1609] Step 2:

[1610] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[1611] Step 3:

[1612] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[1613] Step 4:

[1614] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[1615] Step 5:

[1616] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[1617] Step 6:

[1618] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[1619] Step 7:

[1620] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[1621] Step 8:

[1622] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[1623] Step 9:

[1624] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message shorter."

[1625] Step 10:

[1626] The server collects user feedback and adjusts the advertising pattern based on that feedback, for example speeding up the voice announcements or shortening the text.

[1627] Step 11:

[1628] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[1629] Step 12:

[1630] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[1631] In this way, generative AI can be used to generate high-quality ads in an efficient and low-risk manner.

[1632] Example 1

[1633] 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."

[1634] Ad production is a time-consuming and costly process, especially when multiple ad variations need to be generated for different target audiences. It's also challenging to quickly incorporate feedback and deliver optimal ads while maintaining ad quality, making it difficult to maximize the effectiveness of advertising campaigns.

[1635] 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.

[1636] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for analyzing and preprocessing the collected advertising data, means for selecting and initializing an optimal generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, and means for final adjustment and output of the advertising patterns based on user feedback. This reduces the time and cost of advertising production, makes it possible to quickly and effectively generate multiple advertising patterns, and provide optimal advertisements by reflecting feedback.

[1637] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to automatically generate advertising content from a variety of data, including text, images, and audio.

[1638] "Advertising data" is a general term for information necessary for generating advertisements, such as product information provided by companies, target demographics, appealing points, and past advertising examples.

[1639] "Input data" refers to advertising data that is input into the generative AI model, and is provided to the generative AI model after initial data collection and preprocessing.

[1640] "Preprocessing" refers to the process of data processing and cleaning to convert advertising data into a format suitable for generative AI models.

[1641] "Advertising patterns" is a collective term for multiple advertising formats automatically generated by a generative AI model.

[1642] "Review" is an operation in which a user checks and evaluates the generated advertisement pattern.

[1643] "Feedback" refers to opinions and suggestions for improvement that a user provides regarding the generated advertising pattern.

[1644] "Final adjustment" refers to the process of optimizing the content and format of the advertising pattern by reflecting user feedback.

[1645] "Output" refers to the process of encoding the final adjusted advertisement pattern to generate the final advertisement file.

[1646] A "promotion content creation request" is a request sent by a user to a server to request the creation of new advertising or marketing materials.

[1647] "Quality evaluation" is the process of evaluating the effectiveness and suitability of advertising patterns output by a generative AI model.

[1648] "High definition format" refers to a file format that allows the generated advertisement to be presented in high visual and audio resolution.

[1649] "Media optimization" is the process of converting the generated advertising patterns into formats suitable for different distribution media such as the Internet, television, and radio.

[1650] This invention is a system for automatically generating advertisements using a generative AI model. The main functions of the system consist of the following steps: collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, and finally adjusting and outputting the advertisements.

[1651] Data collection and preparation

[1652] Data collection

[1653] The server collects advertising data provided by companies, such as product information, target demographics, key selling points, and past advertising examples. This data is structured and stored in an advertising database. Data is collected through APIs and database connections and converted into the format required to generate ads.

[1654] For example, if a user requests an advertisement for new sports shoes, information such as the characteristics of the sports shoes, the target demographic of men in their twenties, and their interest in health and fitness is provided and collected by the server.

[1655] Data Preprocessing

[1656] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model. This includes cleaning up text data, processing image data, and formatting audio files. For text data, unnecessary spaces and special characters are removed and natural language processing tools are used to segment the data. For image data, image resizing and formatting are performed, and for audio data, noise reduction and encoding are performed.

[1657] Preparing the generative AI model

[1658] Model Selection and Initialization

[1659] The server analyzes the provided data and selects the optimal generative AI model (text generation, image generation, speech generation, etc.). Specifically, it uses GPT-3 as the text generation model, DALL-E as the image generation model, and WaveNet as the speech generation model. After the selection, it loads the necessary training data and initializes the model.

[1660] Ad generation

[1661] Ad generation instructions

[1662] The server inputs the preprocessed data into the generative AI model to automatically generate multiple ad patterns, and the generated results are recorded in a log to generate multiple different ad patterns (such as energetic ads, technology-focused ads, and ads showing usage scenarios).

[1663] Reviews and Feedback

[1664] Providing a review

[1665] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface for displaying the generated advertisements in thumbnail and playback format and allowing the user to evaluate them.

[1666] Collecting feedback

[1667] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[1668] Final ad adjustment and output

[1669] Applying Feedback

[1670] The server adjusts the advertising patterns based on user feedback, specifically by readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[1671] Final encoding and output

[1672] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1673] Specific examples

[1674] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[1675] 1. Data Collection

[1676] The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1677] 2. Data Preprocessing

[1678] The server preprocesses the collected data using natural language processing tools and image processing tools.

[1679] 3. Model Selection and Initialization

[1680] Based on the collected data, the server selects a generative AI model such as GPT-3, DALL-E, or WaveNet and loads the necessary training data.

