System

Abstract

A system is provided.SOLUTION: A system, comprising: means for receiving an advertisement content inputted by a user; means for analyzing the received advertisement content to generate a generation prompt; means for generating a new content using the generation prompt; means for comparing the original advertisement content and the generated content to evaluate a similarity; means for determining as a harmful advertisement if the similarity exceeds a certain threshold; and means for blocking the harmful advertisement based on the determination result and notifying the user.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] In recent years, advances in generative AI technology have made it easier for malicious advertisers to create and distribute large volumes of harmful ads. This increases the cost of reviewing ads for advertisers and increases the risk that harmful ads will actually be published. If this problem is left unaddressed, it could undermine the credibility of ads, worsen the user experience, and ultimately reduce the value of internet services as a whole. Conventional approaches, which mainly involve account-by-account responses or manual review, have not provided a fundamental solution. Therefore, a method to quickly and efficiently resolve this issue is needed. [Means for solving the problem]

[0005] The present invention receives advertising content entered by a user, analyzes the received advertising content, and generates a generation prompt using a generation AI. Next, new content is generated based on the generation prompt, and the similarity between the original advertising content and the generated content is evaluated. Based on the results of this evaluation, if the similarity is determined to exceed a certain threshold, the system blocks the advertisement as a harmful advertisement and notifies the user. This enables the rapid and efficient detection and blocking of harmful advertisements, thereby maintaining the reliability of advertisements. The present invention supports advertising content in text, image, and video formats, and performs similarity evaluation by combining natural language processing technology, image recognition technology, or video analysis technology.

[0006] "User" refers to an individual or organization providing advertising content.

[0007] "Advertising Content" refers to advertising information in the form of text, images, or videos submitted by a User.

[0008] "Means for receiving" refers to the technology or device for electronically receiving user-provided advertising content.

[0009] "Means for analyzing and generating generative prompts" refers to technology that analyzes advertising content and automatically generates prompts to be used by the generative AI based on that analysis.

[0010] "Generation prompts" refer to input data or instructions that enable a generative AI to generate similar content.

[0011] "Means for generating new content" refers to the technology or device that allows the generative AI to create new content using generative prompts.

[0012] The "means for comparing and assessing similarity" refers to a technology or device that assesses the degree of similarity between the original advertising content and the generated content.

[0013] "Similarity" refers to an index that indicates how similar two pieces of advertising content are.

[0014] "Means for determining that an advertisement is harmful if it exceeds a threshold" refers to technology or equipment that determines that an advertisement is harmful if the similarity exceeds a predetermined standard value (threshold).

[0015] "Harmful advertising" refers to advertising content that the system determines to be harmful to users.

[0016] "Means for blocking and notifying users" refers to technology or devices that prohibit the display of determined harmful advertisements and notify users of this fact.

[0017] "Natural language processing technology" refers to technology for handling text data and performing semantic analysis, document classification, etc.

[0018] "Image recognition technology" refers to technology for analyzing image data and understanding its contents.

[0019] "Video analysis technology" refers to technology for analyzing video data to understand its content and extract specific features. [Brief explanation of the drawings]

[0020] [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

[0021] 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.

[0022] First, the terms used in the following description will be explained.

[0023] 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).

[0024] 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.

[0025] 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.

[0026] 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.

[0027] 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."

[0028] [First embodiment]

[0029] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0030] 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.

[0031] 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).

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0037] 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.

[0038] 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.

[0039] 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.

[0040] 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."

[0041] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. The program of the present invention is made up of a series of processes that are carried out between a server, a terminal, and a user.

[0042] System Configuration

[0043] 1. Users submit advertising content

[0044] A user uses the ad management system interface to input ad content, which may be in the form of text, images, or video.

[0045] 2. Submitting Content

[0046] The device sends the user-entered advertising content to the server via an HTTPS request, including the advertising content and associated metadata (e.g., advertiser ID, category, target audience, etc.).

[0047] 3. Content analysis and generation of prompts

[0048] The server analyzes the received advertising content and generates a prompt using generative AI. Natural language processing (NLP), image recognition, or video analysis techniques are used for the analysis. For example, specific keywords or phrases in the advertising content, parameters such as "weight loss in a short time" or "special price" are extracted.

[0049] 4. Generating Similar Content

[0050] The server uses generative AI to generate new advertising content based on the generative prompts, with the generated content having similar characteristics to the original advertising content.

[0051] 5. Content Comparison and Similarity Evaluation

[0052] The server compares the original advertising content with the generated similar content. This comparison evaluates the similarity using natural language processing, image recognition, or video analysis technology. A similarity score is calculated as the evaluation result.

[0053] 6. Determining harmful advertising

[0054] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold, the advertisement is recognized as harmful. This determination is made automatically.

[0055] 7. Block harmful ads and notify users

[0056] The server blocks ads that are deemed harmful and notifies the user based on that information. The blocked ads are then stopped from being displayed, and the user is notified by email or other means.

[0057] Specific examples

[0058] For example, if a user types in an ad for a "new diet supplement":

[0059] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0060] When a user submits advertising content, the following steps are performed:

[0061] 1. User enters advertising content:

[0062] The user enters the above text into the advertisement management system and clicks the send button.

[0063] 2. The device sends the advertising content:

[0064] The user's terminal transmits the advertising content and associated metadata to the server.

[0065] 3. The server analyzes the ad content:

[0066] The server receives the advertising content and uses natural language processing techniques to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[0067] 4. Generate content based on a generation prompt:

[0068] The server uses generation AI to generate advertising content with similar content.

[0069] 5. The server compares the original ad with the generated ad:

[0070] The server evaluates the similarity between the original advertising content and the generated advertising content.

[0071] 6. The server evaluates the similarity score:

[0072] If the similarity score is high, the ad is determined to be harmful.

[0073] 7. The server blocks harmful ads and notifies the user:

[0074] Advertisements that are determined to be harmful are blocked, and the user is notified that "the advertisement has been determined to be harmful and has been blocked."

[0075] Through this process, the system can efficiently detect and block malicious and harmful ads, while providing appropriate feedback to users, ensuring transparency.

[0076] The processing flow will be explained below.

[0077] Step 1:

[0078] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[0079] Step 2:

[0080] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[0081] Step 3:

[0082] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[0083] Step 4:

[0084] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[0085] Step 5:

[0086] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[0087] Step 6:

[0088] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[0089] Step 7:

[0090] The server determines whether an advertisement is harmful based on the similarity score. If the similarity score exceeds a preset threshold, the server determines the advertisement as harmful. Conversely, if the similarity score is below the threshold, the advertisement is not determined to be harmful.

[0091] Step 8:

[0092] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[0093] Step 9:

[0094] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block and the similarity score.

[0095] These steps enable the system of the present invention to efficiently detect, evaluate, and block harmful advertising content, and provide appropriate feedback to the user.

[0096] Example 1

[0097] 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."

[0098] Conventional ad management systems lacked the ability to automatically detect and block harmful ad content entered by users. This resulted in a high risk of harmful ads being displayed and made it difficult to maintain ad quality. Feedback to users was also insufficient, lacking transparency. To solve these problems, a system that could automatically detect and block harmful ads with high accuracy and efficiency was needed.

[0099] 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.

[0100] In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate similarity, means for determining that the advertising content is harmful if the similarity exceeds a certain threshold, means for blocking the harmful advertising based on the determination result and notifying the user, means for transmitting information including metadata related to the advertising content, means for extracting specific keywords or phrases based on the received advertising content, and means for calculating the similarity of the generated advertising content. This effectively solves the problems of detecting and blocking harmful advertising that conventional systems have had, improves the quality of advertising, and enables transparent feedback to users.

[0101] "Advertising Content" means information that a user inputs into the Ad Management System, and may be provided in the form of text, images, or video.

[0102] A "generative prompt" is input data that allows a generative AI model to generate new advertising content based on information extracted by analyzing received advertising content.

[0103] A "generative AI model" is a machine learning model for automatically generating new advertising content based on generative prompts.

[0104] The "similarity" refers to the degree of similarity between the original advertising content and the generated advertising content, and is expressed as a numerical value.

[0105] The "threshold" refers to a boundary value that is the standard for determining whether the similarity is a harmful advertisement.

[0106] A "harmful ad" is an ad whose content has a similarity score that exceeds a certain threshold and has been automatically determined not to be displayed.

[0107] "Metadata" is additional information related to advertising content, including advertiser ID, category, target audience, etc.

[0108] "Natural language processing technology" is a technology for analyzing text-based content and extracting specific keywords and phrases.

[0109] "Image recognition technology" is a technology for analyzing image-based content and extracting specific features and patterns.

[0110] "Video analysis technology" is a technology for analyzing video-based content and extracting specific frames or scenes.

[0111] An "HTTPS request" is a secure communication protocol used when exchanging data over the Internet.

[0112] "Means for receiving" refers to hardware or software for receiving advertising content from a user.

[0113] "Means for analyzing" refers to hardware or software for examining received advertising content and extracting necessary information.

[0114] The "means for evaluating" refers to hardware or software for measuring the similarity between the original advertising content and the generated content and evaluating it as a numerical value.

[0115] "Blocking measures" refers to hardware or software that stops the display of advertising content that is determined to be harmful.

[0116] The "notification means" refers to hardware or software for notifying the user of the determination result.

[0117] The present invention is a system that automatically analyzes advertising content entered by users and detects and blocks harmful advertisements as necessary. The system configuration includes a series of processes performed between a server, a terminal, and a user. The following describes the specific program processing for implementing the present invention and its detailed explanation.

[0118] Program processing explanation

[0119] The system has the following main features:

[0120] 1. Receiving advertising content: The user enters advertising content using the advertising management system interface. The entered advertising content can be in the form of text, images, or videos, and is sent by the device to the server using an HTTPS request.

[0121] 2. Analyzing advertising content and generating a generative prompt: The server analyzes the received advertising content. This analysis uses natural language processing (NLP), image recognition, or video analysis technology. For example, specific keywords or phrases in the advertising content, or parameters such as "fast weight loss" or "special price offer," are extracted. A generative prompt is generated based on the extracted information.

[0122] 3. Generating similar content based on the prompt: The server generates new advertising content using a generative AI model (e.g., GPT-4) based on the prompt. The generated content has similar characteristics to the original advertising content.

[0123] 4. Comparison of advertising content and similarity evaluation: The server compares the original advertising content with the newly generated advertising content and evaluates their similarity. This evaluation uses natural language processing technology, image recognition technology, or video analysis technology, and calculates a similarity score as the evaluation result.

[0124] 5. Identifying and blocking harmful ads: The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold, the ad is recognized as harmful. This determination is made automatically. Based on the determination result, harmful ads are blocked and the user is notified. This notification is made by means such as email.

[0125] Specific examples

[0126] For example, consider the following ad content for a "new diet supplement" entered by a user:

[0127] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0128] In this case, the system does the following:

[0129] Receiving advertising content: The user enters the above text into the advertising management system and clicks the submit button.

[0130] Content submission: The user's device submits the ad content and metadata to the server using an HTTP POST request.

[0131] Content analysis: The server uses text analysis tools to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[0132] Similar content generation: The server inputs the generation prompt into the generative AI model to generate new advertising content with similar content.

[0133] Similarity evaluation: The server evaluates the similarity between the original advertising content and the newly generated advertising content and calculates a similarity score.

[0134] Identifying and blocking harmful ads: If the server determines that an ad is harmful based on the similarity score, it blocks the ad and notifies the user that "the ad has been identified as harmful and blocked."

[0135] Through these processes, the system can efficiently detect and block harmful ads, thereby improving the quality of ads and providing transparent feedback to users.

[0136] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0137] Step 1:

[0138] User enters advertising content

[0139] Specific operation: The user accesses the interface of the ad management system and inputs the ad content. The ad content can be input in the form of text, image, or video. The user clicks the submit button to submit the content.

[0140] Input: Ad content in the form of text, images, or videos

[0141] Output: Clicking the submit button sends the ad content to the device.

[0142] Step 2:

[0143] The device sends advertising content

[0144] How it works: The user's device sends the entered advertising content to the server using an HTTPS request, along with metadata such as the advertiser's ID, category, and target audience.

[0145] Input: User-entered ad content and metadata

[0146] Output: Ad content and metadata sent to the server as an HTTPS request

[0147] Step 3:

[0148] The server analyzes the ad content

[0149] How it works: The server uses natural language processing (NLP), image recognition, and video analysis techniques to analyze the received ad content. For example, in the case of a text ad, keywords and phrases such as "fast weight loss" or "special offer" are extracted.

[0150] Input: Ad content and metadata received as an HTTPS request

[0151] Output: Generated prompts containing extracted keywords and phrases

[0152] Step 4:

[0153] The server generates similar content based on the prompt.

[0154] How it works: The server generates new ad content using a generative AI model (e.g., GPT-4) based on the prompt. The generated ad content has similar characteristics to the original ad content.

[0155] Input: Generate prompt

[0156] Output: New ad content generated by the generative AI model

[0157] Step 5:

[0158] The server compares the original ad content with the generated ad content

[0159] Specific operation: The server uses natural language processing, image recognition, and video analysis technologies to evaluate the similarity between the original advertising content and the generated advertising content. This evaluation calculates a similarity score.

[0160] Input: Original and generated ad content

[0161] Output: Similarity score

[0162] Step 6:

[0163] The server determines harmful ads

[0164] How it works: The server compares the similarity score with a pre-defined threshold, and if the score exceeds the threshold, it marks the ad as harmful. This decision is made automatically.

[0165] Input: Similarity score

[0166] Output: Harmful ad detection result

[0167] Step 7:

[0168] The server blocks harmful ads and notifies the user.

[0169] Specific operation: The server blocks ads that are determined to be harmful and notifies the user based on that information. Notifications are sent via email or other means.

[0170] Input: Harmful ad verdict

[0171] Output: Blocked ads and notification to the user

[0172] (Application example 1)

[0173] 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."

[0174] Advertisements provided on the Internet often contain harmful content. This leads to the distribution of unreliable advertisements and advertisements that may have a negative impact on users. In particular, the display of such harmful advertisements on smartphone and PC applications has a significant impact on the user experience. The present invention aims to provide a system that efficiently and automatically detects and blocks such harmful advertisements.

[0175] 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.

[0176] In this invention, the server includes: means for receiving advertising content input by a user; means for analyzing the received advertising content and generating a generation prompt; means for generating new content using the generation prompt; means for comparing the original advertising content with the generated content and evaluating the similarity; means for determining that the generated content is a harmful advertisement if the similarity exceeds a certain threshold; means for blocking the harmful advertisement based on the evaluation result and notifying the user; means for evaluating the similarity between the generated content and the original advertising content and determining that the advertisement is harmful based on the evaluation result; and means for notifying the user that the advertisement is blocked if it is determined to be harmful. This makes it possible to automatically determine whether advertising content created by an advertiser is harmful and to quickly block content that is determined to be harmful.

[0177] "User" means the person or entity that creates and inputs advertising content into the system.

[0178] "Advertising Content" means advertising information provided in text, image, or video format.

[0179] The "means for receiving" refers to a method or device for capturing the advertising content entered by the user into the server.

[0180] "Means for analyzing" refers to a method or device for analyzing received advertising content to create generated prompts.

[0181] "Generative prompts" are input data that a generative AI model uses to create new content based on the analysis results.

[0182] A "generating means" is a method or device that uses a generating prompt to create new content.

[0183] The "means for comparing and assessing similarity" refers to a method or device for calculating the degree of similarity in content or characteristics between the original advertising content and the generated content.

[0184] The "certain threshold" is a standard value beyond which the similarity is judged to be harmful advertising.

[0185] "Means for determining" refers to a method or device that determines whether advertising content is harmful based on the similarity assessment.

[0186] "Blocking means" refers to a method or device that stops the display or distribution of advertising content that has been determined to be harmful advertising.

[0187] "Means for notifying" refers to a method or device for informing the user of the advertisements that have been blocked and the reasons for their blocking.

[0188] The present invention is a system for receiving advertising content entered by a user and automatically detecting and blocking harmful advertisements. The system is composed of a user terminal, a server, and a program that performs a series of processes between the user and the user.

[0189] System configuration

[0190] 1. User Device

[0191] The user inputs advertising content using the ad management system interface. The input advertising content can be in the form of text, images, or videos. The user device then sends the advertising content and related metadata (such as the advertiser's ID, category, and target audience) to the server. The transmission uses an HTTPS request to ensure data security.

[0192] 2. Server

[0193] The server analyzes the advertising content received from the user's device. For the analysis, it uses natural language processing (NLP), image recognition, and video analysis technologies, and generates a generated prompt using a generative AI model. The generated prompt is created based on the main keywords and phrases in the analyzed advertising content. For example, "weight loss effect in a short time," "special price," and "innovative supplement" are extracted as generated prompts.

[0194] Next, the server generates new advertising content using a generative AI model based on the generated prompts. The generated advertising content has similar characteristics to the original advertising content. The server compares the original advertising content with the generated advertising content and evaluates their similarity. This evaluation is performed using natural language processing technology, image recognition technology, and video analysis technology. A similarity score is calculated as the evaluation result.

[0195] 3. Determining and notifying harmful advertising

[0196] The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold (e.g., 0.8), the ad is recognized as harmful. This determination is made automatically. If the ad is determined to be harmful, the server blocks the ad and notifies the user. The notification is sent by email, in-app message, or other means.

[0197] Specific examples

[0198] For example, a user enters the following ad content:

[0199] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0200] For this ad content, the following prompt text is generated:

[0201] "Fast weight loss," "Special price," "Innovative supplement"

[0202] The server generates new advertising content based on these prompts and compares it with the original advertising content to evaluate its similarity. If the similarity score exceeds a threshold, the server determines the advertising content to be harmful and notifies the user that it is blocked.

[0203] Hardware and software used

[0204] User device: Inputs and transmits advertising content. Examples include smartphones and PCs.

[0205] Server: Performs reception, analysis, generation, judgment, and notification processes. Uses a cloud server (e.g., Amazon Web Services, Google Cloud Platform, etc.).

[0206] Natural language processing technologies: such as the Hugging Face transformers library and the BERT model.

[0207] Image recognition technology: Uses OpenCV and TensorFlow.

[0208] Video analysis technology: Uses FFmpeg and OpenCV.

[0209] Generative AI models: Large-scale language models such as GPT-3.

[0210] In this way, the present invention can automatically analyze advertising content, effectively block harmful advertisements, and provide appropriate feedback to users.

[0211] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0212] Step 1:

[0213] The user inputs advertising content using the interface of the advertising management system. The input advertising content can be in the form of text, images, or videos. The user terminal then sends the input advertising content and related metadata (such as advertiser ID, category, target audience, etc.) to the server using an HTTPS request.

[0214] Input: Ad content, associated metadata

[0215] Output: Ad content and associated metadata included in the HTTPS request

[0216] Step 2:

[0217] The server receives the advertising content and associated metadata sent from the user device. The server then analyzes the advertising content using natural language processing (NLP), image recognition, or video analysis techniques to generate a generated prompt that includes key keywords or phrases from the analyzed advertising content (e.g., "fast weight loss," "special price," "innovative supplement").

[0218] Input: Ad content, associated metadata

[0219] Output: Generated prompt

[0220] Step 3:

[0221] The server generates new advertising content using a generative AI model based on the prompts. The generative AI model is a large-scale language model, such as GPT-3, that receives the prompts as input and generates new advertising content based on them. The generated advertising content has similar characteristics to the original advertising content.

