system

Abstract

A system is provided.SOLUTION: A system comprising: means for receiving a website access request from an Internet-connected device; means for obtaining content at a specified URL; means for analyzing text and images of the content using a generative model; means for determining inappropriate or fraudulent content based on an analysis result of the generative model; and means for blocking the inappropriate or fraudulent content and displaying a warning page.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 today's world, where internet usage is rapidly expanding, users have access to a wide variety of websites, exposing them to various risks in exchange for convenience. Minors and the elderly are particularly at risk of accessing fraudulent websites and inappropriate content, which can cause serious problems. There is also a risk of their mental health being damaged by unconscious exposure to information that causes mental stress.

[0005] There is a need for a filtering system that can solve the above problems and allow users to use the Internet safely and securely. [Means for solving the problem]

[0006] The present invention provides a filtering system including a means for receiving a website access request from an Internet-connected device, a means for obtaining the content of the specified URL, a means for analyzing the text and images of the content using a generative model, a means for determining inappropriate or fraudulent content based on the analysis results of the generative model, and a means for blocking the inappropriate or fraudulent content and displaying a warning page.

[0007] Furthermore, the present invention contributes to protecting mental health by adding a means to detect content that may cause mental stress and restrict access to that content.

[0008] In particular, by using natural language processing technology as a means of text analysis and computer vision technology as a means of image analysis, more sophisticated and effective content filtering can be achieved, allowing users to enjoy safe and healthy internet usage.

[0009] An "Internet-connected device" is an electronic device, such as a computer, smartphone, or tablet, that a user uses to connect to the Internet and access websites.

[0010] A "website access request" is a request sent to a specified URL through a browser or application when a user attempts to access a specific website.

[0011] "URL" is an abbreviation for "Uniform Resource Locator" and is a string of characters that indicates the location of a resource on the Internet.

[0012] "Content" refers collectively to information such as text, images, video, and audio provided on the Website.

[0013] A "generative model" is a machine learning algorithm or artificial intelligence model used to analyze data and make predictions, including, among other things, natural language processing and computer vision techniques.

[0014] "Text analysis" is the process of using natural language processing techniques to understand the meaning and structure of text and evaluate specific keywords and context.

[0015] "Image analysis" is the process of using computer vision techniques to understand the content of an image and detect inappropriate images or specific patterns.

[0016] "Inappropriate Content" is information that is deemed harmful, offensive, immoral or violent to users.

[0017] "Deceptive Content" is information that is intended to deceive users or that encourages fraudulent activity.

[0018] A "warning page" is a specific web page that is displayed to warn users who are attempting to access inappropriate or fraudulent content.

[0019] "Mental stress-inducing content" is information that may have a negative impact on a user's mental health, and includes specific trigger words or images.

[0020] "Natural language processing technology" is a technology that allows computers to understand, process, and generate human language.

[0021] "Computer vision technology" is a technology that allows computers to extract and analyze information from images and videos. [Brief explanation of the drawings]

[0022] [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

[0023] 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.

[0024] First, the terms used in the following description will be explained.

[0025] 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).

[0026] 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.

[0027] 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.

[0028] 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.

[0029] 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."

[0030] [First embodiment]

[0031] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0032] 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.

[0033] 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).

[0034] 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.

[0035] 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.

[0036] 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.

[0037] 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.

[0038] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0039] 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.

[0040] 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.

[0041] 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.

[0042] 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."

[0043] The present invention is a filtering system that receives website access requests from internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress.

[0044] Program processing

[0045] Server Processing

[0046] The server receives a URL access request sent from the user's device. After receiving the request, the server makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by the generative model.

[0047] The generative model uses natural language processing (NLP) technology to analyze the text content of a page and evaluate whether it contains harmful keywords or expressions, while also using computer vision technology to evaluate image data for inappropriate images.

[0048] If the analysis detects inappropriate or fraudulent content, the server blocks access to the content and sends a warning page to the user's device explaining the reason for the block and the importance of safe web browsing.

[0049] During the analysis, information that may cause psychological stress is also detected, and if such information is found, the server blocks access to it as well and sends a warning page.

[0050] Terminal handling

[0051] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0052] User Interface

[0053] Users can surf the internet using a regular browser or a dedicated application. If the website they are trying to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous content. It also provides advance warnings about information that may cause mental stress, protecting users' mental health.