[1681] 4. Ad generation instructions

[1682] The server inputs the preprocessed data into the model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1683] 5. Providing Reviews

[1684] A user uses a web interface to review and rate the generated ad variations.

[1685] 6. Gathering Feedback

[1686] The user may provide feedback requesting that the tempo of the audio for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback.

[1687] 7. Applying Feedback

[1688] The server adjusts the tempo of the audio and regenerates it based on the feedback.

[1689] 8. Final encoding and output

[1690] The server encodes the tailored advertisement into high definition format and provides the final version to the user via a secure download link.

[1691] Based on the above steps, the system enables companies to generate high-quality advertisements efficiently and with low risk.

[1692] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1693] Step 1: Data collection

[1694] The server collects data provided by companies, such as product information, target demographics, selling points, and past advertising examples. This data is retrieved through APIs and database connections, structured, and stored in an advertising database.

[1695] Input: Promotional content creation requests from companies, product information, target demographic data

[1696] Data processing: Acquired through APIs and database connections, stored in a structured database

[1697] Output: Structured data stored in the ad database

[1698] Step 2: Preprocessing the data

[1699] The server analyzes data collected from the advertising database and converts it into a format suitable for the generative AI model. Text data is stripped of unnecessary spaces and special characters and segmented using natural language processing tools. Image data is resized and formatted, and audio data is noise-reduced and encoded.

[1700] Input: Structured advertising data in an advertising database

[1701] Data processing: cleaning up text data, resizing and formatting images, noise reduction and encoding of audio data

[1702] Output: Preprocessed data (cleaned text, enhanced images, formatted audio)

[1703] Step 3: Model selection and initialization

[1704] The server selects the optimal generative AI model based on the preprocessed data. Specifically, it selects GPT-3 for text generation, DALL-E for image generation, and WaveNet for speech generation. After the selection, it loads the necessary training data and initializes the model.

[1705] Input: Preprocessed data

[1706] Data processing: Analyzing data content, selecting the optimal model, and loading training data

[1707] Output: Initialized generative AI model

[1708] Step 4: Ad generation instructions

[1709] The server inputs the preprocessed data into a generative AI model to automatically generate multiple ad patterns, which are then logged and used to generate different patterns, such as energetic ads, technology-focused ads, and ads showing usage scenarios.

[1710] Input: Initialized generative AI model and preprocessed data

[1711] Data processing: Generating advertising patterns using generative AI models

[1712] Output: Multiple ad variations generated

[1713] Step 5: Provide a review

[1714] The server generates a review interface for providing the generated advertisement patterns to the user, specifically, a web interface is used to display the generated advertisements in thumbnail and playback format and allow the user to evaluate them.

[1715] Input: Generated ad patterns

[1716] Data processing: Generate review interface and display advertising patterns

[1717] Output: The review interface presented to the user

[1718] Step 6: Gather feedback

[1719] The server provides a form to collect feedback from users, allowing them to enter specific feedback such as "speed up the audio tempo" or "correct some of the text."

[1720] Input: User feedback

[1721] Data processing: collecting and storing feedback

[1722] Output: Saved feedback data

[1723] Step 7: Applying feedback

[1724] The server adjusts the ad patterns based on user feedback, readjusting the model parameters based on the collected feedback and rerunning the ad generation process.

[1725] Input: Saved feedback data

[1726] Data processing: Readjust model parameters based on feedback and re-run ad generation

[1727] Output: Adjusted ad patterns

[1728] Step 8: Final encoding and output

[1729] The server encodes the tailored ad into high-definition format and optimizes it for various media such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1730] Input: Adjusted ad pattern

[1731] Data processing: Encoding to high definition formats and optimizing for media

[1732] Output: Final ad file provided with a secure download link

[1733] (Application example 1)

[1734] 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."

[1735] Modern advertising production requires a significant amount of time and effort, making it difficult to generate effective ads, especially for targeted audiences. Furthermore, systems for quickly incorporating user feedback are inadequate, and improvements are needed to improve the quality of advertising. Furthermore, there is a lack of ad generation systems that utilize mobile devices such as smartphones, creating a demand for a more flexible and rapid advertising production process.

[1736] 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.

[1737] In this invention, the server includes: means for collecting advertising data as input data using a generative AI model; means for selecting and initializing a generative AI model based on the collected advertising data; means for automatically generating multiple advertising patterns using the generative AI model; means for providing the generated advertising patterns to users and receiving reviews and feedback; means for making final adjustments to and outputting the advertising patterns based on user feedback; means for providing an interface using a smartphone; means for users to input product information and send the data to the server; and means for analyzing the provided data and converting it into a format suitable for the generative AI model. This enables users to easily collect and input advertising data using a smartphone, quickly generate high-quality advertisements using an optimal generative AI model based on the data, and regenerate advertisements that reflect user feedback.

[1738] A "generative AI model" is an algorithm that uses machine learning to automatically analyze data and perform a specific task, such as generating an ad.