[0222] Input: Generate prompt

[0223] Output: Generated ad content

[0224] Step 4:

[0225] The server compares the original advertising content with the generated advertising content and evaluates the similarity. This similarity evaluation uses natural language processing technology, image recognition technology, or video analysis technology. For example, the server compares the text portions of the original advertising content with the generated advertising content and calculates a similarity score.

[0226] Input: Original ad content, Generated ad content

[0227] Output: Similarity score

[0228] Step 5:

[0229] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold (e.g., 0.8), the advertisement is recognized as harmful. This determination is made automatically.

[0230] Input: Similarity score

[0231] Output: Harmful ad detection result

[0232] Step 6:

[0233] The server blocks advertising content that is determined to be harmful within the system, thereby preventing users from viewing ads that are recognized as harmful. The server also notifies the user that the ad has been determined to be harmful and blocked. This notification is made by means of email, in-app message, etc.

[0234] Input: Harmful advertisement judgment result

[0235] Output: Block execution, notification message

[0236] This series of processes ensures the safety of advertising content and improves the user experience.

[0237] 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.

[0238] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. Furthermore, the present invention incorporates an emotion engine that recognizes the user's emotions, thereby evaluating the emotional impact of advertisements and improving the accuracy of determining harmful advertisements. The specific program processing flow is explained below.

[0239] System Configuration

[0240] 1. Users submit advertising content

[0241] The user uses the ad management system interface to input ad content, which can be in the form of text, images, or videos, and then clicks the "Submit" button.

[0242] 2. Submitting Content

[0243] The device sends the advertising content entered by the user and its associated metadata (e.g., advertiser ID, category, target audience) to the server via an HTTPS request.

[0244] 3. Content analysis and generation of prompts

[0245] The server analyzes the received advertising content and generates prompts using generative AI, which uses natural language processing, image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features.

[0246] 4. User Emotion Analysis

[0247] The server uses an emotion engine to analyze the user's emotions toward the received advertising content. Sentiment analysis extracts emotions, such as positive, negative, or neutral, from text, audio, or image data.

[0248] 5. Generating Similar Content

[0249] Based on the generated prompts, the server uses generative AI to generate new ad content, the format of which matches the original ad content (text, image, or video).

[0250] 6. Content Comparison and Similarity Evaluation

[0251] The server compares the original advertising content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology.

[0252] 7. Identifying Harmful Advertisements Using Sentiment Analysis Results

[0253] The server combines the similarity score and the results of user sentiment analysis to determine whether an advertisement is harmful. If the similarity score and the results of sentiment analysis exceed a certain threshold, the advertisement is determined to be harmful.

[0254] 8. Block harmful ads and notify users

[0255] The server blocks the ads that are determined to be harmful, updates the status of the ads to "blocked," and saves it in the database. At the same time, it automatically sends a notification email to the user informing them that the ads have been determined to be harmful and blocked.

[0256] Specific examples

[0257] For example, suppose a user enters an advertisement for a "new diet supplement" as follows:

[0258] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0259] The specific processing flow when this advertisement is submitted by a user is as follows.

[0260] 1. User enters advertising content:

[0261] The user enters the above text into the form of the ad management system and clicks the "Submit" button.

[0262] 2. The device sends the advertising content:

[0263] The user's device sends the advertising content and associated metadata to the server via an HTTPS request.

[0264] 3. The server analyzes the ad content:

[0265] A server receives the advertising content and extracts specific keywords or phrases to generate generated prompts.

[0266] 4. The server analyzes the user's emotions:

[0267] The server uses an emotion engine to analyze the emotional impact of advertising content on the user.

[0268] 5. Generate content based on a generation prompt:

[0269] The server uses generative AI to generate similar advertising content.

[0270] 6. The server compares the original ad with the generated ad:

[0271] The server evaluates the similarity between the generated advertising content and the original advertisement and calculates a similarity score.

[0272] 7. The server uses the sentiment analysis results to determine harmful ads:

[0273] The server combines the similarity score with the sentiment analysis results to determine whether the ad is harmful.

[0274] 8. The server blocks harmful ads and notifies the user:

[0275] The server blocks ads that are determined to be harmful and notifies the user that "the ad has been determined to be harmful and has been blocked."

[0276] By taking into account the results of user sentiment analysis, the system of the present invention can detect harmful advertisements with higher accuracy than conventional methods that only use similarity evaluation. In addition, appropriate feedback is provided to users, ensuring transparency.

[0277] The processing flow will be explained below.

[0278] Step 1:

[0279] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[0280] Step 2:

[0281] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[0282] Step 3:

[0283] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[0284] Step 4:

[0285] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[0286] Step 5:

[0287] The server analyzes the user's emotions. The server uses an emotion engine to classify the user's emotions toward the advertising content as positive, negative, or neutral. This can be done using text, voice, or image analysis.

[0288] Step 6:

[0289] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[0290] Step 7:

[0291] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[0292] Step 8:

[0293] The server integrates the similarity score and the user's sentiment analysis results, and performs a comprehensive evaluation by combining the similarity score and the sentiment analysis results.

[0294] Step 9:

[0295] The server judges harmful advertisements. If the overall evaluation based on the similarity and sentiment analysis results exceeds a certain threshold, the advertisement is judged as harmful. Conversely, if it is below the threshold, the advertisement is not judged as harmful.

[0296] Step 10:

[0297] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[0298] Step 11:

[0299] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block, the similarity score, and the results of sentiment analysis.

[0300] These steps enable the system of the present invention to not only efficiently detect and block harmful advertising content, but also provide feedback to help users understand the process.

[0301] Example 2

[0302] 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."

[0303] Conventional advertising content filtering systems have limited accuracy in determining whether an ad's content is harmful, leading to false positives. Furthermore, because they do not consider users' emotional responses, they can display emotionally unpleasant ads. This can lead to a poor user experience and a loss of trust in advertisers.

[0304] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0305] In this invention, the server includes a means for analyzing a user's emotions, a means for analyzing received content and generating a generation prompt, and a means for generating new content using the generation prompt, thereby enabling highly accurate determination of harmful content that takes user emotions into consideration.

[0306] "User" means a person who inputs and transmits advertising content using the system.

[0307] "Advertising Content" means a collection of information entered by a user that is analyzed and evaluated by the system, and may take the form of text, images, or video.

[0308] The "means for receiving" is a part of the system that has the functionality for capturing advertising content sent by a user.

[0309] The "means for analyzing" is a part of the system that has the function of analyzing received advertising content and extracting useful information.

[0310] A "generative prompt" is input information given to a generative AI model based on the analysis results of advertising content.

[0311] A "means for generating" is a part of the system that has the ability to create new content using generation prompts.

[0312] A "comparison means" is a part of the system that has the ability to evaluate the original advertising content against the generated content.

[0313] The "means for assessing similarity" is a part of the system that has the function of quantifying the similarity between the original advertising content and the generated content.

[0314] The "means for determining that content is harmful" is a part of the system that has the function of determining whether advertising content is harmful based on the results of similarity evaluation and sentiment analysis.

[0315] "Blocking means" is a part of the system that has the function of preventing advertising content that is determined to be harmful from being displayed.

[0316] The "means for notifying the user" is a part of the system that has the function of notifying the user that advertising content has been determined to be harmful.

[0317] The "means for analyzing emotions" is a part of the system that has the function of evaluating the emotional impact that advertising content has on the user and extracting emotions such as positive, negative, or neutral.

[0318] "Natural language processing technology" is a technology for analyzing the meaning and content of text data.

[0319] "Image recognition technology" is a technology for analyzing the contents of image data.

[0320] "Video analysis technology" is a technology for analyzing the content of video data.

[0321] The present invention is a system that receives advertising content entered by a user and automatically determines whether that content is harmful and blocks it. Furthermore, by analyzing user sentiment, the accuracy of determining harmful advertisements is improved. Below, specific embodiments of the present invention are described.

[0322] System Configuration

[0323] User Roles

[0324] The user inputs the advertising content through the advertising management system. The advertising content can be provided in the form of text, images, or videos. Once the user inputs the advertising content and clicks the "Submit" button, the content is received by the system.

[0325] Device Role

[0326] When a user submits advertising content, the device sends the entered advertising content and related metadata (e.g., advertiser ID, category, target audience) to the server using an HTTPS request. The device uses an existing web browser or a dedicated application.

[0327] Server Roles

[0328] The server receives the advertising content sent by the user and performs the following processes.

[0329] 1. Advertising content analysis:

[0330] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies. Specifically, it uses open source NLP libraries (e.g., spaCy, NLTK), image recognition libraries (e.g., OpenCV, TensorFlow), and video analysis technologies (e.g., FFmpeg).

[0331] Specific keywords, phrases, and visual features are extracted from the analyzed data, and generative prompt sentences are generated based on these.

[0332] 2. User sentiment analysis:

[0333] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the emotional impact of the ad content on the user, categorizing emotions into positive, negative, and neutral categories.

[0334] 3. Generate new content based on a generation prompt:

[0335] The server uses a generative AI model (e.g., GPT-4, DALL-E) to generate new ad content based on the generated prompt. The generated content matches the original format (text, image, video).

[0336] 4. Comparison of original ad content and generated content:

[0337] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score.

[0338] 5. Determining harmful advertising:

[0339] The server determines whether an advertisement is harmful based on the similarity score and the result of sentiment analysis. If the similarity score or sentiment score exceeds a certain threshold, the advertisement is determined to be harmful.

[0340] 6. Block harmful ads and notify users:

[0341] The server blocks the ad that is determined to be harmful, updates the ad's status to "blocked," and notifies the user by email that the ad has been determined to be harmful and blocked.

[0342] Specific examples

[0343] For example, if a user enters an ad for a "new diet supplement" like this:

[0344] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0345] Once this ad content is sent, the server performs the following steps:

[0346] 1. User enters and submits ad:

[0347] The user enters text into the ad management system and clicks the "Submit" button.

[0348] 2. The device sends the advertising content:

[0349] The device sends advertising content to the server via an HTTPS request.

[0350] 3. The server analyzes the advertising content:

[0351] Keywords such as "innovative," "weight loss effect," and "special price" are extracted, and prompt sentences are generated to be given to the generative AI model.

[0352] 4. The server analyzes the user's emotions:

[0353] Using an emotion engine, the emotions that ad text evokes in users are analyzed, and an emotion score is obtained, such as 80% positive, 15% negative, and 5% neutral.

[0354] 5. Generate new content based on a generation prompt:

[0355] A prompt sentence is input into the generative AI model to generate new advertising text, such as "Why not lose weight in a healthy way with this supplement?"

[0356] 6. Server compares original ad with new ad:

[0357] The original ad and the new ad are compared using natural language processing and a similarity score (e.g., 85 points) is calculated.

[0358] 7. The server determines whether the ad is harmful:

[0359] Advertisements are analyzed based on similarity scores and sentiment scores to determine whether they are harmful.

[0360] 8. Blocking harmful ads and notifying users:

[0361] Update the ad status to "Blocked" and notify the user that "The ad has been identified as harmful and blocked."

[0362] The system of the present invention enables automatic detection and blocking of harmful advertisements with high accuracy, taking into account the user's emotional response, thereby improving the user experience and ensuring advertising safety.

[0363] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0364] Step 1:

[0365] User enters and submits advertising content

[0366] The user uses the interface of the ad management system to input the ad content (in the form of text, images, or videos). Then, he clicks the "Submit" button to send the input content. The input data is the ad content provided by the user, and the output data is the ad content sent to the terminal. Specifically, the user attaches text, images, or videos to a form and submits it.

[0367] Step 2:

[0368] The device sends advertising content to the server

[0369] The device sends the advertising content and related metadata entered by the user to the server using an HTTPS request. The input data is the content and metadata entered by the user, and the output data is the content data sent to the server. Specifically, the device converts the user's input data into packets and sends them to the server as an encrypted HTTPS request.

[0370] Step 3:

[0371] The server analyzes the ad content and generates a prompt

[0372] The server analyzes the received advertising content. It uses natural language processing techniques (e.g., spaCy, NLTK), image recognition techniques (e.g., OpenCV, TensorFlow), and video analysis techniques (e.g., FFmpeg) to extract specific keywords, phrases, and visual features from the content. The input data is the advertising content sent to the server, and the output data is the generated prompt. Specifically, the server runs the analysis engine to extract keywords such as "weight loss" and "special price" from the original content and compose the generated prompt.

[0373] Step 4:

[0374] The server analyzes the user's emotions

[0375] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the user's emotions toward the ad content. The input data here is the analyzed ad content, and the output data is an emotion score such as positive, negative, or neutral. Specifically, the server passes the content data to the emotion analysis module and obtains the emotion score.

[0376] Step 5:

[0377] Generate new content based on a prompt

[0378] The server uses the generation prompt sentence to generate new advertising content for a generative AI model (e.g., GPT-4, DALL-E). The input data is the generation prompt sentence, and the output data is the generated new content. Specifically, the server inputs the generation prompt into the AI ​​model, and the model generates new advertising content (e.g., text and images).

[0379] Step 6:

[0380] The server compares the original ad content with the new content

[0381] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score. The input data is the original advertising content and the generated new content, and the output data is the similarity score. Specifically, the server inputs both contents into an analysis engine and quantifies the similarity.

[0382] Step 7:

[0383] The server determines harmful ads

[0384] The server determines whether an ad is harmful based on the similarity score and the results of sentiment analysis. The input data are the similarity score and sentiment score, and the output data is the harmfulness determination result. Specifically, the server compares the obtained scores and determines that an ad is harmful if it exceeds a threshold.

[0385] Step 8:

[0386] The server blocks harmful ads and notifies the user

[0387] The server blocks ads that are determined to be harmful and updates the ad status to "blocked." It also automatically sends a notification email to the user informing them that the ad has been determined to be harmful and blocked. The input data is the harmful determination result, and the output data is the updated ad status and notification email. Specifically, the server updates the ad database and sends a notification email to the user.

[0388] (Application example 2)

[0389] 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."

[0390] With the spread of internet advertising, the risk of harmful advertising content adversely affecting users is increasing. Therefore, a system that automatically monitors advertisements and quickly detects and blocks harmful advertisements is needed. Furthermore, conventional systems that only use similarity evaluation are sometimes insufficient in accuracy, so there is a need for a highly accurate harmful advertisement detection system that takes into account the emotional impact of users.

[0391] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content and generating a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content and evaluating the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user. This makes it possible to detect harmful advertisements quickly and accurately, and maintain the integrity of advertisements while minimizing the adverse impact on users.

[0392] "User-entered advertising content" refers to data in the form of text, images, or videos of advertisements provided by users such as advertisers or advertising agencies through online platforms.

[0393] The "receiving means" refers to a communication interface and software that has the function of receiving the advertising content input by the user on the server side.

[0394] "Means for analyzing and generating generative prompts" refers to the function of analyzing received advertising content and creating instructions (prompts) for generating new content using generative AI based on that content.

[0395] "Means for generating new content using generative prompts" refers to the ability to automatically generate new advertising content based on generative prompts using a generative AI model.

[0396] "Means for comparing generated content and assessing similarity" refers to algorithms and technologies that match original advertising content with generated advertising content and measure their similarity.

[0397] "Means for recognizing user emotions and assessing the emotional impact of advertising" refers to a sentiment analysis engine and associated algorithms for analyzing the emotional response of advertising content to users and assessing its impact.

[0398] The term "certain threshold" refers to a predetermined reference value for determining whether an advertisement is harmful or not, based on the results of similarity assessment and sentiment analysis.

[0399] "Means for determining harmful advertising" refers to an algorithm that automatically determines that advertising content is harmful if the results of similarity assessment and sentiment analysis exceed a certain threshold.

[0400] "Means for blocking and notifying users" refers to the function of excluding advertising content that is determined to be harmful from display and distribution networks, and automatically notifying advertisers of this fact.

[0401] This invention relates to an automated system for accurately determining the harmfulness of advertising content entered by a user and blocking it. The system of the present invention mainly includes the following main means: means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user.

[0402] Program processing explanation

[0403] 1. Receiving advertising content

[0404] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. After entering the content, the user clicks the "Submit" button, which sends the content and its metadata (advertiser ID, category, target audience) to the server. This communication uses an HTTPS request.

[0405] 2. Analyzing advertising content and generating prompts

[0406] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features. Based on the analysis results, a prompt is generated, and new advertising content is automatically generated based on the advertising content.

[0407] 3. Emotion analysis

[0408] The server evaluates the emotional impact of advertising content using a sentiment analysis engine that uses existing sentiment analysis software, such as NLTK's SentimentIntensityAnalyzer, to extract positive, negative, or neutral sentiment scores from text, audio, or image data.

[0409] 4. Similarity evaluation

[0410] The server compares the original ad content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology. For example, it can use Hugging Face's transformer library to calculate the similarity between the similar content generated by the GPT-3 model and the original content.

[0411] 5. Identifying and blocking harmful ads

[0412] The server combines the results of the similarity assessment and sentiment analysis to determine whether an ad is harmful. Ads that are determined to be harmful are automatically blocked and users are notified of this. Notifications are sent via email or system alerts.

[0413] Specific examples

[0414] For example, if a user enters the following ad content:

[0415] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0416] When this ad is submitted, the system works as follows: First, the ad content is received, the text is analyzed, and a generative prompt is generated. Using the generative prompt, a generative AI model (GPT-3) generates similar content, such as:

[0417] "Just take this new supplement and lose weight fast! Try it at a great price!"

[0418] The sentiment analysis engine then evaluates the emotional impact and calculates the similarity between the original content and the generated content. Based on this result, it determines whether the ad is harmful, and if so, the ad is blocked and the user is notified. This allows for early detection of harmful content and prevents adverse effects on users.

[0419] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0420] Step 1:

[0421] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. The user clicks the "Submit" button, which sends the entered content and its metadata (advertiser ID, category, target audience) to the server. The entered data is securely transmitted to the server via an HTTPS request.

[0422] Step 2:

[0423] The server analyzes the received advertising content. First, it performs text analysis and uses natural language processing (NLP) techniques to extract specific keywords and phrases within the content. It also uses image and video analysis techniques to extract visual features. These analyses result in a generated prompt. The input is the advertising content and related metadata, and the output is the analysis results and the generated prompt.

[0424] Step 3:

[0425] The server uses the generative prompts to generate new advertising content. A generative AI model (e.g., GPT-3) is used to create similar advertising content based on the generative prompts. The generated content is in the same format (text, image, or video) as the original. The input is the generative prompts, and the output is the generated advertising content.

[0426] Step 4:

[0427] The server uses a sentiment analysis engine to evaluate the emotional impact of ad content on users. Sentiment analysis software such as NLTK's SentimentIntensityAnalyzer extracts positive, negative, or neutral sentiment scores from text, audio, or image data. The input is the ad content, and the output is the sentiment analysis results.

[0428] Step 5:

[0429] The server compares the generated ad content with the original ad content and calculates a similarity score. It uses natural language processing, image recognition, or video analysis technology to evaluate the similarity based on the features of both pieces of content. For example, it uses Hugging Face's transformers library to compare it with the output of the GPT-3 model. The input is the original ad content and the generated content, and the output is a similarity score.

[0430] Step 6:

[0431] The server determines whether an advertisement is harmful or not based on the similarity score and the results of sentiment analysis. The criteria (certain threshold) for determining whether an advertisement is harmful is set in advance, and if the similarity score and the results of sentiment analysis exceed this threshold, the advertisement is determined to be harmful. The input is the similarity score and the results of sentiment analysis, and the output is the result of determining whether the advertisement is harmful.

[0432] Step 7:

[0433] The server blocks advertising content that is determined to be harmful. It updates the status of the advertisement to "blocked" and saves it in the database. It also automatically sends a notification to the user that the advertisement has been determined to be harmful and blocked. The input is the result of determining whether the advertisement is harmful, and the output is the notification to the user and the blocking status of the advertising content.