[0054] Specific examples

[0055] Example 1: Accessing inappropriate content

[0056] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0057] Example 2: Visiting a fraudulent website

[0058] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[0059] Example 3: Visiting a stressful news site

[0060] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. The server blocks the access and sends a warning page. The terminal displays the warning page to the user, and their mental health is protected.

[0061] The above is a specific embodiment of the present invention.

[0062] The processing flow will be explained below.

[0063] Step 1:

[0064] A user attempts to access a specific URL using a browser or application.

[0065] Step 2:

[0066] The device sends an access request to the server for the specified URL.

[0067] Step 3:

[0068] The server logs URL access requests received from the device.

[0069] Step 4:

[0070] The server makes an HTTP request to the specified URL to retrieve the web page content.

[0071] Step 5:

[0072] The server separates the content of the acquired web page into text data and image data.

[0073] Step 6:

[0074] The server analyzes the text data using the generative model.

[0075] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[0076] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[0077] Step 7:

[0078] The server analyzes the image data using the generative model.

[0079] Computer vision techniques are used to evaluate the content of the image.

[0080] Detect inappropriate images and specific patterns.

[0081] Step 8:

[0082] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[0083] If there is a problem, the server will block access to the URL.

[0084] Step 9:

[0085] If the server blocks inappropriate or deceptive content, it generates a warning page.

[0086] The warning page will include information about why your access was blocked and the importance of safe internet use.

[0087] Step 10:

[0088] The server sends a warning page to the device.

[0089] Step 11:

[0090] The terminal displays the received warning page to the user.

[0091] Step 12:

[0092] The server analyzes content that may cause mental stress.

[0093] Detects trigger words and images to determine if they contain stress-related information.

[0094] Step 13:

[0095] If the server detects stress-related content, it will also block access to it.

[0096] In this case, a warning page is also generated and sent to the terminal.

[0097] Step 14:

[0098] The device will again display a warning page to the user to protect their mental health.

[0099] Through the above steps, the filtering system of the present invention provides an environment in which users can use the Internet safely and securely.

[0100] Example 1

[0101] 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."

[0102] The Internet contains inappropriate and fraudulent content that is harmful to many users, including minors and the elderly, and can also cause psychological stress. Therefore, a filtering system is needed to ensure that users can use the Internet safely and comfortably. However, existing filtering systems lack the ability to analyze text and images, making them unable to effectively block harmful content.

[0103] 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.

[0104] In this invention, the server includes means for receiving a website access request sent from a terminal, means for acquiring the content of the requested URL, means for separating the acquired content into text data and image data, means for analyzing the text data using a generative AI model to detect harmful keywords and expressions, means for analyzing the image data using computer vision technology to detect inappropriate images, means for determining inappropriate or fraudulent content based on the analysis results of the generative AI model, and means for blocking inappropriate or fraudulent content and sending a warning page to the terminal. This makes it possible to effectively block inappropriate or fraudulent content on the Internet and provide an environment in which users can use the Internet safely and comfortably.

[0105] A "terminal" is a device such as a computer or smartphone that allows a user to connect to the Internet.

[0106] A "server" is a computer system that provides information and performs processing on the Internet.

[0107] A "website access request" is a request that a user sends from a terminal to a server to access a specific website.

[0108] A "URL" is an address on the Internet that points to a specific web page.

[0109] "Content" is a general term for information displayed on a web page, and includes text data and image data.

[0110] A "generative AI model" is a statistical model for generating and analyzing data using artificial intelligence techniques.

[0111] "Natural language processing technology" is a technology that enables computers to understand and process human language.

[0112] "Computer vision technology" is a technology that allows computers to analyze and recognize images and videos.

[0113] A "warning page" is a web page that blocks access to inappropriate or fraudulent content and notifies users of this.

[0114] "Harmful keywords and expressions" are words or phrases that may threaten a user's mental health or safe use of the Internet.

[0115] An "inappropriate image" is an image that is perceived by a user as psychologically harmful or unpleasant.

[0116] "Fraudulent content" is information intended to deceive users and cause them financial harm.

[0117] "Mental stress" is the emotional strain that causes a user's mental health to deteriorate.

[0118] There is a lot of inappropriate or fraudulent content on the Internet that can threaten users' safety and mental health. The present invention provides an advanced filtering system to protect users from such harmful content.

[0119] Server hardware and software configuration

[0120] The server uses a high-performance computer system and has the following software configuration:

[0121] 1. Web server software: Use Apache HTTP Server or Nginx.