[1739] "Advertising data" refers to information such as product information, target demographics, and appealing points that are necessary when creating an advertisement.

[1740] A "server" is a computing system that is responsible for collecting, analyzing, generating, storing, and transmitting data.

[1741] "Input data" refers to data such as product information, target demographics, and past advertising examples provided to generate advertisements.

[1742] "Ad Variants" refers to multiple different variations of an advertisement automatically generated by a generative AI model.

[1743] "User" refers to any person or entity that utilizes the Ad Generation System to generate, review, and provide feedback on Ads.

[1744] "Review interface" refers to a user interface that allows a user to review the generated advertising patterns and provide feedback.

[1745] "Feedback" refers to improvements and evaluations provided by users regarding the generated advertising patterns.

[1746] "Smartphone" means a mobile device that allows access to the ad generation system and enables data entry, review and feedback.

[1747] "Interface" refers to the means by which a user interacts with a system.

[1748] "Data analysis" refers to the process of converting collected data into useful information and forms.

[1749] "High-resolution format" refers to a format for exporting high-quality advertisements to a variety of media formats.

[1750] "Encoding" refers to the process of converting a generated advertisement into a particular format for storage or delivery.

[1751] MODE FOR CARRYING OUT THE INVENTION

[1752] This invention relates to a system for automatically generating advertisements by collecting advertising data using a generative AI model. This system provides a smartphone interface, and when a user inputs product information, it generates multiple advertising patterns and adjusts them based on the user's feedback to output the optimal advertisement.

[1753] System configuration

[1754] The system mainly consists of the following components:

[1755] 1. Server: Responsible for collecting, analyzing, generating, storing, and transmitting data.

[1756] 2. Generative AI model: An algorithm that automatically generates ads based on collected data.

[1757] 3. Smartphone: A device that allows users to enter advertising data and provide reviews and feedback.

[1758] Hardware and software used

[1759] Hardware: Smartphones, cloud servers (e.g., AWS)

[1760] software:

[1761] Mobile frontend (e.g., Swift, Kotlin)

[1762] Backend (e.g. Node.js, Django)

[1763] Data analysis tools (Python, Pandas)

[1764] Image processing library (OpenCV)

[1765] Generative AI models (OpenAI GPT-4, TensorFlow, PyTorch)

[1766] API frameworks (e.g. FastAPI)

[1767] Cloud storage (e.g. AWS S3)

[1768] Streaming encoding tool (e.g. FFmpeg)

[1769] System Operation

[1770] 1. Data collection: Users use a smartphone application to enter advertising data such as product information, target demographics, and selling points, and send it to the server.

[1771] 2. Data analysis: The server analyzes the received data and converts it into a format suitable for the generative AI model, such as cleaning up text data and resizing image data.

[1772] 3. Ad generation: The server inputs the analyzed data into a generative AI model to automatically generate multiple ad patterns.

[1773] 4. Review and Feedback: The generated ad variants are provided to a review interface on the smartphone, where users can review them and provide feedback.

[1774] 5. Applying feedback: The server regenerates and adjusts the advertising pattern based on user feedback and outputs the optimal advertisement.

[1775] 6. Encoding into high-definition formats: The adjusted advertisements are encoded into high-definition formats and optimized for various media.

[1776] Specific examples

[1777] For example, if a company requests an ad for a new smartwatch, the process might go something like this:

[1778] 1. Data collection: The server collects data on the smartwatch's features, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1779] 2. Data pre-processing: The server processes the collected data and converts it into a format suitable for the generative AI model.

[1780] 3. Model Selection and Initialization: The server selects suitable text generation and image generation models for smartwatch ad generation and loads the necessary training data.

[1781] 4. Instructions for generating advertisements: The server inputs the preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1782] 5. Providing reviews: The server provides the generated advertising patterns to the user through a review interface, where the user can check and rate them.

[1783] 6. Feedback collection: The user requests the server to speed up the voice announcement tempo of the “health monitoring-emphasizing advertisement” as feedback, and the server collects this.

[1784] 7. Applying feedback: The server adjusts the "health monitoring emphasis advertisement" based on the user's feedback and speeds up the tempo of the voice announcement.

[1785] 8. Final encoding and output: The server encodes the tailored advertisement and provides the final version to the user via a secure download link.

[1786] Examples of prompt statements

[1787] Below are some examples of prompts to input to the generative AI model.

[1788] Prompt statement:

[1789] text

[1790] Product name: XYZ Smart Watch

[1791] Target demographic: Technology enthusiasts in their 20s and 30s

[1792] Features:

[1793] Health Monitoring Features

[1794] Stylish design

[1795] Long battery life

[1796] Selling points:

[1797] Ideal for health management

[1798] Stylish design using the latest technology

[1799] Lasts for a week on a single charge

[1800] Based on this information, generate three ad variations (emphasizing health monitoring, design, and usage scenarios, respectively).