[0434] Through this series of processes, the system can accurately determine whether an advertisement is harmful and prevent the distribution of harmful advertising content.

[0435] 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.

[0436] 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.

[0437] 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.

[0438] [Second embodiment]

[0439] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0440] 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.

[0441] 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).

[0442] 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.

[0443] 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.

[0444] 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).

[0445] 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.

[0446] 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.

[0447] 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.

[0448] 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.

[0449] In the smart glasses 214, 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.

[0450] 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."

[0451] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. The program of the present invention is made up of a series of processes that are carried out between a server, a terminal, and a user.

[0452] System Configuration

[0453] 1. Users submit advertising content

[0454] A user uses the ad management system interface to input ad content, which may be in the form of text, images, or video.

[0455] 2. Submitting Content

[0456] The device sends the user-entered advertising content to the server via an HTTPS request, including the advertising content and associated metadata (e.g., advertiser ID, category, target audience, etc.).

[0457] 3. Content analysis and generation of prompts

[0458] The server analyzes the received advertising content and generates a prompt using generative AI. Natural language processing (NLP), image recognition, or video analysis techniques are used for the analysis. For example, specific keywords or phrases in the advertising content, parameters such as "weight loss in a short time" or "special price" are extracted.

[0459] 4. Generating Similar Content

[0460] The server uses generative AI to generate new advertising content based on the generative prompts, with the generated content having similar characteristics to the original advertising content.

[0461] 5. Content Comparison and Similarity Evaluation

[0462] The server compares the original advertising content with the generated similar content. This comparison evaluates the similarity using natural language processing, image recognition, or video analysis technology. A similarity score is calculated as the evaluation result.

[0463] 6. Determining harmful advertising

[0464] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold, the advertisement is recognized as harmful. This determination is made automatically.

[0465] 7. Block harmful ads and notify users

[0466] The server blocks ads that are deemed harmful and notifies the user based on that information. The blocked ads are then stopped from being displayed, and the user is notified by email or other means.

[0467] Specific examples

[0468] For example, if a user types in an ad for a "new diet supplement":

[0469] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0470] When a user submits advertising content, the following steps are performed:

[0471] 1. User enters advertising content:

[0472] The user enters the above text into the advertisement management system and clicks the send button.

[0473] 2. The device sends the advertising content:

[0474] The user's terminal transmits the advertising content and associated metadata to the server.

[0475] 3. The server analyzes the ad content:

[0476] The server receives the advertising content and uses natural language processing techniques to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[0477] 4. Generate content based on a generation prompt:

[0478] The server uses generation AI to generate advertising content with similar content.

[0479] 5. The server compares the original ad with the generated ad:

[0480] The server evaluates the similarity between the original advertising content and the generated advertising content.

[0481] 6. The server evaluates the similarity score:

[0482] If the similarity score is high, the ad is determined to be harmful.

[0483] 7. The server blocks harmful ads and notifies the user:

[0484] Advertisements that are determined to be harmful are blocked, and the user is notified that "the advertisement has been determined to be harmful and has been blocked."

[0485] Through this process, the system can efficiently detect and block malicious and harmful ads, while providing appropriate feedback to users, ensuring transparency.

[0486] The processing flow will be explained below.

[0487] Step 1:

[0488] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[0489] Step 2:

[0490] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[0491] Step 3:

[0492] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[0493] Step 4:

[0494] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[0495] Step 5:

[0496] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[0497] Step 6:

[0498] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[0499] Step 7:

[0500] The server determines whether an advertisement is harmful based on the similarity score. If the similarity score exceeds a preset threshold, the server determines the advertisement as harmful. Conversely, if the similarity score is below the threshold, the advertisement is not determined to be harmful.

[0501] Step 8:

[0502] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[0503] Step 9:

[0504] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block and the similarity score.

[0505] These steps enable the system of the present invention to efficiently detect, evaluate, and block harmful advertising content, and provide appropriate feedback to the user.

[0506] Example 1

[0507] 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."

[0508] Conventional ad management systems lacked the ability to automatically detect and block harmful ad content entered by users. This resulted in a high risk of harmful ads being displayed and made it difficult to maintain ad quality. Feedback to users was also insufficient, lacking transparency. To solve these problems, a system that could automatically detect and block harmful ads with high accuracy and efficiency was needed.

[0509] 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.

[0510] In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate similarity, means for determining that the advertising content is harmful if the similarity exceeds a certain threshold, means for blocking the harmful advertising based on the determination result and notifying the user, means for transmitting information including metadata related to the advertising content, means for extracting specific keywords or phrases based on the received advertising content, and means for calculating the similarity of the generated advertising content. This effectively solves the problems of detecting and blocking harmful advertising that conventional systems have had, improves the quality of advertising, and enables transparent feedback to users.

[0511] "Advertising Content" means information that a user inputs into the Ad Management System, and may be provided in the form of text, images, or video.

[0512] A "generative prompt" is input data that allows a generative AI model to generate new advertising content based on information extracted by analyzing received advertising content.

[0513] A "generative AI model" is a machine learning model for automatically generating new advertising content based on generative prompts.

[0514] The "similarity" refers to the degree of similarity between the original advertising content and the generated advertising content, and is expressed as a numerical value.

[0515] The "threshold" refers to a boundary value that is the standard for determining whether the similarity is a harmful advertisement.

[0516] A "harmful ad" is an ad whose content has a similarity score that exceeds a certain threshold and has been automatically determined not to be displayed.

[0517] "Metadata" is additional information related to advertising content, including advertiser ID, category, target audience, etc.

[0518] "Natural language processing technology" is a technology for analyzing text-based content and extracting specific keywords and phrases.

[0519] "Image recognition technology" is a technology for analyzing image-based content and extracting specific features and patterns.

[0520] "Video analysis technology" is a technology for analyzing video-based content and extracting specific frames or scenes.

[0521] An "HTTPS request" is a secure communication protocol used when exchanging data over the Internet.

[0522] "Means for receiving" refers to hardware or software for receiving advertising content from a user.

[0523] "Means for analyzing" refers to hardware or software for examining received advertising content and extracting necessary information.

[0524] The "means for evaluating" refers to hardware or software for measuring the similarity between the original advertising content and the generated content and evaluating it as a numerical value.

[0525] "Blocking measures" refers to hardware or software that stops the display of advertising content that is determined to be harmful.

[0526] The "notification means" refers to hardware or software for notifying the user of the determination result.

[0527] The present invention is a system that automatically analyzes advertising content entered by users and detects and blocks harmful advertisements as necessary. The system configuration includes a series of processes performed between a server, a terminal, and a user. The following describes the specific program processing for implementing the present invention and its detailed explanation.

[0528] Program processing explanation

[0529] The system has the following main features:

[0530] 1. Receiving advertising content: The user enters advertising content using the advertising management system interface. The entered advertising content can be in the form of text, images, or videos, and is sent by the device to the server using an HTTPS request.

[0531] 2. Analyzing advertising content and generating a generative prompt: The server analyzes the received advertising content. This analysis uses natural language processing (NLP), image recognition, or video analysis technology. For example, specific keywords or phrases in the advertising content, or parameters such as "fast weight loss" or "special price offer," are extracted. A generative prompt is generated based on the extracted information.

[0532] 3. Generating similar content based on the prompt: The server generates new advertising content using a generative AI model (e.g., GPT-4) based on the prompt. The generated content has similar characteristics to the original advertising content.

[0533] 4. Comparison of advertising content and similarity evaluation: The server compares the original advertising content with the newly generated advertising content and evaluates their similarity. This evaluation uses natural language processing technology, image recognition technology, or video analysis technology, and calculates a similarity score as the evaluation result.

[0534] 5. Identifying and blocking harmful ads: The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold, the ad is recognized as harmful. This determination is made automatically. Based on the determination result, harmful ads are blocked and the user is notified. This notification is made by means such as email.

[0535] Specific examples

[0536] For example, consider the following ad content for a "new diet supplement" entered by a user:

[0537] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0538] In this case, the system does the following:

[0539] Receiving advertising content: The user enters the above text into the advertising management system and clicks the submit button.

[0540] Content submission: The user's device submits the ad content and metadata to the server using an HTTP POST request.

[0541] Content analysis: The server uses text analysis tools to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[0542] Similar content generation: The server inputs the generation prompt into the generative AI model to generate new advertising content with similar content.

[0543] Similarity evaluation: The server evaluates the similarity between the original advertising content and the newly generated advertising content and calculates a similarity score.

[0544] Identifying and blocking harmful ads: If the server determines that an ad is harmful based on the similarity score, it blocks the ad and notifies the user that "the ad has been identified as harmful and blocked."

[0545] Through these processes, the system can efficiently detect and block harmful ads, thereby improving the quality of ads and providing transparent feedback to users.

[0546] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0547] Step 1:

[0548] User enters advertising content

[0549] Specific operation: The user accesses the interface of the ad management system and inputs the ad content. The ad content can be input in the form of text, image, or video. The user clicks the submit button to submit the content.

[0550] Input: Ad content in the form of text, images, or videos

[0551] Output: Clicking the submit button sends the ad content to the device.

[0552] Step 2:

[0553] The device sends advertising content

[0554] How it works: The user's device sends the entered advertising content to the server using an HTTPS request, along with metadata such as the advertiser's ID, category, and target audience.

[0555] Input: User-entered ad content and metadata

[0556] Output: Ad content and metadata sent to the server as an HTTPS request

[0557] Step 3:

[0558] The server analyzes the ad content

[0559] How it works: The server uses natural language processing (NLP), image recognition, and video analysis techniques to analyze the received ad content. For example, in the case of a text ad, keywords and phrases such as "fast weight loss" or "special offer" are extracted.

[0560] Input: Ad content and metadata received as an HTTPS request

[0561] Output: Generated prompts containing extracted keywords and phrases

[0562] Step 4:

[0563] The server generates similar content based on the prompt.

[0564] How it works: The server generates new ad content using a generative AI model (e.g., GPT-4) based on the prompt. The generated ad content has similar characteristics to the original ad content.

[0565] Input: Generate prompt

[0566] Output: New ad content generated by the generative AI model

[0567] Step 5:

[0568] The server compares the original ad content with the generated ad content

[0569] Specific operation: The server uses natural language processing, image recognition, and video analysis technologies to evaluate the similarity between the original advertising content and the generated advertising content. This evaluation calculates a similarity score.

[0570] Input: Original and generated ad content

[0571] Output: Similarity score

[0572] Step 6:

[0573] The server determines harmful ads

[0574] How it works: The server compares the similarity score with a pre-defined threshold, and if the score exceeds the threshold, it marks the ad as harmful. This decision is made automatically.

[0575] Input: Similarity score

[0576] Output: Harmful ad detection result

[0577] Step 7:

[0578] The server blocks harmful ads and notifies the user.

[0579] Specific operation: The server blocks ads that are determined to be harmful and notifies the user based on that information. Notifications are sent via email or other means.

[0580] Input: Harmful ad verdict

[0581] Output: Blocked ads and notification to the user

[0582] (Application example 1)

[0583] 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."

[0584] Advertisements provided on the Internet often contain harmful content. This leads to the distribution of unreliable advertisements and advertisements that may have a negative impact on users. In particular, the display of such harmful advertisements on smartphone and PC applications has a significant impact on the user experience. The present invention aims to provide a system that efficiently and automatically detects and blocks such harmful advertisements.

[0585] 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.

[0586] In this invention, the server includes: means for receiving advertising content input by a user; means for analyzing the received advertising content and generating a generation prompt; means for generating new content using the generation prompt; means for comparing the original advertising content with the generated content and evaluating the similarity; means for determining that the generated content is a harmful advertisement if the similarity exceeds a certain threshold; means for blocking the harmful advertisement based on the evaluation result and notifying the user; means for evaluating the similarity between the generated content and the original advertising content and determining that the advertisement is harmful based on the evaluation result; and means for notifying the user that the advertisement is blocked if it is determined to be harmful. This makes it possible to automatically determine whether advertising content created by an advertiser is harmful and to quickly block content that is determined to be harmful.

[0587] "User" means the person or entity that creates and inputs advertising content into the system.

[0588] "Advertising Content" means advertising information provided in text, image, or video format.

[0589] The "means for receiving" refers to a method or device for capturing the advertising content entered by the user into the server.

[0590] "Means for analyzing" refers to a method or device for analyzing received advertising content to create generated prompts.

[0591] "Generative prompts" are input data that a generative AI model uses to create new content based on the analysis results.

[0592] A "generating means" is a method or device that uses a generating prompt to create new content.

[0593] The "means for comparing and assessing similarity" refers to a method or device for calculating the degree of similarity in content or characteristics between the original advertising content and the generated content.

[0594] The "certain threshold" is a standard value beyond which the similarity is judged to be harmful advertising.

[0595] "Means for determining" refers to a method or device that determines whether advertising content is harmful based on the similarity assessment.

[0596] "Blocking means" refers to a method or device that stops the display or distribution of advertising content that has been determined to be harmful advertising.

[0597] "Means for notifying" refers to a method or device for informing the user of the advertisements that have been blocked and the reasons for their blocking.

[0598] The present invention is a system for receiving advertising content entered by a user and automatically detecting and blocking harmful advertisements. The system is composed of a user terminal, a server, and a program that performs a series of processes between the user and the user.

[0599] System configuration

[0600] 1. User Device

[0601] The user inputs advertising content using the ad management system interface. The input advertising content can be in the form of text, images, or videos. The user device then sends the advertising content and related metadata (such as the advertiser's ID, category, and target audience) to the server. The transmission uses an HTTPS request to ensure data security.

[0602] 2. Server

[0603] The server analyzes the advertising content received from the user's device. For the analysis, it uses natural language processing (NLP), image recognition, and video analysis technologies, and generates a generated prompt using a generative AI model. The generated prompt is created based on the main keywords and phrases in the analyzed advertising content. For example, "weight loss effect in a short time," "special price," and "innovative supplement" are extracted as generated prompts.

[0604] Next, the server generates new advertising content using a generative AI model based on the generated prompts. The generated advertising content has similar characteristics to the original advertising content. The server compares the original advertising content with the generated advertising content and evaluates their similarity. This evaluation is performed using natural language processing technology, image recognition technology, and video analysis technology. A similarity score is calculated as the evaluation result.

[0605] 3. Determining and notifying harmful advertising

[0606] The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold (e.g., 0.8), the ad is recognized as harmful. This determination is made automatically. If the ad is determined to be harmful, the server blocks the ad and notifies the user. The notification is sent by email, in-app message, or other means.

[0607] Specific examples

[0608] For example, a user enters the following ad content:

[0609] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0610] For this ad content, the following prompt text is generated:

[0611] "Fast weight loss," "Special price," "Innovative supplement"

[0612] The server generates new advertising content based on these prompts and compares it with the original advertising content to evaluate its similarity. If the similarity score exceeds a threshold, the server determines the advertising content to be harmful and notifies the user that it is blocked.

[0613] Hardware and software used

[0614] User device: Inputs and transmits advertising content. Examples include smartphones and PCs.

[0615] Server: Performs reception, analysis, generation, judgment, and notification processes. Uses a cloud server (e.g., Amazon Web Services, Google Cloud Platform, etc.).

[0616] Natural language processing technologies: such as the Hugging Face transformers library and the BERT model.

[0617] Image recognition technology: Uses OpenCV and TensorFlow.

[0618] Video analysis technology: Uses FFmpeg and OpenCV.

[0619] Generative AI models: Large-scale language models such as GPT-3.

[0620] In this way, the present invention can automatically analyze advertising content, effectively block harmful advertisements, and provide appropriate feedback to users.

[0621] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0622] Step 1:

[0623] The user inputs advertising content using the interface of the advertising management system. The input advertising content can be in the form of text, images, or videos. The user terminal then sends the input advertising content and related metadata (such as advertiser ID, category, target audience, etc.) to the server using an HTTPS request.

[0624] Input: Ad content, associated metadata

[0625] Output: Ad content and associated metadata included in the HTTPS request

[0626] Step 2:

[0627] The server receives the advertising content and associated metadata sent from the user device. The server then analyzes the advertising content using natural language processing (NLP), image recognition, or video analysis techniques to generate a generated prompt that includes key keywords or phrases from the analyzed advertising content (e.g., "fast weight loss," "special price," "innovative supplement").

[0628] Input: Ad content, associated metadata

[0629] Output: Generated prompt

[0630] Step 3:

[0631] The server generates new advertising content using a generative AI model based on the prompts. The generative AI model is a large-scale language model, such as GPT-3, that receives the prompts as input and generates new advertising content based on them. The generated advertising content has similar characteristics to the original advertising content.

[0632] Input: Generate prompt

[0633] Output: Generated ad content

[0634] Step 4:

[0635] The server compares the original advertising content with the generated advertising content and evaluates the similarity. This similarity evaluation uses natural language processing technology, image recognition technology, or video analysis technology. For example, the server compares the text portions of the original advertising content with the generated advertising content and calculates a similarity score.

[0636] Input: Original ad content, Generated ad content

[0637] Output: Similarity score

[0638] Step 5:

[0639] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold (e.g., 0.8), the advertisement is recognized as harmful. This determination is made automatically.

[0640] Input: Similarity score

[0641] Output: Harmful ad detection result

[0642] Step 6:

[0643] The server blocks advertising content that is determined to be harmful within the system, thereby preventing users from viewing ads that are recognized as harmful. The server also notifies the user that the ad has been determined to be harmful and blocked. This notification is made by means of email, in-app message, etc.

[0644] Input: Harmful advertisement judgment result

[0645] Output: Block execution, notification message

[0646] This series of processes ensures the safety of advertising content and improves the user experience.

[0647] 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.

[0648] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. Furthermore, the present invention incorporates an emotion engine that recognizes the user's emotions, thereby evaluating the emotional impact of advertisements and improving the accuracy of determining harmful advertisements. The specific program processing flow is explained below.

[0649] System Configuration

[0650] 1. Users submit advertising content

[0651] The user uses the ad management system interface to input ad content, which can be in the form of text, images, or videos, and then clicks the "Submit" button.

[0652] 2. Submitting Content

[0653] The device sends the advertising content entered by the user and its associated metadata (e.g., advertiser ID, category, target audience) to the server via an HTTPS request.

[0654] 3. Content analysis and generation of prompts

[0655] The server analyzes the received advertising content and generates prompts using generative AI, which uses natural language processing, image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features.

[0656] 4. User Emotion Analysis

[0657] The server uses an emotion engine to analyze the user's emotions toward the received advertising content. Sentiment analysis extracts emotions, such as positive, negative, or neutral, from text, audio, or image data.

[0658] 5. Generating Similar Content

[0659] Based on the generated prompts, the server uses generative AI to generate new ad content, the format of which matches the original ad content (text, image, or video).

[0660] 6. Content Comparison and Similarity Evaluation

[0661] The server compares the original advertising content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology.

[0662] 7. Identifying Harmful Advertisements Using Sentiment Analysis Results

[0663] The server combines the similarity score and the results of user sentiment analysis to determine whether an advertisement is harmful. If the similarity score and the results of sentiment analysis exceed a certain threshold, the advertisement is determined to be harmful.

[0664] 8. Block harmful ads and notify users

[0665] The server blocks the ads that are determined to be harmful, updates the status of the ads to "blocked," and saves it in the database. At the same time, it automatically sends a notification email to the user informing them that the ads have been determined to be harmful and blocked.