[0122] 2. Python programming environment: Uses the Python language and related libraries to perform natural language processing, image analysis, etc.

[0123] 3. Generative AI models: Use Hugging Face's Transformers model or OpenAI's GPT-3.

[0124] 4. Natural Language Processing techniques: Use pre-trained models such as BERT and RoBERTa.

[0125] 5. Computer vision technology: Image analysis is performed using TensorFlow and Detectron2.

[0126] System Operation Overview

[0127] The system of the present invention receives a website access request sent from a user's device and retrieves the content of the specified URL based on the request. The retrieved content is separated into text data and image data, each of which is analyzed using a generative AI model. If inappropriate or fraudulent content is detected based on the analysis results, the server blocks access to the content and sends a warning page to the user's device. In addition, content that may cause mental stress may also be detected, and access to such information is also restricted.

[0128] Specific examples

[0129] Example 1: Accessing inappropriate content

[0130] When a minor user attempts to access a website containing adult content using a browser, an access request is sent from the device to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is inappropriate, blocks the access. A warning page is displayed on the device, preventing access to the adult content.

[0131] Example 2: Visiting a fraudulent website

[0132] An elderly person clicks on a link in an email and attempts to access a financial fraud website. In this case, the device sends an access request to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is fraudulent, blocks the access. A warning page is displayed on the device, preventing access to the fraudulent website.

[0133] Example 3: Visiting a stressful news site

[0134] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server will analyze the text and image content and determine that it may cause mental stress. The server will block the access and send a warning page, which will be displayed on the device, protecting the user's mental health.

[0135] Prompt Sentence Examples

[0136] "Please explain in natural language how you analyze the URL of the website a user is attempting to access, determine if the URL contains inappropriate or deceptive content, and block access if necessary. Also, please specify the names of any specific hardware or software used."

[0137] As a result, by using the system of the present invention, users are protected from harmful content on the Internet, enabling safe and comfortable web browsing.

[0138] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0139] Step 1:

[0140] The server receives a website access request sent from the user's terminal.

[0141] Input: An HTTP GET request containing the URL of the website the user wants to access.

[0142] What happens: The server receives this request using web server software (e.g., Apache HTTP Server or Nginx).

[0143] Output: The requested URL.

[0144] Step 2:

[0145] The server makes an HTTP request to the received URL to retrieve the web page content.

[0146] Input: The requested URL.

[0147] What happens: The server uses Python's requests library to download HTML data from the specified URL. It then executes requests.get(url) to get the HTML content of the web page.

[0148] Output: The retrieved HTML data.

[0149] Step 3:

[0150] The server separates the acquired content into text data and image data.

[0151] Input: The retrieved HTML data.

[0152] What happens: The server uses the BeautifulSoup library to parse the HTML data and extract text and image data. It runs soup = BeautifulSoup(html_data, 'html.parser') and separates the data using functions like text_data = soup.get_text() and image_urls = [img['src'] for img in soup.find_all('img')] .

[0153] Output: Separated text and image data.

[0154] Step 4:

[0155] The server uses a generative AI model to analyze the extracted text data and detect harmful keywords and expressions.

[0156] Input: Separated text data.

[0157] Specific operation: The server uses Hugging Face's Transformers library to analyze the data by calling a generative AI model (e.g., BERT, RoBERTa). It then executes nlp_pipeline = pipeline('sentiment-analysis') and runs result = nlp_pipeline(text_data) to detect harmful keywords.

[0158] Output: A list of harmful keywords and expressions contained in the text data.

[0159] Step 5:

[0160] The server uses computer vision technology to analyze the extracted image data and detect inappropriate images.

[0161] Input: Segmented image data.

[0162] Specific operation: The server performs image analysis using TensorFlow and Detectron2. It downloads an image, inputs it into the model, executes detectron2_predictor = DefaultPredictor(cfg) and outputs = detectron2_predictor(image), and evaluates the image.

[0163] Output: A list of inappropriate images.

[0164] Step 6:

[0165] The server uses the analysis to determine whether the URL contains inappropriate or deceptive content.

[0166] Input: Analysis results of text data and image data.

[0167] What it does: The server evaluates the results of the generative AI model and determines whether they contain inappropriate or deceptive content, for example, by using conditional branching such as if 'toxic' in text_analysis_result or 'unsafe' in image_analysis_result.

[0168] Output: Access blocking decision result (whether to block or not).