[1801] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1802] Step 1:

[1803] User enters product information

[1804] The user uses a smartphone application to input product information (e.g., product name, features, target demographic, selling points). The input data is sent to the server. Input: Product information data. Output: Data sent to the server.

[1805] Step 2:

[1806] The server collects the data

[1807] The server receives product information data sent by the user and stores it in a database. It prepares the data for analysis. Input: Product information data sent by the user. Output: Product information data stored in the database.

[1808] Step 3:

[1809] The server analyzes the data

[1810] The server analyzes the collected product information data, cleans up the text data (e.g., removes unnecessary spaces and special characters), resizes the image data, and converts it into a format suitable for the generative AI model. Input: Product information data stored in the database. Output: Data in a format suitable for the generative AI model.

[1811] Step 4:

[1812] Selecting and initializing a generative AI model

[1813] The server selects the optimal generative AI model depending on the type of data, loads the necessary training data, and initializes the model. Input: Data in a format suitable for the generative AI model. Output: The initialized generative AI model.

[1814] Step 5:

[1815] Generating advertising variations

[1816] The server supplies input data to the generative AI model to generate multiple ad patterns. For example, ads that emphasize health monitoring, stylish design, or usage scenarios can be generated. Input: Initialized generative AI model and input data. Output: Multiple ad patterns.

[1817] Step 6:

[1818] User reviews ad variations

[1819] The server provides the generated ad patterns to a review interface on the smartphone, where the user can review them. Input: Multiple ad patterns. Output: User reviews and feedback.

[1820] Step 7:

[1821] Users provide feedback

[1822] The user inputs feedback for the advertisement pattern through the review interface, for example, requesting that the tempo of the voice announcement for the "advertisement emphasizing health monitoring" be made faster. Input: Feedback for the advertisement pattern. Output: Feedback sent to the server.

[1823] Step 8:

[1824] The server reflects the feedback

[1825] The server analyzes the feedback from the user and regenerates / adjusts the ad pattern based on it, for example by speeding up the tempo of the voice announcement. Input: Feedback sent to the server. Output: Adjusted ad pattern.

[1826] Step 9:

[1827] Encoding and Output

[1828] The server encodes the adjusted ad variants into high-resolution formats, optimizing them for various media. It generates the final ad file and provides a secure download link to the user. Input: Adjusted ad variant. Output: Final ad file and download link.

[1829] 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.

[1830] MODE FOR CARRYING OUT THE INVENTION

[1831] In accordance with this invention, a specific implementation method of a system for automatically generating advertisements by combining a generative AI model and an emotion engine is as follows: The main functions of the system include collecting advertisement data, preparing a generative AI model, generating advertisements, processing reviews and feedback, adjusting and outputting the final advertisements, and recognizing and reflecting user emotions.

[1832] Data collection and preparation

[1833] 1. Data Collection

[1834] The server collects advertising data provided by companies, such as product information, target demographics, appealing points, and past advertising examples. This data is structured and stored in an advertising database.

[1835] For example, if a user requests an advertisement for a new smartwatch, information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness will be provided and collected by the server.

[1836] 2. Data Preprocessing

[1837] The server analyzes the collected data and converts it into a suitable format for input into the generative AI model and emotion engine, which includes cleaning up text data, processing image data, and formatting audio data.

[1838] Preparing the generative AI model

[1839] 3. Model Selection and Initialization

[1840] The server analyzes the provided advertising data and its contents and selects an appropriate generative AI model and emotion engine. The generative AI model includes text generation, image generation, and voice generation, and the emotion engine is used to recognize the user's emotional state.

[1841] The server initializes these models and loads the necessary training data.

[1842] Ad generation

[1843] 4. Ad generation instructions

[1844] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, multiple advertisement patterns (such as advertisements emphasizing health monitoring functions or stylish designs) are generated.

[1845] Reviews and Feedback

[1846] 5. Providing Reviews

[1847] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each of them.

[1848] 6. Emotion recognition

[1849] The server uses an emotion engine to detect the user's emotional state, for example by analyzing facial expressions and voice tone to recognize emotions while the user is reviewing an advertisement.

[1850] 7. Emotion-based evaluation

[1851] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[1852] 8. Feedback and Emotion Data Storage

[1853] The feedback provided by the user and the emotional data at that time are stored on the server. For example, if a user likes an advertisement that emphasizes health monitoring, the positive emotional feedback is recorded.

[1854] Final ad adjustment and output

[1855] 9. Applying Feedback

[1856] The server then optimally adjusts the advertising pattern based on user feedback and emotional data, for example by changing the tempo of the voice announcement, modifying the text, or changing some of the visuals.

[1857] 10. Encoding and Serving Ads

[1858] The server then encodes the tailored ad variations into high-definition formats, optimizing them for various media formats such as internet, television, radio, etc. The final ad file is then provided to the user via a secure download link.

[1859] Specific examples

[1860] For example, if a company requests the creation of an advertisement for a new smartwatch, the process would proceed as follows:

[1861] 1. Data Collection

[1862] The server collects data on the smartwatch's functions, its target demographic of technology enthusiasts in their 20s and 30s, and its appealing health monitoring features and stylish design.