[0666] Specific examples

[0667] For example, suppose a user enters an advertisement for a "new diet supplement" as follows:

[0668] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0669] The specific processing flow when this advertisement is submitted by a user is as follows.

[0670] 1. User enters advertising content:

[0671] The user enters the above text into the form of the ad management system and clicks the "Submit" button.

[0672] 2. The device sends the advertising content:

[0673] The user's device sends the advertising content and associated metadata to the server via an HTTPS request.

[0674] 3. The server analyzes the ad content:

[0675] A server receives the advertising content and extracts specific keywords or phrases to generate generated prompts.

[0676] 4. The server analyzes the user's emotions:

[0677] The server uses an emotion engine to analyze the emotional impact of advertising content on the user.

[0678] 5. Generate content based on a generation prompt:

[0679] The server uses generative AI to generate similar advertising content.

[0680] 6. The server compares the original ad with the generated ad:

[0681] The server evaluates the similarity between the generated advertising content and the original advertisement and calculates a similarity score.

[0682] 7. The server uses the sentiment analysis results to determine harmful ads:

[0683] The server combines the similarity score with the sentiment analysis results to determine whether the ad is harmful.

[0684] 8. The server blocks harmful ads and notifies the user:

[0685] The server blocks ads that are determined to be harmful and notifies the user that "the ad has been determined to be harmful and has been blocked."

[0686] By taking into account the results of user sentiment analysis, the system of the present invention can detect harmful advertisements with higher accuracy than conventional methods that only use similarity evaluation. In addition, appropriate feedback is provided to users, ensuring transparency.

[0687] The processing flow will be explained below.

[0688] Step 1:

[0689] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[0690] Step 2:

[0691] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[0692] Step 3:

[0693] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[0694] Step 4:

[0695] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[0696] Step 5:

[0697] The server analyzes the user's emotions. The server uses an emotion engine to classify the user's emotions toward the advertising content as positive, negative, or neutral. This can be done using text, voice, or image analysis.

[0698] Step 6:

[0699] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[0700] Step 7:

[0701] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[0702] Step 8:

[0703] The server integrates the similarity score and the user's sentiment analysis results, and performs a comprehensive evaluation by combining the similarity score and the sentiment analysis results.

[0704] Step 9:

[0705] The server judges harmful advertisements. If the overall evaluation based on the similarity and sentiment analysis results exceeds a certain threshold, the advertisement is judged as harmful. Conversely, if it is below the threshold, the advertisement is not judged as harmful.

[0706] Step 10:

[0707] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[0708] Step 11:

[0709] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block, the similarity score, and the results of sentiment analysis.

[0710] These steps enable the system of the present invention to not only efficiently detect and block harmful advertising content, but also provide feedback to help users understand the process.

[0711] Example 2

[0712] 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."

[0713] Conventional advertising content filtering systems have limited accuracy in determining whether an ad's content is harmful, leading to false positives. Furthermore, because they do not consider users' emotional responses, they can display emotionally unpleasant ads. This can lead to a poor user experience and a loss of trust in advertisers.

[0714] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0715] In this invention, the server includes a means for analyzing a user's emotions, a means for analyzing received content and generating a generation prompt, and a means for generating new content using the generation prompt, thereby enabling highly accurate determination of harmful content that takes user emotions into consideration.

[0716] "User" means a person who inputs and transmits advertising content using the system.

[0717] "Advertising Content" means a collection of information entered by a user that is analyzed and evaluated by the system, and may take the form of text, images, or video.

[0718] The "means for receiving" is a part of the system that has the functionality for capturing advertising content sent by a user.

[0719] The "means for analyzing" is a part of the system that has the function of analyzing received advertising content and extracting useful information.

[0720] A "generative prompt" is input information given to a generative AI model based on the analysis results of advertising content.

[0721] A "means for generating" is a part of the system that has the ability to create new content using generation prompts.

[0722] A "comparison means" is a part of the system that has the ability to evaluate the original advertising content against the generated content.

[0723] The "means for assessing similarity" is a part of the system that has the function of quantifying the similarity between the original advertising content and the generated content.

[0724] The "means for determining that content is harmful" is a part of the system that has the function of determining whether advertising content is harmful based on the results of similarity evaluation and sentiment analysis.

[0725] "Blocking means" is a part of the system that has the function of preventing advertising content that is determined to be harmful from being displayed.

[0726] The "means for notifying the user" is a part of the system that has the function of notifying the user that advertising content has been determined to be harmful.

[0727] The "means for analyzing emotions" is a part of the system that has the function of evaluating the emotional impact that advertising content has on the user and extracting emotions such as positive, negative, or neutral.

[0728] "Natural language processing technology" is a technology for analyzing the meaning and content of text data.

[0729] "Image recognition technology" is a technology for analyzing the contents of image data.

[0730] "Video analysis technology" is a technology for analyzing the content of video data.

[0731] The present invention is a system that receives advertising content entered by a user and automatically determines whether that content is harmful and blocks it. Furthermore, by analyzing user sentiment, the accuracy of determining harmful advertisements is improved. Below, specific embodiments of the present invention are described.

[0732] System Configuration

[0733] User Roles

[0734] The user inputs the advertising content through the advertising management system. The advertising content can be provided in the form of text, images, or videos. Once the user inputs the advertising content and clicks the "Submit" button, the content is received by the system.

[0735] Device Role

[0736] When a user submits advertising content, the device sends the entered advertising content and related metadata (e.g., advertiser ID, category, target audience) to the server using an HTTPS request. The device uses an existing web browser or a dedicated application.

[0737] Server Roles

[0738] The server receives the advertising content sent by the user and performs the following processes.

[0739] 1. Advertising content analysis:

[0740] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies. Specifically, it uses open source NLP libraries (e.g., spaCy, NLTK), image recognition libraries (e.g., OpenCV, TensorFlow), and video analysis technologies (e.g., FFmpeg).

[0741] Specific keywords, phrases, and visual features are extracted from the analyzed data, and generative prompt sentences are generated based on these.

[0742] 2. User sentiment analysis:

[0743] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the emotional impact of the ad content on the user, categorizing emotions into positive, negative, and neutral categories.

[0744] 3. Generate new content based on a generation prompt:

[0745] The server uses a generative AI model (e.g., GPT-4, DALL-E) to generate new ad content based on the generated prompt. The generated content matches the original format (text, image, video).

[0746] 4. Comparison of original ad content and generated content:

[0747] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score.

[0748] 5. Determining harmful advertising:

[0749] The server determines whether an advertisement is harmful based on the similarity score and the result of sentiment analysis. If the similarity score or sentiment score exceeds a certain threshold, the advertisement is determined to be harmful.

[0750] 6. Block harmful ads and notify users:

[0751] The server blocks the ad that is determined to be harmful, updates the ad's status to "blocked," and notifies the user by email that the ad has been determined to be harmful and blocked.

[0752] Specific examples

[0753] For example, if a user enters an ad for a "new diet supplement" like this:

[0754] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0755] Once this ad content is sent, the server performs the following steps:

[0756] 1. User enters and submits ad:

[0757] The user enters text into the ad management system and clicks the "Submit" button.

[0758] 2. The device sends the advertising content:

[0759] The device sends advertising content to the server via an HTTPS request.

[0760] 3. The server analyzes the advertising content:

[0761] Keywords such as "innovative," "weight loss effect," and "special price" are extracted, and prompt sentences are generated to be given to the generative AI model.

[0762] 4. The server analyzes the user's emotions:

[0763] Using an emotion engine, the emotions that ad text evokes in users are analyzed, and an emotion score is obtained, such as 80% positive, 15% negative, and 5% neutral.

[0764] 5. Generate new content based on a generation prompt:

[0765] A prompt sentence is input into the generative AI model to generate new advertising text, such as "Why not lose weight in a healthy way with this supplement?"

[0766] 6. Server compares original ad with new ad:

[0767] The original ad and the new ad are compared using natural language processing and a similarity score (e.g., 85 points) is calculated.

[0768] 7. The server determines whether the ad is harmful:

[0769] Advertisements are analyzed based on similarity scores and sentiment scores to determine whether they are harmful.

[0770] 8. Blocking harmful ads and notifying users:

[0771] Update the ad status to "Blocked" and notify the user that "The ad has been identified as harmful and blocked."

[0772] The system of the present invention enables automatic detection and blocking of harmful advertisements with high accuracy, taking into account the user's emotional response, thereby improving the user experience and ensuring advertising safety.

[0773] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0774] Step 1:

[0775] User enters and submits advertising content

[0776] The user uses the interface of the ad management system to input the ad content (in the form of text, images, or videos). Then, he clicks the "Submit" button to send the input content. The input data is the ad content provided by the user, and the output data is the ad content sent to the terminal. Specifically, the user attaches text, images, or videos to a form and submits it.

[0777] Step 2:

[0778] The device sends advertising content to the server

[0779] The device sends the advertising content and related metadata entered by the user to the server using an HTTPS request. The input data is the content and metadata entered by the user, and the output data is the content data sent to the server. Specifically, the device converts the user's input data into packets and sends them to the server as an encrypted HTTPS request.

[0780] Step 3:

[0781] The server analyzes the ad content and generates a prompt

[0782] The server analyzes the received advertising content. It uses natural language processing techniques (e.g., spaCy, NLTK), image recognition techniques (e.g., OpenCV, TensorFlow), and video analysis techniques (e.g., FFmpeg) to extract specific keywords, phrases, and visual features from the content. The input data is the advertising content sent to the server, and the output data is the generated prompt. Specifically, the server runs the analysis engine to extract keywords such as "weight loss" and "special price" from the original content and compose the generated prompt.

[0783] Step 4:

[0784] The server analyzes the user's emotions

[0785] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the user's emotions toward the ad content. The input data here is the analyzed ad content, and the output data is an emotion score such as positive, negative, or neutral. Specifically, the server passes the content data to the emotion analysis module and obtains the emotion score.

[0786] Step 5:

[0787] Generate new content based on a prompt

[0788] The server uses the generation prompt sentence to generate new advertising content for a generative AI model (e.g., GPT-4, DALL-E). The input data is the generation prompt sentence, and the output data is the generated new content. Specifically, the server inputs the generation prompt into the AI ​​model, and the model generates new advertising content (e.g., text and images).

[0789] Step 6:

[0790] The server compares the original ad content with the new content

[0791] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score. The input data is the original advertising content and the generated new content, and the output data is the similarity score. Specifically, the server inputs both contents into an analysis engine and quantifies the similarity.

[0792] Step 7:

[0793] The server determines harmful ads

[0794] The server determines whether an ad is harmful based on the similarity score and the results of sentiment analysis. The input data are the similarity score and sentiment score, and the output data is the harmfulness determination result. Specifically, the server compares the obtained scores and determines that an ad is harmful if it exceeds a threshold.

[0795] Step 8:

[0796] The server blocks harmful ads and notifies the user

[0797] The server blocks ads that are determined to be harmful and updates the ad status to "blocked." It also automatically sends a notification email to the user informing them that the ad has been determined to be harmful and blocked. The input data is the harmful determination result, and the output data is the updated ad status and notification email. Specifically, the server updates the ad database and sends a notification email to the user.

[0798] (Application example 2)

[0799] 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."

[0800] With the spread of internet advertising, the risk of harmful advertising content adversely affecting users is increasing. Therefore, a system that automatically monitors advertisements and quickly detects and blocks harmful advertisements is needed. Furthermore, conventional systems that only use similarity evaluation are sometimes insufficient in accuracy, so there is a need for a highly accurate harmful advertisement detection system that takes into account the emotional impact of users.

[0801] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content and generating a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content and evaluating the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user. This makes it possible to detect harmful advertisements quickly and accurately, and maintain the integrity of advertisements while minimizing the adverse impact on users.

[0802] "User-entered advertising content" refers to data in the form of text, images, or videos of advertisements provided by users such as advertisers or advertising agencies through online platforms.

[0803] The "receiving means" refers to a communication interface and software that has the function of receiving the advertising content input by the user on the server side.

[0804] "Means for analyzing and generating generative prompts" refers to the function of analyzing received advertising content and creating instructions (prompts) for generating new content using generative AI based on that content.

[0805] "Means for generating new content using generative prompts" refers to the ability to automatically generate new advertising content based on generative prompts using a generative AI model.

[0806] "Means for comparing generated content and assessing similarity" refers to algorithms and technologies that match original advertising content with generated advertising content and measure their similarity.

[0807] "Means for recognizing user emotions and assessing the emotional impact of advertising" refers to a sentiment analysis engine and associated algorithms for analyzing the emotional response of advertising content to users and assessing its impact.

[0808] The term "certain threshold" refers to a predetermined reference value for determining whether an advertisement is harmful or not, based on the results of similarity assessment and sentiment analysis.

[0809] "Means for determining harmful advertising" refers to an algorithm that automatically determines that advertising content is harmful if the results of similarity assessment and sentiment analysis exceed a certain threshold.

[0810] "Means for blocking and notifying users" refers to the function of excluding advertising content that is determined to be harmful from display and distribution networks, and automatically notifying advertisers of this fact.

[0811] This invention relates to an automated system for accurately determining the harmfulness of advertising content entered by a user and blocking it. The system of the present invention mainly includes the following main means: means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user.

[0812] Program processing explanation

[0813] 1. Receiving advertising content

[0814] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. After entering the content, the user clicks the "Submit" button, which sends the content and its metadata (advertiser ID, category, target audience) to the server. This communication uses an HTTPS request.

[0815] 2. Analyzing advertising content and generating prompts

[0816] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features. Based on the analysis results, a prompt is generated, and new advertising content is automatically generated based on the advertising content.

[0817] 3. Emotion analysis

[0818] The server evaluates the emotional impact of advertising content using a sentiment analysis engine that uses existing sentiment analysis software, such as NLTK's SentimentIntensityAnalyzer, to extract positive, negative, or neutral sentiment scores from text, audio, or image data.

[0819] 4. Similarity evaluation

[0820] The server compares the original ad content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology. For example, it can use Hugging Face's transformer library to calculate the similarity between the similar content generated by the GPT-3 model and the original content.

[0821] 5. Identifying and blocking harmful ads

[0822] The server combines the results of the similarity assessment and sentiment analysis to determine whether an ad is harmful. Ads that are determined to be harmful are automatically blocked and users are notified of this. Notifications are sent via email or system alerts.

[0823] Specific examples

[0824] For example, if a user enters the following ad content:

[0825] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0826] When this ad is submitted, the system works as follows: First, the ad content is received, the text is analyzed, and a generative prompt is generated. Using the generative prompt, a generative AI model (GPT-3) generates similar content, such as:

[0827] "Just take this new supplement and lose weight fast! Try it at a great price!"

[0828] The sentiment analysis engine then evaluates the emotional impact and calculates the similarity between the original content and the generated content. Based on this result, it determines whether the ad is harmful, and if so, the ad is blocked and the user is notified. This allows for early detection of harmful content and prevents adverse effects on users.

[0829] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0830] Step 1:

[0831] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. The user clicks the "Submit" button, which sends the entered content and its metadata (advertiser ID, category, target audience) to the server. The entered data is securely transmitted to the server via an HTTPS request.

[0832] Step 2:

[0833] The server analyzes the received advertising content. First, it performs text analysis and uses natural language processing (NLP) techniques to extract specific keywords and phrases within the content. It also uses image and video analysis techniques to extract visual features. These analyses result in a generated prompt. The input is the advertising content and related metadata, and the output is the analysis results and the generated prompt.

[0834] Step 3:

[0835] The server uses the generative prompts to generate new advertising content. A generative AI model (e.g., GPT-3) is used to create similar advertising content based on the generative prompts. The generated content is in the same format (text, image, or video) as the original. The input is the generative prompts, and the output is the generated advertising content.

[0836] Step 4:

[0837] The server uses a sentiment analysis engine to evaluate the emotional impact of ad content on users. Sentiment analysis software such as NLTK's SentimentIntensityAnalyzer extracts positive, negative, or neutral sentiment scores from text, audio, or image data. The input is the ad content, and the output is the sentiment analysis results.

[0838] Step 5:

[0839] The server compares the generated ad content with the original ad content and calculates a similarity score. It uses natural language processing, image recognition, or video analysis technology to evaluate the similarity based on the features of both pieces of content. For example, it uses Hugging Face's transformers library to compare it with the output of the GPT-3 model. The input is the original ad content and the generated content, and the output is a similarity score.

[0840] Step 6:

[0841] The server determines whether an advertisement is harmful or not based on the similarity score and the results of sentiment analysis. The criteria (certain threshold) for determining whether an advertisement is harmful is set in advance, and if the similarity score and the results of sentiment analysis exceed this threshold, the advertisement is determined to be harmful. The input is the similarity score and the results of sentiment analysis, and the output is the result of determining whether the advertisement is harmful.

[0842] Step 7:

[0843] The server blocks advertising content that is determined to be harmful. It updates the status of the advertisement to "blocked" and saves it in the database. It also automatically sends a notification to the user that the advertisement has been determined to be harmful and blocked. The input is the result of determining whether the advertisement is harmful, and the output is the notification to the user and the blocking status of the advertising content.

[0844] Through this series of processes, the system can accurately determine whether an advertisement is harmful and prevent the distribution of harmful advertising content.

[0845] 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.

[0846] 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.

[0847] 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.

[0848] [Third embodiment]

[0849] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0850] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0851] 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).

[0852] 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.

[0853] 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.

[0854] 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).

[0855] 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.

[0856] 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.

[0857] 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.

[0858] 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.

[0859] 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.

[0860] 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."

[0861] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. The program of the present invention is made up of a series of processes that are carried out between a server, a terminal, and a user.

[0862] System Configuration

[0863] 1. Users submit advertising content

[0864] A user uses the ad management system interface to input ad content, which may be in the form of text, images, or video.

[0865] 2. Submitting Content

[0866] The device sends the user-entered advertising content to the server via an HTTPS request, including the advertising content and associated metadata (e.g., advertiser ID, category, target audience, etc.).

[0867] 3. Content analysis and generation of prompts

[0868] The server analyzes the received advertising content and generates a prompt using generative AI. Natural language processing (NLP), image recognition, or video analysis techniques are used for the analysis. For example, specific keywords or phrases in the advertising content, parameters such as "weight loss in a short time" or "special price" are extracted.

[0869] 4. Generating Similar Content

[0870] The server uses generative AI to generate new advertising content based on the generative prompts, with the generated content having similar characteristics to the original advertising content.

[0871] 5. Content Comparison and Similarity Evaluation

[0872] The server compares the original advertising content with the generated similar content. This comparison evaluates the similarity using natural language processing, image recognition, or video analysis technology. A similarity score is calculated as the evaluation result.

[0873] 6. Determining harmful advertising

[0874] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold, the advertisement is recognized as harmful. This determination is made automatically.

[0875] 7. Block harmful ads and notify users

[0876] The server blocks ads that are deemed harmful and notifies the user based on that information. The blocked ads are then stopped from being displayed, and the user is notified by email or other means.

[0877] Specific examples

[0878] For example, if a user types in an ad for a "new diet supplement":

[0879] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0880] When a user submits advertising content, the following steps are performed:

[0881] 1. User enters advertising content:

[0882] The user enters the above text into the advertisement management system and clicks the send button.

[0883] 2. The device sends the advertising content:

[0884] The user's terminal transmits the advertising content and associated metadata to the server.

[0885] 3. The server analyzes the ad content:

[0886] The server receives the advertising content and uses natural language processing techniques to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[0887] 4. Generate content based on a generation prompt:

[0888] The server uses generation AI to generate advertising content with similar content.

[0889] 5. The server compares the original ad with the generated ad:

[0890] The server evaluates the similarity between the original advertising content and the generated advertising content.