[0169] Step 7:

[0170] If the server determines that the content is inappropriate or fraudulent, it will send a warning page to the user's device.

[0171] Input: Block decision result.

[0172] Specific operation: The server generates a warning page using HTML and CSS and sends it to the terminal as an HTTP response. response_html = " <h1> Warning< / h1> This site is blocked. " and send it as return response_html.

[0173] Output: A warning page that is displayed on the user's device.

[0174] Step 8:

[0175] The terminal displays a warning page to the user.

[0176] Input: The HTML of the warning page received from the server.

[0177] Specific operation: The user's device renders the received HTML in the browser and displays the warning content.

[0178] Output: The warning page that the user sees.

[0179] (Application example 1)

[0180] 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."

[0181] As the Internet becomes more widespread, users are increasingly at risk of accessing inappropriate content or fraudulent websites. Furthermore, there is a large amount of information that can potentially cause psychological stress, making it difficult for users to use the Internet safely and comfortably. Therefore, there is a need for a system that can detect dangerous content in real time and protect users.

[0182] 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.

[0183] In this invention, the server includes means for receiving a website access request from an Internet connection device, means for acquiring data from a specified URL, means for analyzing text and images in the data using a generative model, means for determining whether the data is inappropriate or fraudulent based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent data and displaying a warning page, and means for scanning data in real time and monitoring user Internet usage, thereby enabling users to use the Internet safely and comfortably.

[0184] An "Internet connection device" is a device for connecting to the Internet and capable of sending access requests to websites.

[0185] A "website access request" is a request sent from an Internet-connected device when a user attempts to access a specific website on the Internet.

[0186] A "specified URL" refers to the address of a particular web page that a user wishes to access.

[0187] "Data" means the content on the Website, including text and image data.

[0188] A "generative model" is a model based on artificial intelligence techniques used to analyze text and image data.

[0189] "Analysis results" refers to the analysis results of data obtained by a generative model, which are used to determine whether the data is inappropriate or fraudulent.

[0190] "Inappropriate Content" means content that is deemed harmful or offensive to users.

[0191] "Deceptive data" is content determined to be intended to deceive or mislead users.

[0192] A "warning page" is a page that is displayed when inappropriate or fraudulent data is detected to notify the user and block access.

[0193] "Real-time scanning" refers to the process of analyzing data as soon as it is acquired.

[0194] "Internet usage monitoring measures" means measures that continuously track and analyze users' Internet activity to detect inappropriate or fraudulent data.

[0195] Server Processing

[0196] The server receives a website access request sent from the user's Internet connection device. Based on the received access request, the server makes an HTTP request to the specified URL to obtain the web page data. This data is separated into text data and image data and analyzed using a generative model.

[0197] Generative models use natural language processing (NLP) techniques to analyze the text content of web pages, for example, to assess whether the text contains harmful keywords or inappropriate language, and simultaneously use computer vision techniques to analyze image data and assess whether it contains inappropriate images.

[0198] If the analysis detects inappropriate or fraudulent data, the server blocks access to it and sends a warning page to the user's internet connection, explaining the reason for the block and the importance of safe web browsing.

[0199] Additionally, the server detects any information that may cause psychological stress, and if such information is included, the server will similarly block access and send a warning page.

[0200] Terminal handling

[0201] When a user attempts to access a specific URL, the user's Internet connection device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate data or a warning page and displays it to the user. If access to inappropriate data or a fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0202] User Interface

[0203] Users can use a regular browser or dedicated application to access the Internet. If the website they are about to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous data. Advance warnings are also provided for information that may cause mental stress, protecting users' mental health.

[0204] Hardware and software used

[0205] The system uses the following hardware and software:

[0206] Hardware: Internet-connected devices (smartphones, tablets, PCs, etc.)

[0207] Software: Server-side HTTP request processing frameworks (e.g., Python's requests library), natural language processing models (e.g., Hugging Face's transformers library), computer vision models (e.g., OpenCV)

[0208] Specific examples

[0209] If a user attempts to access a news site, but it is determined that the content of the news site may cause mental stress, the following prompt sentence is presented to the generative model for analysis:

[0210] "Does this news article contain content that may cause emotional stress to users?"

[0211] The generative model generates an answer to this prompt and decides whether to block access based on the answer, allowing users to always use the Internet safely and with peace of mind.