[1863] 2. Data Preprocessing

[1864] The server processes the collected data and converts it into a format suitable for generative AI models and emotion engines.

[1865] 3. Model Selection and Initialization

[1866] The server selects text generation and image generation models and emotion engines suitable for generating ads for smartwatches and loads the necessary training data.

[1867] 4. Ad generation instructions

[1868] The server inputs preprocessed data into the generative AI model and automatically generates three types of advertisements: one that emphasizes health monitoring, one that emphasizes design, and one that shows usage scenarios.

[1869] 5. Providing Reviews

[1870] The server provides the generated advertisement patterns to the user through a review interface, and the user checks and evaluates them.

[1871] 6. Emotion recognition

[1872] The server analyzes the user's emotions during the review using an emotion engine and collects the user's reactions for each advertising pattern.

[1873] 7. Gathering Feedback

[1874] As feedback, users request that the tempo of the voice announcement for the "health monitoring-emphasizing advertisement" be made faster, and the server collects this feedback while also recording the user's positive emotional data.

[1875] 8. Applying Feedback

[1876] The server adjusts the advertising pattern and speeds up the tempo of the voice announcements based on the user's feedback and emotional data.

[1877] 9. Encoding and Serving Ads

[1878] The server encodes the tailored advertisement into a high-resolution format and provides the final version to the user via a secure download link.

[1879] Based on the above steps, this system enables companies to generate high-quality advertisements in an efficient and low-risk manner, while also enabling advertisements that reflect user emotions.

[1880] The processing flow will be explained below.

[1881] MODE FOR CARRYING OUT THE INVENTION

[1882] Program processing flow

[1883] Step 1:

[1884] The user sends a request to create an advertisement to the server. The request includes product information, target demographic, and selling points. For example, the user requests "creating an advertisement for a new smartwatch."

[1885] Step 2:

[1886] The server collects the necessary advertising data based on the request received from the user. This data includes product features, target demographic attributes, and selling points. The collected data is stored in an advertising database.

[1887] Step 3:

[1888] The server preprocesses the collected data, for example, cleaning up text data (removing unnecessary characters), resizing image data, and converting the format of audio data.

[1889] Step 4:

[1890] The server selects an appropriate generative AI model based on the preprocessed data, which can include text generation models, image generation models, and speech generation models.

[1891] Step 5:

[1892] The server initializes the selected generative AI model and loads the necessary training data, which prepares the model for generating ads.

[1893] Step 6:

[1894] The server inputs the preprocessed data into the generative AI model and instructs it to automatically generate advertisements. For example, it generates multiple advertisement patterns (such as advertisements that emphasize health monitoring functions or stylish designs).

[1895] Step 7:

[1896] The server applies a quality evaluation algorithm to the generated advertisement patterns to evaluate their quality, and ranks the advertisement patterns based on the evaluation results.

[1897] Step 8:

[1898] The server generates a review interface for providing the generated plurality of advertisement patterns to the user, who uses the interface to review the advertisement patterns and input feedback for each one.

[1899] Step 9:

[1900] The server uses an emotion engine to analyze the user's emotional state during the review, for example, by evaluating the user's facial expressions and tone of voice in real time to recognize the user's emotions.

[1901] Step 10:

[1902] The server evaluates whether the user's emotions are positive or negative based on the analysis results of the emotion engine, and prioritizes ad patterns that generate more positive responses.

[1903] Step 11:

[1904] The user reviews the ad patterns provided, selects the most suitable one, and inputs feedback for improvement, such as specific requests like "make the music faster" or "make the message more concise."

[1905] Step 12:

[1906] The server adjusts the advertising pattern based on user feedback and analyzed emotion data, for example, by changing the tempo of the voice announcement and modifying some of the text.

[1907] Step 13:

[1908] The server encodes the tailored ad variations into a high-definition format, which includes rendering the video and encoding the audio.

[1909] Step 14:

[1910] The server generates the final ad file and provides a secure download link to the user, who can use this link to download the final ad file for use in various media.

[1911] In this way, generative AI can generate high-quality ads in an efficient and low-risk manner. In addition, by combining it with an emotion engine, it becomes possible to create ads that reflect user emotions, resulting in highly appealing ads.

[1912] Example 2

[1913] 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."

[1914] Conventional ad generation systems have difficulty creating ads that are optimized for the target audience because they do not adequately adjust ads to reflect specific user feedback and emotions. Furthermore, there is no established method for integrating emotion recognition technology into the ad generation process to effectively reflect users' positive reactions.

[1915] 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 advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for using an emotion engine to recognize the user's emotional state, and means for final adjustment and output of the advertising patterns based on the user's feedback and emotion data. This enables the generation of optimized advertisements that reflect user emotions in real time.

[1916] A "generative AI model" is an algorithm that uses artificial intelligence technology to automatically generate advertising content such as text, images, and audio.