[0891] 6. The server evaluates the similarity score:

[0892] If the similarity score is high, the ad is determined to be harmful.

[0893] 7. The server blocks harmful ads and notifies the user:

[0894] Advertisements that are determined to be harmful are blocked, and the user is notified that "the advertisement has been determined to be harmful and has been blocked."

[0895] Through this process, the system can efficiently detect and block malicious and harmful ads, while providing appropriate feedback to users, ensuring transparency.

[0896] The processing flow will be explained below.

[0897] Step 1:

[0898] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[0899] Step 2:

[0900] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[0901] Step 3:

[0902] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[0903] Step 4:

[0904] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[0905] Step 5:

[0906] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[0907] Step 6:

[0908] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[0909] Step 7:

[0910] The server determines whether an advertisement is harmful based on the similarity score. If the similarity score exceeds a preset threshold, the server determines the advertisement as harmful. Conversely, if the similarity score is below the threshold, the advertisement is not determined to be harmful.

[0911] Step 8:

[0912] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[0913] Step 9:

[0914] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block and the similarity score.

[0915] These steps enable the system of the present invention to efficiently detect, evaluate, and block harmful advertising content, and provide appropriate feedback to the user.

[0916] Example 1

[0917] 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."

[0918] Conventional ad management systems lacked the ability to automatically detect and block harmful ad content entered by users. This resulted in a high risk of harmful ads being displayed and made it difficult to maintain ad quality. Feedback to users was also insufficient, lacking transparency. To solve these problems, a system that could automatically detect and block harmful ads with high accuracy and efficiency was needed.

[0919] 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.

[0920] In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate similarity, means for determining that the advertising content is harmful if the similarity exceeds a certain threshold, means for blocking the harmful advertising based on the determination result and notifying the user, means for transmitting information including metadata related to the advertising content, means for extracting specific keywords or phrases based on the received advertising content, and means for calculating the similarity of the generated advertising content. This effectively solves the problems of detecting and blocking harmful advertising that conventional systems have had, improves the quality of advertising, and enables transparent feedback to users.

[0921] "Advertising Content" means information that a user inputs into the Ad Management System, and may be provided in the form of text, images, or video.

[0922] A "generative prompt" is input data that allows a generative AI model to generate new advertising content based on information extracted by analyzing received advertising content.

[0923] A "generative AI model" is a machine learning model for automatically generating new advertising content based on generative prompts.

[0924] The "similarity" refers to the degree of similarity between the original advertising content and the generated advertising content, and is expressed as a numerical value.

[0925] The "threshold" refers to a boundary value that is the standard for determining whether the similarity is a harmful advertisement.

[0926] A "harmful ad" is an ad whose content has a similarity score that exceeds a certain threshold and has been automatically determined not to be displayed.

[0927] "Metadata" is additional information related to advertising content, including advertiser ID, category, target audience, etc.

[0928] "Natural language processing technology" is a technology for analyzing text-based content and extracting specific keywords and phrases.

[0929] "Image recognition technology" is a technology for analyzing image-based content and extracting specific features and patterns.

[0930] "Video analysis technology" is a technology for analyzing video-based content and extracting specific frames or scenes.

[0931] An "HTTPS request" is a secure communication protocol used when exchanging data over the Internet.

[0932] "Means for receiving" refers to hardware or software for receiving advertising content from a user.

[0933] "Means for analyzing" refers to hardware or software for examining received advertising content and extracting necessary information.

[0934] The "means for evaluating" refers to hardware or software for measuring the similarity between the original advertising content and the generated content and evaluating it as a numerical value.

[0935] "Blocking measures" refers to hardware or software that stops the display of advertising content that is determined to be harmful.

[0936] The "notification means" refers to hardware or software for notifying the user of the determination result.

[0937] The present invention is a system that automatically analyzes advertising content entered by users and detects and blocks harmful advertisements as necessary. The system configuration includes a series of processes performed between a server, a terminal, and a user. The following describes the specific program processing for implementing the present invention and its detailed explanation.

[0938] Program processing explanation

[0939] The system has the following main features:

[0940] 1. Receiving advertising content: The user enters advertising content using the advertising management system interface. The entered advertising content can be in the form of text, images, or videos, and is sent by the device to the server using an HTTPS request.

[0941] 2. Analyzing advertising content and generating a generative prompt: The server analyzes the received advertising content. This analysis uses natural language processing (NLP), image recognition, or video analysis technology. For example, specific keywords or phrases in the advertising content, or parameters such as "fast weight loss" or "special price offer," are extracted. A generative prompt is generated based on the extracted information.

[0942] 3. Generating similar content based on the prompt: The server generates new advertising content using a generative AI model (e.g., GPT-4) based on the prompt. The generated content has similar characteristics to the original advertising content.

[0943] 4. Comparison of advertising content and similarity evaluation: The server compares the original advertising content with the newly generated advertising content and evaluates their similarity. This evaluation uses natural language processing technology, image recognition technology, or video analysis technology, and calculates a similarity score as the evaluation result.

[0944] 5. Identifying and blocking harmful ads: The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold, the ad is recognized as harmful. This determination is made automatically. Based on the determination result, harmful ads are blocked and the user is notified. This notification is made by means such as email.

[0945] Specific examples

[0946] For example, consider the following ad content for a "new diet supplement" entered by a user:

[0947] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[0948] In this case, the system does the following:

[0949] Receiving advertising content: The user enters the above text into the advertising management system and clicks the submit button.

[0950] Content submission: The user's device submits the ad content and metadata to the server using an HTTP POST request.

[0951] Content analysis: The server uses text analysis tools to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[0952] Similar content generation: The server inputs the generation prompt into the generative AI model to generate new advertising content with similar content.

[0953] Similarity evaluation: The server evaluates the similarity between the original advertising content and the newly generated advertising content and calculates a similarity score.

[0954] Identifying and blocking harmful ads: If the server determines that an ad is harmful based on the similarity score, it blocks the ad and notifies the user that "the ad has been identified as harmful and blocked."

[0955] Through these processes, the system can efficiently detect and block harmful ads, thereby improving the quality of ads and providing transparent feedback to users.

[0956] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0957] Step 1:

[0958] User enters advertising content

[0959] Specific operation: The user accesses the interface of the ad management system and inputs the ad content. The ad content can be input in the form of text, image, or video. The user clicks the submit button to submit the content.

[0960] Input: Ad content in the form of text, images, or videos

[0961] Output: Clicking the submit button sends the ad content to the device.

[0962] Step 2:

[0963] The device sends advertising content

[0964] How it works: The user's device sends the entered advertising content to the server using an HTTPS request, along with metadata such as the advertiser's ID, category, and target audience.

[0965] Input: User-entered ad content and metadata

[0966] Output: Ad content and metadata sent to the server as an HTTPS request

[0967] Step 3:

[0968] The server analyzes the ad content

[0969] How it works: The server uses natural language processing (NLP), image recognition, and video analysis techniques to analyze the received ad content. For example, in the case of a text ad, keywords and phrases such as "fast weight loss" or "special offer" are extracted.

[0970] Input: Ad content and metadata received as an HTTPS request

[0971] Output: Generated prompts containing extracted keywords and phrases

[0972] Step 4:

[0973] The server generates similar content based on the prompt.

[0974] How it works: The server generates new ad content using a generative AI model (e.g., GPT-4) based on the prompt. The generated ad content has similar characteristics to the original ad content.

[0975] Input: Generate prompt

[0976] Output: New ad content generated by the generative AI model

[0977] Step 5:

[0978] The server compares the original ad content with the generated ad content

[0979] Specific operation: The server uses natural language processing, image recognition, and video analysis technologies to evaluate the similarity between the original advertising content and the generated advertising content. This evaluation calculates a similarity score.

[0980] Input: Original and generated ad content

[0981] Output: Similarity score

[0982] Step 6:

[0983] The server determines harmful ads

[0984] How it works: The server compares the similarity score with a pre-defined threshold, and if the score exceeds the threshold, it marks the ad as harmful. This decision is made automatically.

[0985] Input: Similarity score

[0986] Output: Harmful ad detection result

[0987] Step 7:

[0988] The server blocks harmful ads and notifies the user.

[0989] Specific operation: The server blocks ads that are determined to be harmful and notifies the user based on that information. Notifications are sent via email or other means.

[0990] Input: Harmful ad verdict

[0991] Output: Blocked ads and notification to the user

[0992] (Application example 1)

[0993] 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."

[0994] Advertisements provided on the Internet often contain harmful content. This leads to the distribution of unreliable advertisements and advertisements that may have a negative impact on users. In particular, the display of such harmful advertisements on smartphone and PC applications has a significant impact on the user experience. The present invention aims to provide a system that efficiently and automatically detects and blocks such harmful advertisements.

[0995] 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.

[0996] In this invention, the server includes: means for receiving advertising content input by a user; means for analyzing the received advertising content and generating a generation prompt; means for generating new content using the generation prompt; means for comparing the original advertising content with the generated content and evaluating the similarity; means for determining that the generated content is a harmful advertisement if the similarity exceeds a certain threshold; means for blocking the harmful advertisement based on the evaluation result and notifying the user; means for evaluating the similarity between the generated content and the original advertising content and determining that the advertisement is harmful based on the evaluation result; and means for notifying the user that the advertisement is blocked if it is determined to be harmful. This makes it possible to automatically determine whether advertising content created by an advertiser is harmful and to quickly block content that is determined to be harmful.

[0997] "User" means the person or entity that creates and inputs advertising content into the system.

[0998] "Advertising Content" means advertising information provided in text, image, or video format.

[0999] The "means for receiving" refers to a method or device for capturing the advertising content entered by the user into the server.

[1000] "Means for analyzing" refers to a method or device for analyzing received advertising content to create generated prompts.

[1001] "Generative prompts" are input data that a generative AI model uses to create new content based on the analysis results.

[1002] A "generating means" is a method or device that uses a generating prompt to create new content.

[1003] The "means for comparing and assessing similarity" refers to a method or device for calculating the degree of similarity in content or characteristics between the original advertising content and the generated content.

[1004] The "certain threshold" is a standard value beyond which the similarity is judged to be harmful advertising.

[1005] "Means for determining" refers to a method or device that determines whether advertising content is harmful based on the similarity assessment.

[1006] "Blocking means" refers to a method or device that stops the display or distribution of advertising content that has been determined to be harmful advertising.

[1007] "Means for notifying" refers to a method or device for informing the user of the advertisements that have been blocked and the reasons for their blocking.

[1008] The present invention is a system for receiving advertising content entered by a user and automatically detecting and blocking harmful advertisements. The system is composed of a user terminal, a server, and a program that performs a series of processes between the user and the user.

[1009] System configuration

[1010] 1. User Device

[1011] The user inputs advertising content using the ad management system interface. The input advertising content can be in the form of text, images, or videos. The user device then sends the advertising content and related metadata (such as the advertiser's ID, category, and target audience) to the server. The transmission uses an HTTPS request to ensure data security.

[1012] 2. Server

[1013] The server analyzes the advertising content received from the user's device. For the analysis, it uses natural language processing (NLP), image recognition, and video analysis technologies, and generates a generated prompt using a generative AI model. The generated prompt is created based on the main keywords and phrases in the analyzed advertising content. For example, "weight loss effect in a short time," "special price," and "innovative supplement" are extracted as generated prompts.

[1014] Next, the server generates new advertising content using a generative AI model based on the generated prompts. The generated advertising content has similar characteristics to the original advertising content. The server compares the original advertising content with the generated advertising content and evaluates their similarity. This evaluation is performed using natural language processing technology, image recognition technology, and video analysis technology. A similarity score is calculated as the evaluation result.

[1015] 3. Determining and notifying harmful advertising

[1016] The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold (e.g., 0.8), the ad is recognized as harmful. This determination is made automatically. If the ad is determined to be harmful, the server blocks the ad and notifies the user. The notification is sent by email, in-app message, or other means.

[1017] Specific examples

[1018] For example, a user enters the following ad content:

[1019] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1020] For this ad content, the following prompt text is generated:

[1021] "Fast weight loss," "Special price," "Innovative supplement"

[1022] The server generates new advertising content based on these prompts and compares it with the original advertising content to evaluate its similarity. If the similarity score exceeds a threshold, the server determines the advertising content to be harmful and notifies the user that it is blocked.

[1023] Hardware and software used

[1024] User device: Inputs and transmits advertising content. Examples include smartphones and PCs.

[1025] Server: Performs reception, analysis, generation, judgment, and notification processes. Uses a cloud server (e.g., Amazon Web Services, Google Cloud Platform, etc.).

[1026] Natural language processing technologies: such as the Hugging Face transformers library and the BERT model.

[1027] Image recognition technology: Uses OpenCV and TensorFlow.

[1028] Video analysis technology: Uses FFmpeg and OpenCV.

[1029] Generative AI models: Large-scale language models such as GPT-3.

[1030] In this way, the present invention can automatically analyze advertising content, effectively block harmful advertisements, and provide appropriate feedback to users.

[1031] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1032] Step 1:

[1033] The user inputs advertising content using the interface of the advertising management system. The input advertising content can be in the form of text, images, or videos. The user terminal then sends the input advertising content and related metadata (such as advertiser ID, category, target audience, etc.) to the server using an HTTPS request.

[1034] Input: Ad content, associated metadata

[1035] Output: Ad content and associated metadata included in the HTTPS request

[1036] Step 2:

[1037] The server receives the advertising content and associated metadata sent from the user device. The server then analyzes the advertising content using natural language processing (NLP), image recognition, or video analysis techniques to generate a generated prompt that includes key keywords or phrases from the analyzed advertising content (e.g., "fast weight loss," "special price," "innovative supplement").

[1038] Input: Ad content, associated metadata

[1039] Output: Generated prompt

[1040] Step 3:

[1041] The server generates new advertising content using a generative AI model based on the prompts. The generative AI model is a large-scale language model, such as GPT-3, that receives the prompts as input and generates new advertising content based on them. The generated advertising content has similar characteristics to the original advertising content.

[1042] Input: Generate prompt

[1043] Output: Generated ad content

[1044] Step 4:

[1045] The server compares the original advertising content with the generated advertising content and evaluates the similarity. This similarity evaluation uses natural language processing technology, image recognition technology, or video analysis technology. For example, the server compares the text portions of the original advertising content with the generated advertising content and calculates a similarity score.

[1046] Input: Original ad content, Generated ad content

[1047] Output: Similarity score

[1048] Step 5:

[1049] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold (e.g., 0.8), the advertisement is recognized as harmful. This determination is made automatically.

[1050] Input: Similarity score

[1051] Output: Harmful ad detection result

[1052] Step 6:

[1053] The server blocks advertising content that is determined to be harmful within the system, thereby preventing users from viewing ads that are recognized as harmful. The server also notifies the user that the ad has been determined to be harmful and blocked. This notification is made by means of email, in-app message, etc.

[1054] Input: Harmful advertisement judgment result

[1055] Output: Block execution, notification message

[1056] This series of processes ensures the safety of advertising content and improves the user experience.

[1057] 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.

[1058] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. Furthermore, the present invention incorporates an emotion engine that recognizes the user's emotions, thereby evaluating the emotional impact of advertisements and improving the accuracy of determining harmful advertisements. The specific program processing flow is explained below.

[1059] System Configuration

[1060] 1. Users submit advertising content

[1061] The user uses the ad management system interface to input ad content, which can be in the form of text, images, or videos, and then clicks the "Submit" button.

[1062] 2. Submitting Content

[1063] The device sends the advertising content entered by the user and its associated metadata (e.g., advertiser ID, category, target audience) to the server via an HTTPS request.

[1064] 3. Content analysis and generation of prompts

[1065] The server analyzes the received advertising content and generates prompts using generative AI, which uses natural language processing, image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features.

[1066] 4. User Emotion Analysis

[1067] The server uses an emotion engine to analyze the user's emotions toward the received advertising content. Sentiment analysis extracts emotions, such as positive, negative, or neutral, from text, audio, or image data.

[1068] 5. Generating Similar Content

[1069] Based on the generated prompts, the server uses generative AI to generate new ad content, the format of which matches the original ad content (text, image, or video).

[1070] 6. Content Comparison and Similarity Evaluation

[1071] The server compares the original advertising content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology.

[1072] 7. Identifying Harmful Advertisements Using Sentiment Analysis Results

[1073] The server combines the similarity score and the results of user sentiment analysis to determine whether an advertisement is harmful. If the similarity score and the results of sentiment analysis exceed a certain threshold, the advertisement is determined to be harmful.

[1074] 8. Block harmful ads and notify users

[1075] The server blocks the ads that are determined to be harmful, updates the status of the ads to "blocked," and saves it in the database. At the same time, it automatically sends a notification email to the user informing them that the ads have been determined to be harmful and blocked.

[1076] Specific examples

[1077] For example, suppose a user enters an advertisement for a "new diet supplement" as follows:

[1078] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1079] The specific processing flow when this advertisement is submitted by a user is as follows.

[1080] 1. User enters advertising content:

[1081] The user enters the above text into the form of the ad management system and clicks the "Submit" button.

[1082] 2. The device sends the advertising content:

[1083] The user's device sends the advertising content and associated metadata to the server via an HTTPS request.

[1084] 3. The server analyzes the ad content:

[1085] A server receives the advertising content and extracts specific keywords or phrases to generate generated prompts.

[1086] 4. The server analyzes the user's emotions:

[1087] The server uses an emotion engine to analyze the emotional impact of advertising content on the user.

[1088] 5. Generate content based on a generation prompt:

[1089] The server uses generative AI to generate similar advertising content.

[1090] 6. The server compares the original ad with the generated ad:

[1091] The server evaluates the similarity between the generated advertising content and the original advertisement and calculates a similarity score.

[1092] 7. The server uses the sentiment analysis results to determine harmful ads:

[1093] The server combines the similarity score with the sentiment analysis results to determine whether the ad is harmful.

[1094] 8. The server blocks harmful ads and notifies the user:

[1095] The server blocks ads that are determined to be harmful and notifies the user that "the ad has been determined to be harmful and has been blocked."

[1096] By taking into account the results of user sentiment analysis, the system of the present invention can detect harmful advertisements with higher accuracy than conventional methods that only use similarity evaluation. In addition, appropriate feedback is provided to users, ensuring transparency.

[1097] The processing flow will be explained below.

[1098] Step 1:

[1099] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[1100] Step 2:

[1101] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[1102] Step 3:

[1103] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[1104] Step 4:

[1105] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[1106] Step 5:

[1107] The server analyzes the user's emotions. The server uses an emotion engine to classify the user's emotions toward the advertising content as positive, negative, or neutral. This can be done using text, voice, or image analysis.

[1108] Step 6:

[1109] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[1110] Step 7:

[1111] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[1112] Step 8:

[1113] The server integrates the similarity score and the user's sentiment analysis results, and performs a comprehensive evaluation by combining the similarity score and the sentiment analysis results.

[1114] Step 9:

[1115] The server judges harmful advertisements. If the overall evaluation based on the similarity and sentiment analysis results exceeds a certain threshold, the advertisement is judged as harmful. Conversely, if it is below the threshold, the advertisement is not judged as harmful.

[1116] Step 10:

[1117] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[1118] Step 11:

[1119] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block, the similarity score, and the results of sentiment analysis.

[1120] These steps enable the system of the present invention to not only efficiently detect and block harmful advertising content, but also provide feedback to help users understand the process.

[1121] Example 2

[1122] 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."