[0212] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0213] Step 1:

[0214] Receives website access requests from internet-connected devices

[0215] The server receives a website access request sent from the user's Internet connection device. As input, this request contains a URL. The server takes this URL and proceeds to the next step.

[0216] Step 2:

[0217] Gets the data of the specified URL

[0218] The server makes an HTTP request to the received URL to retrieve the web page content of that URL. The URL is given as input, and the text and image data of the web page are obtained as output. The server separates these data for the next step.

[0219] Step 3:

[0220] Analyzing data using generative models

[0221] The server analyzes the acquired text data and image data using a generative model. The text data and image data are given as input, and the analysis results are obtained as output. Specifically, the text data is analyzed using a natural language processing (NLP) model, and the image data is analyzed using a computer vision model.

[0222] Step 4:

[0223] Identifying inappropriate or fraudulent data

[0224] The server determines whether data is inappropriate or fraudulent based on the analysis results of the generative model. The analysis results are given as input, and the output is a judgment result on whether the data is inappropriate or not. Specifically, the NLP model detects harmful keywords and expressions in text, and the computer vision model detects inappropriate content in images.

[0225] Step 5:

[0226] Blocking data and displaying a warning page

[0227] If the server detects inappropriate or fraudulent data, it blocks access to that data. The input is the verdict, and the output is a blocked access flag and a warning page that explains the inappropriate content and the importance of safe web browsing.

[0228] Step 6:

[0229] Sending and displaying a warning page

[0230] The server sends the generated warning page to the user's Internet connection device. The warning page is given as input, and the warning page is displayed on the user's terminal as output. The terminal receives this warning page and displays it to the user.

[0231] Step 7:

[0232] Detecting and limiting data that may cause mental stress

[0233] During the analysis process, the server detects information that may cause mental stress. Text data and image data are input, and a judgment result regarding mental stress is obtained as output. If the information is determined to cause mental stress, it similarly blocks access and sends a warning page.

[0234] Specific actions

[0235] The natural language processing model uses pre-trained AI models (e.g., Hugging Face's transformers library) to analyze text data and generate answers to prompts such as, "Does this news article contain content that would cause mental stress to the user?"

[0236] Computer vision technology uses OpenCV and other image recognition models to scan image data in real time to detect whether it contains inappropriate images.

[0237] These specific processes enable users to use the Internet safely and comfortably.

[0238] 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.

[0239] The present invention provides a filtering system that receives a website access request from an internet-connected device, retrieves the content of the specified URL, analyzes the text and images of the content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health state in real time and suggest appropriate content according to the user's specific emotional state.

[0240] Program processing

[0241] Server Processing

[0242] The server receives a URL access request sent from the user's device. The server then makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by a generative model. The generative model uses natural language processing (NLP) technology to analyze the text content and detect harmful keywords and context. At the same time, computer vision technology is used to detect inappropriate images in the image data.

[0243] Emotion engine processing

[0244] The server recognizes the user's emotions using an emotion engine. The emotion engine analyzes the user's device's camera, microphone, and text input data to determine the user's emotional state from their facial expressions, voice, and text input. Based on the results, the server grasps the user's current emotional state in real time.

[0245] Filtering and Content Suggestions

[0246] If inappropriate or fraudulent content is detected based on the analysis results of the generative model, the server blocks access to the content and sends a warning page to the user's device. On the other hand, if content that may cause mental stress is detected, the server blocks access and displays a warning page in combination with the results of the emotion engine.

[0247] Furthermore, the server can suggest content that takes into account the user's mental health based on the results of the emotion engine. For example, if a user is under stress, it can suggest relaxing content, soothing music, or videos for relaxation.

[0248] Terminal handling

[0249] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0250] The device also uses a camera and microphone to provide data to the emotion engine, monitoring the user's emotional state in real time.

[0251] User Interface

[0252] Users surf the Internet using a regular browser or a dedicated application. If the website they are attempting to access is deemed inappropriate or fraudulent, a warning page is displayed, protecting the user from dangerous content. The system also provides advance warnings about information that may cause mental stress, maintaining the user's mental health. Furthermore, an emotion engine monitors the user's emotional state in real time and suggests relaxing content as needed.

[0253] Specific examples

[0254] Example 1: Accessing inappropriate content

[0255] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0256] Example 2: Visiting a fraudulent website

[0257] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[0258] Example 3: Visiting a stressful news site

[0259] If a user attempts to access a news site, but the content of the site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. Furthermore, an emotion engine monitors the user's emotional state. If the server determines that the user is in a stressful state, the server blocks the access and sends a warning page to the device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[0260] The above is a specific embodiment of the present invention.