[1917] "Advertising data" refers to input data necessary to generate advertisements, such as product information, target demographics, appealing points, and past advertising examples.

[1918] An "emotion engine" is a software component for detecting and analyzing a user's emotional state, specifically analyzing data such as facial expressions and vocal tone.

[1919] "Feedback" refers to specific evaluations and opinions provided by users regarding the generated advertising patterns, as well as instructions for improvement.

[1920] "Ad Variants" are different variations or formats of an advertisement automatically generated by a generative AI model.

[1921] The "review interface" is an interface that allows a user to check the generated advertisement pattern and input feedback.

[1922] "Initialization" is the process of loading the settings and training data required to run a generative AI model.

[1923] "Encoding" is the process of outputting the generated advertising content in a high-resolution format or in a format optimized for a particular medium.

[1924] "Data collection" is the process of obtaining, structuring, and storing advertising data needed to generate ads from companies and other sources.

[1925] "Adjustment" refers to the process of optimizing advertising patterns based on user feedback and sentiment data, including text correction, audio tempo adjustment, and visual changes.

[1926] This invention is a system that automatically generates advertisements by combining a generative AI model and an emotion engine. Detailed embodiments are described below. The invention is composed of various means and processes centered on a server, a terminal, and a user.

[1927] Data collection and preparation

[1928] The server collects advertising data provided by companies, such as product information, target demographics, selling points, and past advertising examples. The collected data is stored in an advertising database and classified as text data, image data, and audio data. This data is then converted into the appropriate format for input into the generative AI model and emotion engine. Unnecessary information is removed from the text data and the formatting is adjusted. Image data is resized and its resolution is adjusted, and audio data is subjected to noise removal and sound quality adjustment.

[1929] For example, if a user requests an advertisement for a new smartwatch, the server collects information such as the features of the smartwatch, the target demographic of men in their 20s, and their interest in health and fitness, and stores it in a database.

[1930] Preparing the generative AI model

[1931] The server selects the appropriate generative AI model (e.g., GPT-3 for text generation, DALL-E for image generation) and emotion engine based on the collected ad data. These models are initialized and loaded with the necessary training data to apply them to the specific ad generation task.

[1932] Ad generation

[1933] The server then inputs the pre-processed data into the selected generative AI model and sends a prompt to generate an ad, such as:

[1934] "Generate an ad for a smartwatch targeted at men in their 20s. Features include health monitoring, stylish design, and fitness features."

[1935] As a result, the generative AI model automatically generates multiple advertising patterns, such as ads that emphasize health monitoring, ads that emphasize design, and ads that show usage scenarios.

[1936] Reviews and Feedback

[1937] The server provides the generated ad template to the user through a review interface that can be viewed on the device. The user reviews the ad template and enters feedback. For example, the user can review the text, images, and audio of the ad through a smartphone or PC interface and enter specific feedback.

[1938] The server uses an emotion engine to analyze the user's facial expressions and tone of voice while reviewing the ad, for example, by using a webcam and microphone to collect the user's real-time reactions and recognize the user's emotional state based on the data obtained.

[1939] Final ad adjustment and output

[1940] The server then adjusts the ad template based on the collected feedback and sentiment data, modifying the text, adjusting the tempo of the voice announcement, and changing the visuals. The final ad template is then encoded in high-resolution format and output in a format optimized for each medium. The final ad file is then provided to the user via a secure download link.

[1941] For example, if an "advertisement emphasizing health monitoring" is adjusted to have a faster audio tempo, the adjusted advertisement will be output as the final version.

[1942] The above is a specific embodiment for carrying out the present invention, which makes it possible to generate an optimized advertisement that reflects the user's emotions in real time.

[1943] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1944] Step 1:

[1945] Data collection

[1946] The server collects advertising data such as product information, target demographics, appealing points, and past advertising examples provided by companies.

[1947] The server receives the data through an API or a management portal.

[1948] The server stores the received data in an advertisement database.

[1949] Input: Provided product information, target demographic, selling points, past advertising examples

[1950] Output: Structured data in an advertising database

[1951] Step 2:

[1952] Data preprocessing

[1953] The server analyzes and cleans the raw data collected.

[1954] The server cleans the text data, removing unnecessary information and formatting.

[1955] The server resizes and adjusts the resolution of the image data.

[1956] The server performs noise removal and sound quality adjustment on the audio data.

[1957] Input: Raw data stored in the advertising database

[1958] Output: Data converted into a format suitable for generative AI models and emotion engines

[1959] Step 3:

[1960] Model Selection and Initialization

[1961] The server selects the appropriate generative AI model and emotion engine based on the collected advertising data. Specifically, it uses GPT-3 for text generation and DALL-E for image generation.

[1962] The server initializes the selected generative AI model and emotion engine and loads the training data.

[1963] Input: Data converted into a format suitable for generative AI models and emotion engines

[1964] Output: Initialized generative AI model and emotion engine

[1965] Step 4:

[1966] Ad generation instructions

[1967] The server inputs the preprocessed data into the generative AI model and sends prompt sentences.