[1123] Conventional advertising content filtering systems have limited accuracy in determining whether an ad's content is harmful, leading to false positives. Furthermore, because they do not consider users' emotional responses, they can display emotionally unpleasant ads. This can lead to a poor user experience and a loss of trust in advertisers.

[1124] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1125] In this invention, the server includes a means for analyzing a user's emotions, a means for analyzing received content and generating a generation prompt, and a means for generating new content using the generation prompt, thereby enabling highly accurate determination of harmful content that takes user emotions into consideration.

[1126] "User" means a person who inputs and transmits advertising content using the system.

[1127] "Advertising Content" means a collection of information entered by a user that is analyzed and evaluated by the system, and may take the form of text, images, or video.

[1128] The "means for receiving" is a part of the system that has the functionality for capturing advertising content sent by a user.

[1129] The "means for analyzing" is a part of the system that has the function of analyzing received advertising content and extracting useful information.

[1130] A "generative prompt" is input information given to a generative AI model based on the analysis results of advertising content.

[1131] A "means for generating" is a part of the system that has the ability to create new content using generation prompts.

[1132] A "comparison means" is a part of the system that has the ability to evaluate the original advertising content against the generated content.

[1133] The "means for assessing similarity" is a part of the system that has the function of quantifying the similarity between the original advertising content and the generated content.

[1134] The "means for determining that content is harmful" is a part of the system that has the function of determining whether advertising content is harmful based on the results of similarity evaluation and sentiment analysis.

[1135] "Blocking means" is a part of the system that has the function of preventing advertising content that is determined to be harmful from being displayed.

[1136] The "means for notifying the user" is a part of the system that has the function of notifying the user that advertising content has been determined to be harmful.

[1137] The "means for analyzing emotions" is a part of the system that has the function of evaluating the emotional impact that advertising content has on the user and extracting emotions such as positive, negative, or neutral.

[1138] "Natural language processing technology" is a technology for analyzing the meaning and content of text data.

[1139] "Image recognition technology" is a technology for analyzing the contents of image data.

[1140] "Video analysis technology" is a technology for analyzing the content of video data.

[1141] The present invention is a system that receives advertising content entered by a user and automatically determines whether that content is harmful and blocks it. Furthermore, by analyzing user sentiment, the accuracy of determining harmful advertisements is improved. Below, specific embodiments of the present invention are described.

[1142] System Configuration

[1143] User Roles

[1144] The user inputs the advertising content through the advertising management system. The advertising content can be provided in the form of text, images, or videos. Once the user inputs the advertising content and clicks the "Submit" button, the content is received by the system.

[1145] Device Role

[1146] When a user submits advertising content, the device sends the entered advertising content and related metadata (e.g., advertiser ID, category, target audience) to the server using an HTTPS request. The device uses an existing web browser or a dedicated application.

[1147] Server Roles

[1148] The server receives the advertising content sent by the user and performs the following processes.

[1149] 1. Advertising content analysis:

[1150] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies. Specifically, it uses open source NLP libraries (e.g., spaCy, NLTK), image recognition libraries (e.g., OpenCV, TensorFlow), and video analysis technologies (e.g., FFmpeg).

[1151] Specific keywords, phrases, and visual features are extracted from the analyzed data, and generative prompt sentences are generated based on these.

[1152] 2. User sentiment analysis:

[1153] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the emotional impact of the ad content on the user, categorizing emotions into positive, negative, and neutral categories.

[1154] 3. Generate new content based on a generation prompt:

[1155] The server uses a generative AI model (e.g., GPT-4, DALL-E) to generate new ad content based on the generated prompt. The generated content matches the original format (text, image, video).

[1156] 4. Comparison of original ad content and generated content:

[1157] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score.

[1158] 5. Determining harmful advertising:

[1159] The server determines whether an advertisement is harmful based on the similarity score and the result of sentiment analysis. If the similarity score or sentiment score exceeds a certain threshold, the advertisement is determined to be harmful.

[1160] 6. Block harmful ads and notify users:

[1161] The server blocks the ad that is determined to be harmful, updates the ad's status to "blocked," and notifies the user by email that the ad has been determined to be harmful and blocked.

[1162] Specific examples

[1163] For example, if a user enters an ad for a "new diet supplement" like this:

[1164] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1165] Once this ad content is sent, the server performs the following steps:

[1166] 1. User enters and submits ad:

[1167] The user enters text into the ad management system and clicks the "Submit" button.

[1168] 2. The device sends the advertising content:

[1169] The device sends advertising content to the server via an HTTPS request.

[1170] 3. The server analyzes the advertising content:

[1171] Keywords such as "innovative," "weight loss effect," and "special price" are extracted, and prompt sentences are generated to be given to the generative AI model.

[1172] 4. The server analyzes the user's emotions:

[1173] Using an emotion engine, the emotions that ad text evokes in users are analyzed, and an emotion score is obtained, such as 80% positive, 15% negative, and 5% neutral.

[1174] 5. Generate new content based on a generation prompt:

[1175] A prompt sentence is input into the generative AI model to generate new advertising text, such as "Why not lose weight in a healthy way with this supplement?"

[1176] 6. Server compares original ad with new ad:

[1177] The original ad and the new ad are compared using natural language processing and a similarity score (e.g., 85 points) is calculated.

[1178] 7. The server determines whether the ad is harmful:

[1179] Advertisements are analyzed based on similarity scores and sentiment scores to determine whether they are harmful.

[1180] 8. Blocking harmful ads and notifying users:

[1181] Update the ad status to "Blocked" and notify the user that "The ad has been identified as harmful and blocked."

[1182] The system of the present invention enables automatic detection and blocking of harmful advertisements with high accuracy, taking into account the user's emotional response, thereby improving the user experience and ensuring advertising safety.

[1183] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1184] Step 1:

[1185] User enters and submits advertising content

[1186] The user uses the interface of the ad management system to input the ad content (in the form of text, images, or videos). Then, he clicks the "Submit" button to send the input content. The input data is the ad content provided by the user, and the output data is the ad content sent to the terminal. Specifically, the user attaches text, images, or videos to a form and submits it.

[1187] Step 2:

[1188] The device sends advertising content to the server

[1189] The device sends the advertising content and related metadata entered by the user to the server using an HTTPS request. The input data is the content and metadata entered by the user, and the output data is the content data sent to the server. Specifically, the device converts the user's input data into packets and sends them to the server as an encrypted HTTPS request.

[1190] Step 3:

[1191] The server analyzes the ad content and generates a prompt

[1192] The server analyzes the received advertising content. It uses natural language processing techniques (e.g., spaCy, NLTK), image recognition techniques (e.g., OpenCV, TensorFlow), and video analysis techniques (e.g., FFmpeg) to extract specific keywords, phrases, and visual features from the content. The input data is the advertising content sent to the server, and the output data is the generated prompt. Specifically, the server runs the analysis engine to extract keywords such as "weight loss" and "special price" from the original content and compose the generated prompt.

[1193] Step 4:

[1194] The server analyzes the user's emotions

[1195] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the user's emotions toward the ad content. The input data here is the analyzed ad content, and the output data is an emotion score such as positive, negative, or neutral. Specifically, the server passes the content data to the emotion analysis module and obtains the emotion score.

[1196] Step 5:

[1197] Generate new content based on a prompt

[1198] The server uses the generation prompt sentence to generate new advertising content for a generative AI model (e.g., GPT-4, DALL-E). The input data is the generation prompt sentence, and the output data is the generated new content. Specifically, the server inputs the generation prompt into the AI ​​model, and the model generates new advertising content (e.g., text and images).

[1199] Step 6:

[1200] The server compares the original ad content with the new content

[1201] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score. The input data is the original advertising content and the generated new content, and the output data is the similarity score. Specifically, the server inputs both contents into an analysis engine and quantifies the similarity.

[1202] Step 7:

[1203] The server determines harmful ads

[1204] The server determines whether an ad is harmful based on the similarity score and the results of sentiment analysis. The input data are the similarity score and sentiment score, and the output data is the harmfulness determination result. Specifically, the server compares the obtained scores and determines that an ad is harmful if it exceeds a threshold.

[1205] Step 8:

[1206] The server blocks harmful ads and notifies the user

[1207] The server blocks ads that are determined to be harmful and updates the ad status to "blocked." It also automatically sends a notification email to the user informing them that the ad has been determined to be harmful and blocked. The input data is the harmful determination result, and the output data is the updated ad status and notification email. Specifically, the server updates the ad database and sends a notification email to the user.

[1208] (Application example 2)

[1209] 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."

[1210] With the spread of internet advertising, the risk of harmful advertising content adversely affecting users is increasing. Therefore, a system that automatically monitors advertisements and quickly detects and blocks harmful advertisements is needed. Furthermore, conventional systems that only use similarity evaluation are sometimes insufficient in accuracy, so there is a need for a highly accurate harmful advertisement detection system that takes into account the emotional impact of users.

[1211] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content and generating a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content and evaluating the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user. This makes it possible to detect harmful advertisements quickly and accurately, and maintain the integrity of advertisements while minimizing the adverse impact on users.

[1212] "User-entered advertising content" refers to data in the form of text, images, or videos of advertisements provided by users such as advertisers or advertising agencies through online platforms.

[1213] The "receiving means" refers to a communication interface and software that has the function of receiving the advertising content input by the user on the server side.

[1214] "Means for analyzing and generating generative prompts" refers to the function of analyzing received advertising content and creating instructions (prompts) for generating new content using generative AI based on that content.

[1215] "Means for generating new content using generative prompts" refers to the ability to automatically generate new advertising content based on generative prompts using a generative AI model.

[1216] "Means for comparing generated content and assessing similarity" refers to algorithms and technologies that match original advertising content with generated advertising content and measure their similarity.

[1217] "Means for recognizing user emotions and assessing the emotional impact of advertising" refers to a sentiment analysis engine and associated algorithms for analyzing the emotional response of advertising content to users and assessing its impact.

[1218] The term "certain threshold" refers to a predetermined reference value for determining whether an advertisement is harmful or not, based on the results of similarity assessment and sentiment analysis.

[1219] "Means for determining harmful advertising" refers to an algorithm that automatically determines that advertising content is harmful if the results of similarity assessment and sentiment analysis exceed a certain threshold.

[1220] "Means for blocking and notifying users" refers to the function of excluding advertising content that is determined to be harmful from display and distribution networks, and automatically notifying advertisers of this fact.

[1221] This invention relates to an automated system for accurately determining the harmfulness of advertising content entered by a user and blocking it. The system of the present invention mainly includes the following main means: means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user.

[1222] Program processing explanation

[1223] 1. Receiving advertising content

[1224] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. After entering the content, the user clicks the "Submit" button, which sends the content and its metadata (advertiser ID, category, target audience) to the server. This communication uses an HTTPS request.

[1225] 2. Analyzing advertising content and generating prompts

[1226] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features. Based on the analysis results, a prompt is generated, and new advertising content is automatically generated based on the advertising content.

[1227] 3. Emotion analysis

[1228] The server evaluates the emotional impact of advertising content using a sentiment analysis engine that uses existing sentiment analysis software, such as NLTK's SentimentIntensityAnalyzer, to extract positive, negative, or neutral sentiment scores from text, audio, or image data.

[1229] 4. Similarity evaluation

[1230] The server compares the original ad content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology. For example, it can use Hugging Face's transformer library to calculate the similarity between the similar content generated by the GPT-3 model and the original content.

[1231] 5. Identifying and blocking harmful ads

[1232] The server combines the results of the similarity assessment and sentiment analysis to determine whether an ad is harmful. Ads that are determined to be harmful are automatically blocked and users are notified of this. Notifications are sent via email or system alerts.

[1233] Specific examples

[1234] For example, if a user enters the following ad content:

[1235] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1236] When this ad is submitted, the system works as follows: First, the ad content is received, the text is analyzed, and a generative prompt is generated. Using the generative prompt, a generative AI model (GPT-3) generates similar content, such as:

[1237] "Just take this new supplement and lose weight fast! Try it at a great price!"

[1238] The sentiment analysis engine then evaluates the emotional impact and calculates the similarity between the original content and the generated content. Based on this result, it determines whether the ad is harmful, and if so, the ad is blocked and the user is notified. This allows for early detection of harmful content and prevents adverse effects on users.

[1239] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1240] Step 1:

[1241] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. The user clicks the "Submit" button, which sends the entered content and its metadata (advertiser ID, category, target audience) to the server. The entered data is securely transmitted to the server via an HTTPS request.

[1242] Step 2:

[1243] The server analyzes the received advertising content. First, it performs text analysis and uses natural language processing (NLP) techniques to extract specific keywords and phrases within the content. It also uses image and video analysis techniques to extract visual features. These analyses result in a generated prompt. The input is the advertising content and related metadata, and the output is the analysis results and the generated prompt.

[1244] Step 3:

[1245] The server uses the generative prompts to generate new advertising content. A generative AI model (e.g., GPT-3) is used to create similar advertising content based on the generative prompts. The generated content is in the same format (text, image, or video) as the original. The input is the generative prompts, and the output is the generated advertising content.

[1246] Step 4:

[1247] The server uses a sentiment analysis engine to evaluate the emotional impact of ad content on users. Sentiment analysis software such as NLTK's SentimentIntensityAnalyzer extracts positive, negative, or neutral sentiment scores from text, audio, or image data. The input is the ad content, and the output is the sentiment analysis results.

[1248] Step 5:

[1249] The server compares the generated ad content with the original ad content and calculates a similarity score. It uses natural language processing, image recognition, or video analysis technology to evaluate the similarity based on the features of both pieces of content. For example, it uses Hugging Face's transformers library to compare it with the output of the GPT-3 model. The input is the original ad content and the generated content, and the output is a similarity score.

[1250] Step 6:

[1251] The server determines whether an advertisement is harmful or not based on the similarity score and the results of sentiment analysis. The criteria (certain threshold) for determining whether an advertisement is harmful is set in advance, and if the similarity score and the results of sentiment analysis exceed this threshold, the advertisement is determined to be harmful. The input is the similarity score and the results of sentiment analysis, and the output is the result of determining whether the advertisement is harmful.

[1252] Step 7:

[1253] The server blocks advertising content that is determined to be harmful. It updates the status of the advertisement to "blocked" and saves it in the database. It also automatically sends a notification to the user that the advertisement has been determined to be harmful and blocked. The input is the result of determining whether the advertisement is harmful, and the output is the notification to the user and the blocking status of the advertising content.

[1254] Through this series of processes, the system can accurately determine whether an advertisement is harmful and prevent the distribution of harmful advertising content.

[1255] 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.

[1256] 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.

[1257] 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.

[1258] [Fourth embodiment]

[1259] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1260] 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.

[1261] 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).

[1262] 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.

[1263] 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.

[1264] 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).

[1265] 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.

[1266] 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.

[1267] 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.

[1268] 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.

[1269] 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.

[1270] 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.

[1271] 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."

[1272] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. The program of the present invention is made up of a series of processes that are carried out between a server, a terminal, and a user.

[1273] System Configuration

[1274] 1. Users submit advertising content

[1275] A user uses the ad management system interface to input ad content, which may be in the form of text, images, or video.

[1276] 2. Submitting Content

[1277] The device sends the user-entered advertising content to the server via an HTTPS request, including the advertising content and associated metadata (e.g., advertiser ID, category, target audience, etc.).

[1278] 3. Content analysis and generation of prompts

[1279] The server analyzes the received advertising content and generates a prompt using generative AI. Natural language processing (NLP), image recognition, or video analysis techniques are used for the analysis. For example, specific keywords or phrases in the advertising content, parameters such as "weight loss in a short time" or "special price" are extracted.

[1280] 4. Generating Similar Content

[1281] The server uses generative AI to generate new advertising content based on the generative prompts, with the generated content having similar characteristics to the original advertising content.

[1282] 5. Content Comparison and Similarity Evaluation

[1283] The server compares the original advertising content with the generated similar content. This comparison evaluates the similarity using natural language processing, image recognition, or video analysis technology. A similarity score is calculated as the evaluation result.

[1284] 6. Determining harmful advertising

[1285] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold, the advertisement is recognized as harmful. This determination is made automatically.

[1286] 7. Block harmful ads and notify users

[1287] The server blocks ads that are deemed harmful and notifies the user based on that information. The blocked ads are then stopped from being displayed, and the user is notified by email or other means.

[1288] Specific examples

[1289] For example, if a user types in an ad for a "new diet supplement":

[1290] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1291] When a user submits advertising content, the following steps are performed:

[1292] 1. User enters advertising content:

[1293] The user enters the above text into the advertisement management system and clicks the send button.

[1294] 2. The device sends the advertising content:

[1295] The user's terminal transmits the advertising content and associated metadata to the server.

[1296] 3. The server analyzes the ad content:

[1297] The server receives the advertising content and uses natural language processing techniques to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[1298] 4. Generate content based on a generation prompt:

[1299] The server uses generation AI to generate advertising content with similar content.

[1300] 5. The server compares the original ad with the generated ad:

[1301] The server evaluates the similarity between the original advertising content and the generated advertising content.

[1302] 6. The server evaluates the similarity score:

[1303] If the similarity score is high, the ad is determined to be harmful.

[1304] 7. The server blocks harmful ads and notifies the user:

[1305] Advertisements that are determined to be harmful are blocked, and the user is notified that "the advertisement has been determined to be harmful and has been blocked."

[1306] Through this process, the system can efficiently detect and block malicious and harmful ads, while providing appropriate feedback to users, ensuring transparency.

[1307] The processing flow will be explained below.

[1308] Step 1:

[1309] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[1310] Step 2:

[1311] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[1312] Step 3:

[1313] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[1314] Step 4:

[1315] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[1316] Step 5:

[1317] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[1318] Step 6:

[1319] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[1320] Step 7:

[1321] The server determines whether an advertisement is harmful based on the similarity score. If the similarity score exceeds a preset threshold, the server determines the advertisement as harmful. Conversely, if the similarity score is below the threshold, the advertisement is not determined to be harmful.

[1322] Step 8:

[1323] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[1324] Step 9:

[1325] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block and the similarity score.

[1326] These steps enable the system of the present invention to efficiently detect, evaluate, and block harmful advertising content, and provide appropriate feedback to the user.

[1327] Example 1

[1328] 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."

[1329] Conventional ad management systems lacked the ability to automatically detect and block harmful ad content entered by users. This resulted in a high risk of harmful ads being displayed and made it difficult to maintain ad quality. Feedback to users was also insufficient, lacking transparency. To solve these problems, a system that could automatically detect and block harmful ads with high accuracy and efficiency was needed.

[1330] 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.

[1331] In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate similarity, means for determining that the advertising content is harmful if the similarity exceeds a certain threshold, means for blocking the harmful advertising based on the determination result and notifying the user, means for transmitting information including metadata related to the advertising content, means for extracting specific keywords or phrases based on the received advertising content, and means for calculating the similarity of the generated advertising content. This effectively solves the problems of detecting and blocking harmful advertising that conventional systems have had, improves the quality of advertising, and enables transparent feedback to users.

[1332] "Advertising Content" means information that a user inputs into the Ad Management System, and may be provided in the form of text, images, or video.

[1333] A "generative prompt" is input data that allows a generative AI model to generate new advertising content based on information extracted by analyzing received advertising content.

[1334] A "generative AI model" is a machine learning model for automatically generating new advertising content based on generative prompts.

[1335] The "similarity" refers to the degree of similarity between the original advertising content and the generated advertising content, and is expressed as a numerical value.

[1336] The "threshold" refers to a boundary value that is the standard for determining whether the similarity is a harmful advertisement.

[1337] A "harmful ad" is an ad whose content has a similarity score that exceeds a certain threshold and has been automatically determined not to be displayed.