[0261] The processing flow will be explained below.

[0262] Step 1:

[0263] A user attempts to access a specific URL using a browser or application.

[0264] Step 2:

[0265] The device sends an access request to the server for the specified URL.

[0266] Step 3:

[0267] The server logs URL access requests received from the device.

[0268] Step 4:

[0269] The server makes an HTTP request to the specified URL to retrieve the web page content.

[0270] Step 5:

[0271] The server separates the content of the acquired web page into text data and image data.

[0272] Step 6:

[0273] The server analyzes the text data using the generative model.

[0274] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[0275] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[0276] Step 7:

[0277] The server analyzes the image data using the generative model.

[0278] It uses computer vision techniques to evaluate the content of images and detect inappropriate images or specific patterns.

[0279] Step 8:

[0280] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[0281] If there is a problem, the server will block access to the URL.

[0282] Step 9:

[0283] If the server blocks inappropriate or deceptive content, it generates a warning page.

[0284] The warning page will include information about why your access was blocked and the importance of safe internet use.

[0285] Step 10:

[0286] The server sends a warning page to the device.

[0287] Step 11:

[0288] The terminal displays the received warning page to the user.

[0289] Step 12:

[0290] The server analyzes content that may cause mental stress.

[0291] Detects trigger words and images to determine if they contain stress-related information.

[0292] Step 13:

[0293] If the server detects stress-related content, it will also block access to it.

[0294] In this case, a warning page is also generated and sent to the terminal.

[0295] Step 14:

[0296] The device will again display a warning page to the user to protect their mental health.

[0297] Emotion engine processing

[0298] Step 15:

[0299] The server uses an emotion engine to recognize the user's emotional state.

[0300] It collects and analyzes data from the device's camera, microphone, and text input, and determines the user's emotional state from their facial expressions, voice, and text input.

[0301] Step 16:

[0302] If the emotion engine determines that the user is stressed, the server uses this information to make decisions about suggesting more appropriate content.

[0303] Step 17:

[0304] The server selects content to reduce stress based on the results of the emotion engine.

[0305] It includes relaxing videos, music, articles, etc.

[0306] Step 18:

[0307] The server generates a page for proposing the selected stress reduction content to the user.

[0308] Step 19:

[0309] The server generates a proposal page and sends it to the terminal.

[0310] Step 20:

[0311] The terminal displays the received proposal page to the user.

[0312] It is expected that users will view the suggested content and reduce their mental stress.

[0313] The above is the flow of processing in a specific embodiment of the present invention.

[0314] Example 2

[0315] 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."

[0316] There is a lot of inappropriate or fraudulent content on the Internet, and accessing it poses a significant risk to users. Furthermore, involuntary access to content that causes mental stress can have a negative impact on users' mental health. Conventional filtering systems have had difficulty completely eliminating such risks. The present invention aims to solve these problems, protect users from harmful content on the Internet, and maintain their mental health.

[0317] 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.

[0318] In this invention, the server includes means for receiving a website access request from an Internet-connected device, means for retrieving content from a specified URL, means for analyzing text and images in the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for collecting data to recognize a user's emotional state, and means for analyzing the collected data using an emotion engine to determine the user's emotional state. This protects users from inappropriate or fraudulent content and from content that may cause mental stress, thereby enabling them to maintain their mental health.

[0319] An "Internet-connected device" is a device that has the ability to connect to the Internet and is capable of web browsing and data communication. Examples include smartphones, tablets, and personal computers.

[0320] A "website access request" is an HTTP request sent when a user attempts to access a particular web page using an Internet-connected device.

[0321] A "specified URL" is the address of a particular web page that a user wishes to access, written in Uniform Resource Locator (URL) format.

[0322] A "generative model" is an AI model that is trained on large datasets and used to analyze text and images, allowing it to detect specific patterns and anomalies.

[0323] "Text analytics" is the process of analyzing text data using generative models, specifically using natural language processing techniques to understand and classify the content of the text.

[0324] "Image analysis" is the process of analyzing image data using generative models. It refers to the use of computer vision techniques to detect features and anomalies in images.

[0325] "Inappropriate Content" refers to content that is inappropriate for minors, or that contains violent, sexual, or discriminatory material, and that may have a harmful effect on users.