[1968] For example, enter the prompt text, "Generate a smartwatch ad for men in their 20s. Features include health monitoring, stylish design, and fitness functions."

[1969] The server receives the generated advertisement pattern.

[1970] Input: prompts, data input by generative AI models

[1971] Output: Multiple ad variations

[1972] Step 5:

[1973] Providing a review

[1974] The server provides the generated advertisement patterns to the user through a review interface that can be viewed on a terminal.

[1975] The user reviews each ad variant and enters specific feedback.

[1976] Input: Generated ad patterns

[1977] Output: User feedback

[1978] Step 6:

[1979] emotion recognition

[1980] The server uses an emotion engine to analyze the user's facial expressions and tone of voice during the advertisement review.

[1981] The server uses a webcam and microphone to collect the user's real-time responses.

[1982] The server recognizes the user's emotional state based on the obtained data.

[1983] Input: Real-time data of the user while reviewing the ad (facial expressions, tone of voice)

[1984] Output: User's emotional state data

[1985] Step 7:

[1986] Emotion-based ratings

[1987] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[1988] The server sets a higher priority to the advertisement pattern that receives the most positive responses.

[1989] Input: User's emotional state data

[1990] Output: Priority of the evaluated ad variants

[1991] Step 8:

[1992] Feedback and sentiment data storage

[1993] The server stores the user feedback and emotion data in a database.

[1994] The server saves it for future use in the ad generation task.

[1995] Input: User feedback and sentiment data

[1996] Output: Stored feedback and emotion data

[1997] Step 9:

[1998] Applying Feedback

[1999] The server adjusts advertising patterns based on collected feedback and sentiment data.

[2000] The server modifies the text of the advertisement, adjusts the tempo of the voice announcement, changes the visuals, etc.

[2001] Input: User feedback and sentiment data

[2002] Output: Adjusted ad patterns

[2003] Step 10:

[2004] Encoding and Serving Ads

[2005] The server encodes the finalized ad variations in high-definition format.

[2006] The server outputs the advertisement in a format optimized for each medium.

[2007] The server provides the final ad file to the user via a secure download link.

[2008] Input: Adjusted ad pattern

[2009] Output: Final ad file in high resolution format

[2010] The above is the specific processing flow of the program for this system.

[2011] (Application example 2)

[2012] 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."

[2013] Conventional ad generation systems have difficulty taking into account user emotional data, making it difficult for the ads they provide to reflect the user's psychological state and preferences. Furthermore, there are limited means for effectively utilizing real-time user feedback, making it difficult to improve the quality of ads. There is a need for a system that can analyze user emotions in real time and reflect that data in ad generation to provide more effective and attractive ads.

[2014] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[2015] In this invention, the server includes means for collecting advertising data as input data using a generative AI model, means for selecting and initializing a generative AI model based on the collected advertising data, means for automatically generating multiple advertising patterns using the generative AI model, means for providing the generated advertising patterns to users and receiving reviews and feedback, means for collecting and analyzing user emotional data in real time, and means for making final adjustments to and outputting advertising patterns based on the user emotional data and feedback, thereby making it possible to instantly reflect the user's emotions and provide advertisements optimized for the user's emotional state.

[2016] A "generative AI model" is a model that uses artificial intelligence to generate advertising data such as text, images, and audio.

[2017] "Advertising data" refers to information for creating advertisements, including product information, target demographics, selling points, and past advertising examples.

[2018] "Emotion data" refers to data relating to the user's emotional state obtained by analyzing the user's facial expressions, voice tone, and other sensor information.

[2019] "Advertising variations" are multiple variations of advertisements automatically generated by a generative AI model.

[2020] "Reviews and feedback" refers to the evaluation and opinions provided by users regarding the provided advertising patterns.

[2021] "Final adjustment and output" refers to modifying the advertising pattern based on user feedback and emotional data and generating the final version.

[2022] A "virtual store" is a virtual shop that exists on the Internet and is a place where users can browse and purchase products.

[2023] "Means for collecting and analyzing in real time" refers to methods and technologies for instantly collecting user emotional data and analyzing it on the spot.

[2024] The "means for prioritizing" is a means for selecting the most suitable advertising pattern from among a plurality of advertising patterns based on the user's emotional data.

[2025] In this invention, we use a generative AI model and an emotion engine to build a system that provides advertisements that reflect customer emotions in a virtual store. The specific system configuration is as follows.

[2026] The server first collects advertising data provided by each company, such as product information, target demographics, selling points, and past advertising examples. This data is structured and stored in an advertising database. Devices such as smartphones and head-mounted displays collect users' purchasing history and preference tag information through apps. Based on this collected data, it is converted into a format suitable for the generative AI model. During this process, text data is cleaned up, image data is processed, and audio data is shaped.

[2027] The server uses this processed data to select and initialize a generative AI model suitable for the product. The generative AI models include a text generation model, an image generation model, and a voice generation model. The server uses these models to automatically generate multiple ad variations. The user then reviews these ads on a smartphone app and provides feedback.