[1338] "Metadata" is additional information related to advertising content, including advertiser ID, category, target audience, etc.

[1339] "Natural language processing technology" is a technology for analyzing text-based content and extracting specific keywords and phrases.

[1340] "Image recognition technology" is a technology for analyzing image-based content and extracting specific features and patterns.

[1341] "Video analysis technology" is a technology for analyzing video-based content and extracting specific frames or scenes.

[1342] An "HTTPS request" is a secure communication protocol used when exchanging data over the Internet.

[1343] "Means for receiving" refers to hardware or software for receiving advertising content from a user.

[1344] "Means for analyzing" refers to hardware or software for examining received advertising content and extracting necessary information.

[1345] The "means for evaluating" refers to hardware or software for measuring the similarity between the original advertising content and the generated content and evaluating it as a numerical value.

[1346] "Blocking measures" refers to hardware or software that stops the display of advertising content that is determined to be harmful.

[1347] The "notification means" refers to hardware or software for notifying the user of the determination result.

[1348] The present invention is a system that automatically analyzes advertising content entered by users and detects and blocks harmful advertisements as necessary. The system configuration includes a series of processes performed between a server, a terminal, and a user. The following describes the specific program processing for implementing the present invention and its detailed explanation.

[1349] Program processing explanation

[1350] The system has the following main features:

[1351] 1. Receiving advertising content: The user enters advertising content using the advertising management system interface. The entered advertising content can be in the form of text, images, or videos, and is sent by the device to the server using an HTTPS request.

[1352] 2. Analyzing advertising content and generating a generative prompt: The server analyzes the received advertising content. This analysis uses natural language processing (NLP), image recognition, or video analysis technology. For example, specific keywords or phrases in the advertising content, or parameters such as "fast weight loss" or "special price offer," are extracted. A generative prompt is generated based on the extracted information.

[1353] 3. Generating similar content based on the prompt: The server generates new advertising content using a generative AI model (e.g., GPT-4) based on the prompt. The generated content has similar characteristics to the original advertising content.

[1354] 4. Comparison of advertising content and similarity evaluation: The server compares the original advertising content with the newly generated advertising content and evaluates their similarity. This evaluation uses natural language processing technology, image recognition technology, or video analysis technology, and calculates a similarity score as the evaluation result.

[1355] 5. Identifying and blocking harmful ads: The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold, the ad is recognized as harmful. This determination is made automatically. Based on the determination result, harmful ads are blocked and the user is notified. This notification is made by means such as email.

[1356] Specific examples

[1357] For example, consider the following ad content for a "new diet supplement" entered by a user:

[1358] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1359] In this case, the system does the following:

[1360] Receiving advertising content: The user enters the above text into the advertising management system and clicks the submit button.

[1361] Content submission: The user's device submits the ad content and metadata to the server using an HTTP POST request.

[1362] Content analysis: The server uses text analysis tools to generate prompts such as "fast weight loss," "special prices," and "innovative supplements."

[1363] Similar content generation: The server inputs the generation prompt into the generative AI model to generate new advertising content with similar content.

[1364] Similarity evaluation: The server evaluates the similarity between the original advertising content and the newly generated advertising content and calculates a similarity score.

[1365] Identifying and blocking harmful ads: If the server determines that an ad is harmful based on the similarity score, it blocks the ad and notifies the user that "the ad has been identified as harmful and blocked."

[1366] Through these processes, the system can efficiently detect and block harmful ads, thereby improving the quality of ads and providing transparent feedback to users.

[1367] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1368] Step 1:

[1369] User enters advertising content

[1370] Specific operation: The user accesses the interface of the ad management system and inputs the ad content. The ad content can be input in the form of text, image, or video. The user clicks the submit button to submit the content.

[1371] Input: Ad content in the form of text, images, or videos

[1372] Output: Clicking the submit button sends the ad content to the device.

[1373] Step 2:

[1374] The device sends advertising content

[1375] How it works: The user's device sends the entered advertising content to the server using an HTTPS request, along with metadata such as the advertiser's ID, category, and target audience.

[1376] Input: User-entered ad content and metadata

[1377] Output: Ad content and metadata sent to the server as an HTTPS request

[1378] Step 3:

[1379] The server analyzes the ad content

[1380] How it works: The server uses natural language processing (NLP), image recognition, and video analysis techniques to analyze the received ad content. For example, in the case of a text ad, keywords and phrases such as "fast weight loss" or "special offer" are extracted.

[1381] Input: Ad content and metadata received as an HTTPS request

[1382] Output: Generated prompts containing extracted keywords and phrases

[1383] Step 4:

[1384] The server generates similar content based on the prompt.

[1385] How it works: The server generates new ad content using a generative AI model (e.g., GPT-4) based on the prompt. The generated ad content has similar characteristics to the original ad content.

[1386] Input: Generate prompt

[1387] Output: New ad content generated by the generative AI model

[1388] Step 5:

[1389] The server compares the original ad content with the generated ad content

[1390] Specific operation: The server uses natural language processing, image recognition, and video analysis technologies to evaluate the similarity between the original advertising content and the generated advertising content. This evaluation calculates a similarity score.

[1391] Input: Original and generated ad content

[1392] Output: Similarity score

[1393] Step 6:

[1394] The server determines harmful ads

[1395] How it works: The server compares the similarity score with a pre-defined threshold, and if the score exceeds the threshold, it marks the ad as harmful. This decision is made automatically.

[1396] Input: Similarity score

[1397] Output: Harmful ad detection result

[1398] Step 7:

[1399] The server blocks harmful ads and notifies the user.

[1400] Specific operation: The server blocks ads that are determined to be harmful and notifies the user based on that information. Notifications are sent via email or other means.

[1401] Input: Harmful ad verdict

[1402] Output: Blocked ads and notification to the user

[1403] (Application example 1)

[1404] 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."

[1405] Advertisements provided on the Internet often contain harmful content. This leads to the distribution of unreliable advertisements and advertisements that may have a negative impact on users. In particular, the display of such harmful advertisements on smartphone and PC applications has a significant impact on the user experience. The present invention aims to provide a system that efficiently and automatically detects and blocks such harmful advertisements.

[1406] 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.

[1407] In this invention, the server includes: means for receiving advertising content input by a user; means for analyzing the received advertising content and generating a generation prompt; means for generating new content using the generation prompt; means for comparing the original advertising content with the generated content and evaluating the similarity; means for determining that the generated content is a harmful advertisement if the similarity exceeds a certain threshold; means for blocking the harmful advertisement based on the evaluation result and notifying the user; means for evaluating the similarity between the generated content and the original advertising content and determining that the advertisement is harmful based on the evaluation result; and means for notifying the user that the advertisement is blocked if it is determined to be harmful. This makes it possible to automatically determine whether advertising content created by an advertiser is harmful and to quickly block content that is determined to be harmful.

[1408] "User" means the person or entity that creates and inputs advertising content into the system.

[1409] "Advertising Content" means advertising information provided in text, image, or video format.

[1410] The "means for receiving" refers to a method or device for capturing the advertising content entered by the user into the server.

[1411] "Means for analyzing" refers to a method or device for analyzing received advertising content to create generated prompts.

[1412] "Generative prompts" are input data that a generative AI model uses to create new content based on the analysis results.

[1413] A "generating means" is a method or device that uses a generating prompt to create new content.

[1414] The "means for comparing and assessing similarity" refers to a method or device for calculating the degree of similarity in content or characteristics between the original advertising content and the generated content.

[1415] The "certain threshold" is a standard value beyond which the similarity is judged to be harmful advertising.

[1416] "Means for determining" refers to a method or device that determines whether advertising content is harmful based on the similarity assessment.

[1417] "Blocking means" refers to a method or device that stops the display or distribution of advertising content that has been determined to be harmful advertising.

[1418] "Means for notifying" refers to a method or device for informing the user of the advertisements that have been blocked and the reasons for their blocking.

[1419] The present invention is a system for receiving advertising content entered by a user and automatically detecting and blocking harmful advertisements. The system is composed of a user terminal, a server, and a program that performs a series of processes between the user and the user.

[1420] System configuration

[1421] 1. User Device

[1422] The user inputs advertising content using the ad management system interface. The input advertising content can be in the form of text, images, or videos. The user device then sends the advertising content and related metadata (such as the advertiser's ID, category, and target audience) to the server. The transmission uses an HTTPS request to ensure data security.

[1423] 2. Server

[1424] The server analyzes the advertising content received from the user's device. For the analysis, it uses natural language processing (NLP), image recognition, and video analysis technologies, and generates a generated prompt using a generative AI model. The generated prompt is created based on the main keywords and phrases in the analyzed advertising content. For example, "weight loss effect in a short time," "special price," and "innovative supplement" are extracted as generated prompts.

[1425] Next, the server generates new advertising content using a generative AI model based on the generated prompts. The generated advertising content has similar characteristics to the original advertising content. The server compares the original advertising content with the generated advertising content and evaluates their similarity. This evaluation is performed using natural language processing technology, image recognition technology, and video analysis technology. A similarity score is calculated as the evaluation result.

[1426] 3. Determining and notifying harmful advertising

[1427] The server determines whether the ad content is harmful based on the similarity score. If the similarity exceeds a certain threshold (e.g., 0.8), the ad is recognized as harmful. This determination is made automatically. If the ad is determined to be harmful, the server blocks the ad and notifies the user. The notification is sent by email, in-app message, or other means.

[1428] Specific examples

[1429] For example, a user enters the following ad content:

[1430] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1431] For this ad content, the following prompt text is generated:

[1432] "Fast weight loss," "Special price," "Innovative supplement"

[1433] The server generates new advertising content based on these prompts and compares it with the original advertising content to evaluate its similarity. If the similarity score exceeds a threshold, the server determines the advertising content to be harmful and notifies the user that it is blocked.

[1434] Hardware and software used

[1435] User device: Inputs and transmits advertising content. Examples include smartphones and PCs.

[1436] Server: Performs reception, analysis, generation, judgment, and notification processes. Uses a cloud server (e.g., Amazon Web Services, Google Cloud Platform, etc.).

[1437] Natural language processing technologies: such as the Hugging Face transformers library and the BERT model.

[1438] Image recognition technology: Uses OpenCV and TensorFlow.

[1439] Video analysis technology: Uses FFmpeg and OpenCV.

[1440] Generative AI models: Large-scale language models such as GPT-3.

[1441] In this way, the present invention can automatically analyze advertising content, effectively block harmful advertisements, and provide appropriate feedback to users.

[1442] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1443] Step 1:

[1444] The user inputs advertising content using the interface of the advertising management system. The input advertising content can be in the form of text, images, or videos. The user terminal then sends the input advertising content and related metadata (such as advertiser ID, category, target audience, etc.) to the server using an HTTPS request.

[1445] Input: Ad content, associated metadata

[1446] Output: Ad content and associated metadata included in the HTTPS request

[1447] Step 2:

[1448] The server receives the advertising content and associated metadata sent from the user device. The server then analyzes the advertising content using natural language processing (NLP), image recognition, or video analysis techniques to generate a generated prompt that includes key keywords or phrases from the analyzed advertising content (e.g., "fast weight loss," "special price," "innovative supplement").

[1449] Input: Ad content, associated metadata

[1450] Output: Generated prompt

[1451] Step 3:

[1452] The server generates new advertising content using a generative AI model based on the prompts. The generative AI model is a large-scale language model, such as GPT-3, that receives the prompts as input and generates new advertising content based on them. The generated advertising content has similar characteristics to the original advertising content.

[1453] Input: Generate prompt

[1454] Output: Generated ad content

[1455] Step 4:

[1456] The server compares the original advertising content with the generated advertising content and evaluates the similarity. This similarity evaluation uses natural language processing technology, image recognition technology, or video analysis technology. For example, the server compares the text portions of the original advertising content with the generated advertising content and calculates a similarity score.

[1457] Input: Original ad content, Generated ad content

[1458] Output: Similarity score

[1459] Step 5:

[1460] The server determines whether the advertising content is harmful based on the similarity score. If the similarity score exceeds a certain threshold (e.g., 0.8), the advertisement is recognized as harmful. This determination is made automatically.

[1461] Input: Similarity score

[1462] Output: Harmful ad detection result

[1463] Step 6:

[1464] The server blocks advertising content that is determined to be harmful within the system, thereby preventing users from viewing ads that are recognized as harmful. The server also notifies the user that the ad has been determined to be harmful and blocked. This notification is made by means of email, in-app message, etc.

[1465] Input: Harmful advertisement judgment result

[1466] Output: Block execution, notification message

[1467] This series of processes ensures the safety of advertising content and improves the user experience.

[1468] 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.

[1469] The present invention relates to a system that receives advertising content entered by a user and automatically detects and blocks harmful advertisements. Furthermore, the present invention incorporates an emotion engine that recognizes the user's emotions, thereby evaluating the emotional impact of advertisements and improving the accuracy of determining harmful advertisements. The specific program processing flow is explained below.

[1470] System Configuration

[1471] 1. Users submit advertising content

[1472] The user uses the ad management system interface to input ad content, which can be in the form of text, images, or videos, and then clicks the "Submit" button.

[1473] 2. Submitting Content

[1474] The device sends the advertising content entered by the user and its associated metadata (e.g., advertiser ID, category, target audience) to the server via an HTTPS request.

[1475] 3. Content analysis and generation of prompts

[1476] The server analyzes the received advertising content and generates prompts using generative AI, which uses natural language processing, image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features.

[1477] 4. User Emotion Analysis

[1478] The server uses an emotion engine to analyze the user's emotions toward the received advertising content. Sentiment analysis extracts emotions, such as positive, negative, or neutral, from text, audio, or image data.

[1479] 5. Generating Similar Content

[1480] Based on the generated prompts, the server uses generative AI to generate new ad content, the format of which matches the original ad content (text, image, or video).

[1481] 6. Content Comparison and Similarity Evaluation

[1482] The server compares the original advertising content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology.

[1483] 7. Identifying Harmful Advertisements Using Sentiment Analysis Results

[1484] The server combines the similarity score and the results of user sentiment analysis to determine whether an advertisement is harmful. If the similarity score and the results of sentiment analysis exceed a certain threshold, the advertisement is determined to be harmful.

[1485] 8. Block harmful ads and notify users

[1486] The server blocks the ads that are determined to be harmful, updates the status of the ads to "blocked," and saves it in the database. At the same time, it automatically sends a notification email to the user informing them that the ads have been determined to be harmful and blocked.

[1487] Specific examples

[1488] For example, suppose a user enters an advertisement for a "new diet supplement" as follows:

[1489] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1490] The specific processing flow when this advertisement is submitted by a user is as follows.

[1491] 1. User enters advertising content:

[1492] The user enters the above text into the form of the ad management system and clicks the "Submit" button.

[1493] 2. The device sends the advertising content:

[1494] The user's device sends the advertising content and associated metadata to the server via an HTTPS request.

[1495] 3. The server analyzes the ad content:

[1496] A server receives the advertising content and extracts specific keywords or phrases to generate generated prompts.

[1497] 4. The server analyzes the user's emotions:

[1498] The server uses an emotion engine to analyze the emotional impact of advertising content on the user.

[1499] 5. Generate content based on a generation prompt:

[1500] The server uses generative AI to generate similar advertising content.

[1501] 6. The server compares the original ad with the generated ad:

[1502] The server evaluates the similarity between the generated advertising content and the original advertisement and calculates a similarity score.

[1503] 7. The server uses the sentiment analysis results to determine harmful ads:

[1504] The server combines the similarity score with the sentiment analysis results to determine whether the ad is harmful.

[1505] 8. The server blocks harmful ads and notifies the user:

[1506] The server blocks ads that are determined to be harmful and notifies the user that "the ad has been determined to be harmful and has been blocked."

[1507] By taking into account the results of user sentiment analysis, the system of the present invention can detect harmful advertisements with higher accuracy than conventional methods that only use similarity evaluation. In addition, appropriate feedback is provided to users, ensuring transparency.

[1508] The processing flow will be explained below.

[1509] Step 1:

[1510] The user enters the advertising content. The user logs in to the advertising management system and enters the advertising content in the form of text, image, or video in the advertising input form. After entering the content, the user clicks the "Submit" button to submit the content.

[1511] Step 2:

[1512] The device sends the ad content to the server. The user's device generates an HTTPS request containing the entered ad content and its associated metadata (e.g., advertiser ID, category, target audience) and sends it to the server.

[1513] Step 3:

[1514] The server receives the ad content. The server receives the request and temporarily stores the ad content and metadata in a database.

[1515] Step 4:

[1516] The server analyzes the ad content. The server uses generative AI to generate potential prompts from the received ad content. This analysis utilizes natural language processing, image recognition, and video analysis technologies to extract specific keywords and phrases, for example.

[1517] Step 5:

[1518] The server analyzes the user's emotions. The server uses an emotion engine to classify the user's emotions toward the advertising content as positive, negative, or neutral. This can be done using text, voice, or image analysis.

[1519] Step 6:

[1520] The server generates new content based on the generated prompt. The generative AI uses the prompt generated in step 4 to generate new, similar advertising content. This generated content can also be in the form of text, images, or videos.

[1521] Step 7:

[1522] The server compares the original advertising content with the generated content. The server uses natural language processing technology, image recognition technology, or video analysis technology to compare the similarity between the generated content and the original advertising content. Based on the similarity evaluation, the server calculates a similarity score.

[1523] Step 8:

[1524] The server integrates the similarity score and the user's sentiment analysis results, and performs a comprehensive evaluation by combining the similarity score and the sentiment analysis results.

[1525] Step 9:

[1526] The server judges harmful advertisements. If the overall evaluation based on the similarity and sentiment analysis results exceeds a certain threshold, the advertisement is judged as harmful. Conversely, if it is below the threshold, the advertisement is not judged as harmful.

[1527] Step 10:

[1528] The server blocks harmful ads. The server blocks ads that are determined to be harmful, updates the ad's status to "blocked," and saves it in the database.

[1529] Step 11:

[1530] The server notifies the user. The server automatically sends an email to the user informing them that the ad was determined to be harmful and has been blocked. The notification also includes details of the reason for the block, the similarity score, and the results of sentiment analysis.

[1531] These steps enable the system of the present invention to not only efficiently detect and block harmful advertising content, but also provide feedback to help users understand the process.

[1532] Example 2

[1533] 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."

[1534] Conventional advertising content filtering systems have limited accuracy in determining whether an ad's content is harmful, leading to false positives. Furthermore, because they do not consider users' emotional responses, they can display emotionally unpleasant ads. This can lead to a poor user experience and a loss of trust in advertisers.

[1535] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1536] In this invention, the server includes a means for analyzing a user's emotions, a means for analyzing received content and generating a generation prompt, and a means for generating new content using the generation prompt, thereby enabling highly accurate determination of harmful content that takes user emotions into consideration.

[1537] "User" means a person who inputs and transmits advertising content using the system.

[1538] "Advertising Content" means a collection of information entered by a user that is analyzed and evaluated by the system, and may take the form of text, images, or video.

[1539] The "means for receiving" is a part of the system that has the functionality for capturing advertising content sent by a user.

[1540] The "means for analyzing" is a part of the system that has the function of analyzing received advertising content and extracting useful information.

[1541] A "generative prompt" is input information given to a generative AI model based on the analysis results of advertising content.

[1542] A "means for generating" is a part of the system that has the ability to create new content using generation prompts.

[1543] A "comparison means" is a part of the system that has the ability to evaluate the original advertising content against the generated content.

[1544] The "means for assessing similarity" is a part of the system that has the function of quantifying the similarity between the original advertising content and the generated content.