[2028] During this review, the device collects the user's emotional data in real time and analyzes it using an emotion engine. For example, it uses a camera and microphone to analyze the user's facial expressions and tone of voice to obtain emotional data. The server then uses this emotional data and user feedback to make final adjustments to the advertising patterns. For example, if the user expresses positive emotions toward an advertisement emphasizing health monitoring, that advertisement will be displayed preferentially. The finalized advertisements are then encoded in high-resolution formats and optimized for various media.

[2029] As a concrete example, if a company requests the creation of an advertisement for a "new smartwatch," the following process takes place: The server collects and preprocesses data on the smartwatch's functional information, its target demographic of men in their 20s, and its appealing health monitoring function and stylish design. A generative AI model automatically generates patterns, such as advertisements that emphasize health monitoring, advertisements that emphasize design, and advertisements that show usage scenarios. Users view these advertisements on their smartphone app, and emotional data analyzed by an emotion engine is sent to the server. The server prioritizes and provides advertisement patterns that elicit a positive response to the user. Finally, these advertisements are encoded into high-resolution format and displayed in a virtual store.

[2030] An example of a prompt is:

[2031] "We collect data for product ID 12345, perform text and image preprocessing, and then use a generative AI model to generate product ads. Based on user sentiment data, we select the most appropriate ad pattern."

[2032] As described above, this system can instantly reflect the user's emotions and provide personalized advertisements.

[2033] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2034] Step 1:

[2035] The server collects advertising data such as product information, target demographics, selling points, and past advertising examples provided by each company. This collected data is structured and stored in an advertising database. The input is various advertising data provided by companies, and the output is a structured advertising database. Specifically, the server obtains data through APIs and data upload functions and stores it in the database.

[2036] Step 2:

[2037] The server performs preprocessing of the collected advertising data, such as cleaning up text data, processing image data, and formatting audio data. The input is data from the advertising database, and the output is preprocessed data. Specifically, the server converts the data into an appropriate format using natural language processing technology and image processing software (e.g., OpenCV).

[2038] Step 3:

[2039] The server selects and initializes a generative AI model using the preprocessed data. The input is the preprocessed data, and the output is the initialized generative AI model. Specifically, the server loads the model using a generative AI model library (e.g., TensorFlow or PyTorch) and feeds it training data.

[2040] Step 4:

[2041] The server inputs preprocessed data into the generative AI model to automatically generate multiple ad patterns. The input is the initialized generative AI model and preprocessed data, and the output is multiple ad patterns. Specifically, it uses the model's inference function to generate ad text, images, and audio.

[2042] Step 5:

[2043] The terminal provides the generated advertisement pattern to the user and receives reviews and feedback. The input is the advertisement pattern from the server, and the output is the user's review and feedback. Specific operations include displaying the generated advertisement through a user interface and collecting ratings and comments.

[2044] Step 6:

[2045] The device collects the user's emotional data in real time and analyzes it using an emotion engine. The input is the user's facial expressions and vocal tone, and the output is the analyzed emotional data. Specifically, it uses a camera and microphone and analyzes the data using emotion recognition software (e.g., Affectiva SDK).

[2046] Step 7:

[2047] The server then makes final adjustments to the ad patterns based on the collected user emotion data and feedback. The input is the analyzed emotion data and feedback, and the output is the adjusted ad pattern. Specific operations include correcting text and images to improve the quality of the ad.

[2048] Step 8:

[2049] The server then encodes the final adjusted ad template in a high-resolution format, optimizing it for various media. The input is the adjusted ad template, and the output is a high-resolution ad file. Specifically, the server uses video encoding software (e.g., FFmpeg) to convert the data into the appropriate format.

[2050] Step 9:

[2051] Finally, users can view the optimized advertisements in the virtual store, develop interest in the products, and make a purchase. The input is a high-resolution advertisement file, and the output is the user's purchasing behavior. Specifically, users view the advertisements using a smartphone or head-mounted display and click the purchase button.

[2052] 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.

[2053] 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.

[2054] 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.

[2055] 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.

[2056] 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.

[2057] 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.

[2058] 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).

[2059] 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.

[2060] 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."

[2061] 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.

[2062] 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).

[2063] 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.

[2064] 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.

[2065] 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 netwo...

Claims

1. A means for collecting advertising data as input data using a generative AI model; A means for selecting and initializing a generative AI model based on collected advertising data; A means for automatically generating multiple advertising patterns using a generative AI model; means for providing the generated advertisement patterns to users and receiving reviews and feedback; A means for finalizing and outputting the advertisement pattern based on the user's feedback; A system including:

2. A means for accepting requests for creating commercials from users and structuring and saving input data for the generative AI model; A means for evaluating the quality of the advertising patterns output from the generative AI model; The system of claim 1 .

3. a means for storing user feedback for use in subsequent ad generation tasks; a means for encoding the generated advertisements into high-resolution formats and optimizing them for various media; The system of claim 1 .

Images