[1545] The "means for determining that content is harmful" is a part of the system that has the function of determining whether advertising content is harmful based on the results of similarity evaluation and sentiment analysis.

[1546] "Blocking means" is a part of the system that has the function of preventing advertising content that is determined to be harmful from being displayed.

[1547] The "means for notifying the user" is a part of the system that has the function of notifying the user that advertising content has been determined to be harmful.

[1548] The "means for analyzing emotions" is a part of the system that has the function of evaluating the emotional impact that advertising content has on the user and extracting emotions such as positive, negative, or neutral.

[1549] "Natural language processing technology" is a technology for analyzing the meaning and content of text data.

[1550] "Image recognition technology" is a technology for analyzing the contents of image data.

[1551] "Video analysis technology" is a technology for analyzing the content of video data.

[1552] The present invention is a system that receives advertising content entered by a user and automatically determines whether that content is harmful and blocks it. Furthermore, by analyzing user sentiment, the accuracy of determining harmful advertisements is improved. Below, specific embodiments of the present invention are described.

[1553] System Configuration

[1554] User Roles

[1555] The user inputs the advertising content through the advertising management system. The advertising content can be provided in the form of text, images, or videos. Once the user inputs the advertising content and clicks the "Submit" button, the content is received by the system.

[1556] Device Role

[1557] When a user submits advertising content, the device sends the entered advertising content and related metadata (e.g., advertiser ID, category, target audience) to the server using an HTTPS request. The device uses an existing web browser or a dedicated application.

[1558] Server Roles

[1559] The server receives the advertising content sent by the user and performs the following processes.

[1560] 1. Advertising content analysis:

[1561] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies. Specifically, it uses open source NLP libraries (e.g., spaCy, NLTK), image recognition libraries (e.g., OpenCV, TensorFlow), and video analysis technologies (e.g., FFmpeg).

[1562] Specific keywords, phrases, and visual features are extracted from the analyzed data, and generative prompt sentences are generated based on these.

[1563] 2. User sentiment analysis:

[1564] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the emotional impact of the ad content on the user, categorizing emotions into positive, negative, and neutral categories.

[1565] 3. Generate new content based on a generation prompt:

[1566] The server uses a generative AI model (e.g., GPT-4, DALL-E) to generate new ad content based on the generated prompt. The generated content matches the original format (text, image, video).

[1567] 4. Comparison of original ad content and generated content:

[1568] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score.

[1569] 5. Determining harmful advertising:

[1570] The server determines whether an advertisement is harmful based on the similarity score and the result of sentiment analysis. If the similarity score or sentiment score exceeds a certain threshold, the advertisement is determined to be harmful.

[1571] 6. Block harmful ads and notify users:

[1572] The server blocks the ad that is determined to be harmful, updates the ad's status to "blocked," and notifies the user by email that the ad has been determined to be harmful and blocked.

[1573] Specific examples

[1574] For example, if a user enters an ad for a "new diet supplement" like this:

[1575] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1576] Once this ad content is sent, the server performs the following steps:

[1577] 1. User enters and submits ad:

[1578] The user enters text into the ad management system and clicks the "Submit" button.

[1579] 2. The device sends the advertising content:

[1580] The device sends advertising content to the server via an HTTPS request.

[1581] 3. The server analyzes the advertising content:

[1582] Keywords such as "innovative," "weight loss effect," and "special price" are extracted, and prompt sentences are generated to be given to the generative AI model.

[1583] 4. The server analyzes the user's emotions:

[1584] Using an emotion engine, the emotions that ad text evokes in users are analyzed, and an emotion score is obtained, such as 80% positive, 15% negative, and 5% neutral.

[1585] 5. Generate new content based on a generation prompt:

[1586] A prompt sentence is input into the generative AI model to generate new advertising text, such as "Why not lose weight in a healthy way with this supplement?"

[1587] 6. Server compares original ad with new ad:

[1588] The original ad and the new ad are compared using natural language processing and a similarity score (e.g., 85 points) is calculated.

[1589] 7. The server determines whether the ad is harmful:

[1590] Advertisements are analyzed based on similarity scores and sentiment scores to determine whether they are harmful.

[1591] 8. Blocking harmful ads and notifying users:

[1592] Update the ad status to "Blocked" and notify the user that "The ad has been identified as harmful and blocked."

[1593] The system of the present invention enables automatic detection and blocking of harmful advertisements with high accuracy, taking into account the user's emotional response, thereby improving the user experience and ensuring advertising safety.

[1594] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1595] Step 1:

[1596] User enters and submits advertising content

[1597] The user uses the interface of the ad management system to input the ad content (in the form of text, images, or videos). Then, he clicks the "Submit" button to send the input content. The input data is the ad content provided by the user, and the output data is the ad content sent to the terminal. Specifically, the user attaches text, images, or videos to a form and submits it.

[1598] Step 2:

[1599] The device sends advertising content to the server

[1600] The device sends the advertising content and related metadata entered by the user to the server using an HTTPS request. The input data is the content and metadata entered by the user, and the output data is the content data sent to the server. Specifically, the device converts the user's input data into packets and sends them to the server as an encrypted HTTPS request.

[1601] Step 3:

[1602] The server analyzes the ad content and generates a prompt

[1603] The server analyzes the received advertising content. It uses natural language processing techniques (e.g., spaCy, NLTK), image recognition techniques (e.g., OpenCV, TensorFlow), and video analysis techniques (e.g., FFmpeg) to extract specific keywords, phrases, and visual features from the content. The input data is the advertising content sent to the server, and the output data is the generated prompt. Specifically, the server runs the analysis engine to extract keywords such as "weight loss" and "special price" from the original content and compose the generated prompt.

[1604] Step 4:

[1605] The server analyzes the user's emotions

[1606] The server uses an emotion engine (e.g., Affectiva, Google Cloud Natural Language) to analyze the user's emotions toward the ad content. The input data here is the analyzed ad content, and the output data is an emotion score such as positive, negative, or neutral. Specifically, the server passes the content data to the emotion analysis module and obtains the emotion score.

[1607] Step 5:

[1608] Generate new content based on a prompt

[1609] The server uses the generation prompt sentence to generate new advertising content for a generative AI model (e.g., GPT-4, DALL-E). The input data is the generation prompt sentence, and the output data is the generated new content. Specifically, the server inputs the generation prompt into the AI ​​model, and the model generates new advertising content (e.g., text and images).

[1610] Step 6:

[1611] The server compares the original ad content with the new content

[1612] The server uses natural language processing and image recognition technologies to compare the original advertising content with the generated content and calculate a similarity score. The input data is the original advertising content and the generated new content, and the output data is the similarity score. Specifically, the server inputs both contents into an analysis engine and quantifies the similarity.

[1613] Step 7:

[1614] The server determines harmful ads

[1615] The server determines whether an ad is harmful based on the similarity score and the results of sentiment analysis. The input data are the similarity score and sentiment score, and the output data is the harmfulness determination result. Specifically, the server compares the obtained scores and determines that an ad is harmful if it exceeds a threshold.

[1616] Step 8:

[1617] The server blocks harmful ads and notifies the user

[1618] The server blocks ads that are determined to be harmful and updates the ad status to "blocked." It also automatically sends a notification email to the user informing them that the ad has been determined to be harmful and blocked. The input data is the harmful determination result, and the output data is the updated ad status and notification email. Specifically, the server updates the ad database and sends a notification email to the user.

[1619] (Application example 2)

[1620] 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."

[1621] With the spread of internet advertising, the risk of harmful advertising content adversely affecting users is increasing. Therefore, a system that automatically monitors advertisements and quickly detects and blocks harmful advertisements is needed. Furthermore, conventional systems that only use similarity evaluation are sometimes insufficient in accuracy, so there is a need for a highly accurate harmful advertisement detection system that takes into account the emotional impact of users.

[1622] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving advertising content entered by a user, means for analyzing the received advertising content and generating a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content and evaluating the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user. This makes it possible to detect harmful advertisements quickly and accurately, and maintain the integrity of advertisements while minimizing the adverse impact on users.

[1623] "User-entered advertising content" refers to data in the form of text, images, or videos of advertisements provided by users such as advertisers or advertising agencies through online platforms.

[1624] The "receiving means" refers to a communication interface and software that has the function of receiving the advertising content input by the user on the server side.

[1625] "Means for analyzing and generating generative prompts" refers to the function of analyzing received advertising content and creating instructions (prompts) for generating new content using generative AI based on that content.

[1626] "Means for generating new content using generative prompts" refers to the ability to automatically generate new advertising content based on generative prompts using a generative AI model.

[1627] "Means for comparing generated content and assessing similarity" refers to algorithms and technologies that match original advertising content with generated advertising content and measure their similarity.

[1628] "Means for recognizing user emotions and assessing the emotional impact of advertising" refers to a sentiment analysis engine and associated algorithms for analyzing the emotional response of advertising content to users and assessing its impact.

[1629] The term "certain threshold" refers to a predetermined reference value for determining whether an advertisement is harmful or not, based on the results of similarity assessment and sentiment analysis.

[1630] "Means for determining harmful advertising" refers to an algorithm that automatically determines that advertising content is harmful if the results of similarity assessment and sentiment analysis exceed a certain threshold.

[1631] "Means for blocking and notifying users" refers to the function of excluding advertising content that is determined to be harmful from display and distribution networks, and automatically notifying advertisers of this fact.

[1632] This invention relates to an automated system for accurately determining the harmfulness of advertising content entered by a user and blocking it. The system of the present invention mainly includes the following main means: means for receiving advertising content entered by a user, means for analyzing the received advertising content to generate a generation prompt, means for generating new content using the generation prompt, means for comparing the original advertising content with the generated content to evaluate the similarity, means for recognizing the user's emotions and evaluating the emotional impact of the advertisement, means for determining that the advertisement is harmful if the evaluation results of the similarity and emotional impact exceed certain thresholds, and means for blocking the harmful advertisement based on the determination result and notifying the user.

[1633] Program processing explanation

[1634] 1. Receiving advertising content

[1635] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. After entering the content, the user clicks the "Submit" button, which sends the content and its metadata (advertiser ID, category, target audience) to the server. This communication uses an HTTPS request.

[1636] 2. Analyzing advertising content and generating prompts

[1637] The server analyzes the received advertising content using natural language processing (NLP), image recognition, and video analysis technologies to extract specific keywords, phrases, and visual features. Based on the analysis results, a prompt is generated, and new advertising content is automatically generated based on the advertising content.

[1638] 3. Emotion analysis

[1639] The server evaluates the emotional impact of advertising content using a sentiment analysis engine that uses existing sentiment analysis software, such as NLTK's SentimentIntensityAnalyzer, to extract positive, negative, or neutral sentiment scores from text, audio, or image data.

[1640] 4. Similarity evaluation

[1641] The server compares the original ad content with the generated similar content and calculates a similarity score using natural language processing, image recognition, or video analysis technology. For example, it can use Hugging Face's transformer library to calculate the similarity between the similar content generated by the GPT-3 model and the original content.

[1642] 5. Identifying and blocking harmful ads

[1643] The server combines the results of the similarity assessment and sentiment analysis to determine whether an ad is harmful. Ads that are determined to be harmful are automatically blocked and users are notified of this. Notifications are sent via email or system alerts.

[1644] Specific examples

[1645] For example, if a user enters the following ad content:

[1646] "Try this revolutionary supplement and experience amazing weight loss results in a short amount of time! Now available at a special price!"

[1647] When this ad is submitted, the system works as follows: First, the ad content is received, the text is analyzed, and a generative prompt is generated. Using the generative prompt, a generative AI model (GPT-3) generates similar content, such as:

[1648] "Just take this new supplement and lose weight fast! Try it at a great price!"

[1649] The sentiment analysis engine then evaluates the emotional impact and calculates the similarity between the original content and the generated content. Based on this result, it determines whether the ad is harmful, and if so, the ad is blocked and the user is notified. This allows for early detection of harmful content and prevents adverse effects on users.

[1650] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1651] Step 1:

[1652] The user enters advertising content into an online ad management system. The advertising content can be in the form of text, images, or videos. The user clicks the "Submit" button, which sends the entered content and its metadata (advertiser ID, category, target audience) to the server. The entered data is securely transmitted to the server via an HTTPS request.

[1653] Step 2:

[1654] The server analyzes the received advertising content. First, it performs text analysis and uses natural language processing (NLP) techniques to extract specific keywords and phrases within the content. It also uses image and video analysis techniques to extract visual features. These analyses result in a generated prompt. The input is the advertising content and related metadata, and the output is the analysis results and the generated prompt.

[1655] Step 3:

[1656] The server uses the generative prompts to generate new advertising content. A generative AI model (e.g., GPT-3) is used to create similar advertising content based on the generative prompts. The generated content is in the same format (text, image, or video) as the original. The input is the generative prompts, and the output is the generated advertising content.

[1657] Step 4:

[1658] The server uses a sentiment analysis engine to evaluate the emotional impact of ad content on users. Sentiment analysis software such as NLTK's SentimentIntensityAnalyzer extracts positive, negative, or neutral sentiment scores from text, audio, or image data. The input is the ad content, and the output is the sentiment analysis results.

[1659] Step 5:

[1660] The server compares the generated ad content with the original ad content and calculates a similarity score. It uses natural language processing, image recognition, or video analysis technology to evaluate the similarity based on the features of both pieces of content. For example, it uses Hugging Face's transformers library to compare it with the output of the GPT-3 model. The input is the original ad content and the generated content, and the output is a similarity score.

[1661] Step 6:

[1662] The server determines whether an advertisement is harmful or not based on the similarity score and the results of sentiment analysis. The criteria (certain threshold) for determining whether an advertisement is harmful is set in advance, and if the similarity score and the results of sentiment analysis exceed this threshold, the advertisement is determined to be harmful. The input is the similarity score and the results of sentiment analysis, and the output is the result of determining whether the advertisement is harmful.

[1663] Step 7:

[1664] The server blocks advertising content that is determined to be harmful. It updates the status of the advertisement to "blocked" and saves it in the database. It also automatically sends a notification to the user that the advertisement has been determined to be harmful and blocked. The input is the result of determining whether the advertisement is harmful, and the output is the notification to the user and the blocking status of the advertising content.

[1665] Through this series of processes, the system can accurately determine whether an advertisement is harmful and prevent the distribution of harmful advertising content.

[1666] 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.

[1667] 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.

[1668] 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.

[1669] 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.

[1670] 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.

[1671] 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.

[1672] 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).

[1673] 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.

[1674] 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."

[1675] 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.

[1676] 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).

[1677] 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.

[1678] 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.

[1679] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1680] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1681] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1682] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1683] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1684] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1685] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1686] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1687] The following is further disclosed regarding the above embodiment.

[1688] (Claim 1)

[1689] means for receiving user-entered advertising content;

[1690] means for analyzing the received advertising content to generate a generated prompt;

[1691] means for generating new content using generation prompts;

[1692] a means for comparing the original advertising content with the generated content to assess similarity;

[1693] A means for determining that an advertisement is harmful when the degree of similarity exceeds a certain threshold;

[1694] A means for blocking harmful advertisements based on the determination result and notifying the user;

[1695] A system including:

[1696] (Claim 2)

[1697] 10. The system of claim 1, wherein the advertising content is in the form of text, an image, or a video.

[1698] (Claim 3)

[1699] 2. The system according to claim 1, wherein the similarity evaluation is performed using natural language processing technology, image recognition technology, or video analysis technology.

[1700] "Example 1"

[1701] (Claim 1)

[1702] means for receiving user-entered advertising content;

[1703] means for analyzing the received advertising content to generate a generated prompt;

[1704] means for generating new content using generation prompts;

[1705] a means for comparing the original advertising content with the generated content to assess similarity;

[1706] A means for determining that an advertisement is harmful when the degree of similarity exceeds a certain threshold;

[1707] A means for blocking harmful advertisements based on the determination result and notifying the user;

[1708] means for transmitting information including metadata related to advertising content;

[1709] means for extracting specific keywords or phrases based on the received advertising content;

[1710] a means for calculating a similarity of the generated advertising content;

[1711] A system including:

[1712] (Claim 2)

[1713] 10. The system of claim 1, wherein the advertising content is in the form of text, an image, or a video.

[1714] (Claim 3)

[1715] 2. The system according to claim 1, wherein the similarity evaluation is performed using natural language processing technology, image recognition technology, or video analysis technology.

[1716] "Application Example 1"

[1717] (Claim 1)

[1718] means for receiving user-entered advertising content;

[1719] means for analyzing the received advertising content to generate a generated prompt;

[1720] means for generating new content using generation prompts;

[1721] a means for comparing the original advertising content with the generated content to assess similarity;

[1722] A means for determining that an advertisement is harmful when the degree of similarity exceeds a certain threshold;

[1723] A means for blocking harmful advertisements based on the determination result and notifying the user;

[1724] a means for evaluating the similarity between the generated content and the original advertisement content and determining harmful advertisements based on the evaluation results;

[1725] A means for notifying the user of a block if the advertisement is determined to be harmful;

[1726] A system including:

[1727] (Claim 2)

[1728] 10. The system of claim 1, wherein the advertising content is in the form of text, an image, or a video.

[1729] (Claim 3)

[1730] 2. The system according to claim 1, wherein the similarity evaluation is performed using natural language processing technology, image recognition technology, or video analysis technology.

[1731] "Example 2: Combining Emotion Engines"

[1732] (Claim 1)

[1733] means for receiving user-entered content;

[1734] means for parsing the received content to generate a generation prompt;

[1735] means for generating new content using generation prompts;

[1736] a means for comparing the original content with the generated content to assess similarity;

[1737] means for determining that the content is harmful if the similarity exceeds a certain threshold;

[1738] a means for blocking harmful content based on the judgment result and notifying the user;

[1739] means for analyzing user emotions;

[1740] A means for improving the accuracy of judgment using the emotion analysis results;

[1741] A system including:

[1742] (Claim 2)

[1743] 10. The system of claim 1, wherein the content is in the form of text, images, or videos.

[1744] (Claim 3)

[1745] 2. The system according to claim 1, wherein the similarity evaluation is performed using natural language processing technology, image recognition technology, or video analysis technology.

[1746] "Application example 2 when combining emotion engines"

[1747] (Claim 1)

[1748] means for receiving user-entered advertising content;

[1749] means for analyzing the received advertising content to generate a generated prompt;

[1750] means for generating new content using generation prompts;

[1751] a means for comparing the original advertising content with the generated content to assess similarity;

[1752] means for recognizing a user's emotions and assessing the emotional impact of an advertisement;

[1753] A means for determining that an advertisement is harmful when the evaluation results of the similarity and emotional impact exceed a certain threshold;

[1754] A means for blocking harmful advertisements based on the determination result and notifying the user;

[1755] A system including:

[1756] (Claim 2)

[1757] 10. The system of claim 1, wherein the advertising content is in the form of text, an image, or a video.

[1758] (Claim 3)

[1759] 2. The system according to claim 1, wherein the similarity evaluation is performed using natural language processing technology, image recognition technology, or video analysis technology. [Explanation of symbols]

[1760] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving user-entered advertising content; means for analyzing the received advertising content to generate a generated prompt; means for generating new content using generation prompts; a means for comparing the original advertising content with the generated content to assess similarity; A means for determining that an advertisement is harmful when the degree of similarity exceeds a certain threshold; A means for blocking harmful advertisements based on the determination result and notifying the user; A system including:

2. 10. The system of claim 1, wherein the advertising content is in the form of text, images, or videos.

3. The system according to claim 1 , wherein the similarity evaluation is performed using natural language processing technology, image recognition technology, or video analysis technology.

Images