[0326] "Fraudulent content" refers to web content created with the intention of deceiving users, and which may result in financial loss or the leakage of personal information.

[0327] A "warning page" is a page that is displayed when the web page that a user is attempting to access is determined to be inappropriate or fraudulent, and serves to warn the user.

[0328] An "emotion engine" is a device or software that analyzes a user's emotional state and determines the emotional state based on the user's facial expressions, voice, text input, etc.

[0329] "Mental stress" refers to the psychological burden or state of tension caused by external stimuli or information.

[0330] "Content suggestion" is the process of recommending suitable content, such as videos or music with a relaxing effect, taking into account the user's current emotional state and mental health.

[0331] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[0332] The system's hardware includes internet-connected devices (e.g., smartphones, tablets, PCs, etc.), and its software includes generative AI models (e.g., the BERT model using NLP techniques and the YOLOv5 model using computer vision techniques) and emotion engines (e.g., the OpenFace library).

[0333] When a user attempts to access a specific URL using an Internet-connected device (hereinafter referred to as a terminal), the access request is sent to a server. The server makes an HTTP request to the specified URL and retrieves the web page content. The retrieved content is in HTML format, and web scraping technology (such as Beautiful Soup or Selenium) is used to separate it into text data and image data.

[0334] Text data is analyzed using generative AI models, which leverage NLP techniques to detect harmful keywords and contexts. A specific example is the BERT model, which uses the Transformers library. Similarly, image data is analyzed using computer vision techniques, such as YOLOv5 and OpenCV, to detect inappropriate images.

[0335] To recognize the user's emotional state, the emotion engine collects data from the device's camera, microphone, and text input data. The emotion engine (e.g., the OpenFace library) analyzes the collected data and determines the user's current emotional state from their facial expressions and voice.

[0336] If the server detects inappropriate or fraudulent content based on the analysis results of the generative AI model and the emotion engine, it blocks access to the content and sends a warning page to the device. If it detects content that may cause mental stress, it combines the results of the emotion engine to block access and display a warning page. Furthermore, it suggests appropriate content (e.g., videos or music with a relaxing effect) based on the user's emotional state.

[0337] As a concrete example, in the case of access to inappropriate content, an underage user attempts to access a website containing adult content. The access request from the device is sent to the server, which retrieves and analyzes the content of the URL. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0338] An example prompt might be, "Using a generative AI model, analyze the content of the specified URL to detect inappropriate or deceptive content. Also, consider the results of the user's sentiment engine to suggest appropriate content for the user."

[0339] The above is a detailed description of a specific embodiment of the present invention.

[0340] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0341] Step 1:

[0342] When a user attempts to access a specific URL, the device sends the access request to the server. The input is the URL entered by the user into the device, and the output is an HTTP request to the server. Specifically, a web browser or a dedicated application sends the URL to the server.

[0343] Step 2:

[0344] Based on the received URL access request, the server makes an HTTP request to the specified URL and retrieves the web page content. The input is the URL sent from the terminal, and the output is the web page content in HTML format. Specifically, the server communicates with the web server using the HTTP protocol and downloads the web page.

[0345] Step 3:

[0346] The server separates the retrieved HTML content into text data and image data. This process uses web scraping technology (e.g., Beautiful Soup or Selenium). The input is the web page content in HTML format, and the output is text data and image data. Specifically, the HTML parser analyzes the HTML tags and extracts the text and images.

[0347] Step 4:

[0348] The server analyzes the text data using a generative AI model. Here, natural language processing technology (e.g., the BERT model) is used to analyze the text content and detect harmful keywords and context. The input is the separated text data, and the output is an evaluation score for the text as the analysis result. Specifically, the text data is input into the generative model, and natural language processing is performed to evaluate the harmfulness.

[0349] Step 5:

[0350] The server analyzes image data using a generative AI model. Here, computer vision techniques (e.g., YOLOv5 or OpenCV) are used to detect inappropriate images. The input is the separated image data, and the output is an evaluation score for the image as the analysis result. Specifically, the image data is input into the generative model, and a computer vision algorithm is applied to evaluate its harmfulness.

[0351] Step 6:

[0352] The device collects camera, microphone, and text input data to recognize the user's emotional state. The input is the user's facial expression, voice, and text input data, and the output is to send this data to the server. Specific operations include the process in which the device's sensors capture data and send it to the server.

[0353] Step 7:

[0354] The server uses an emotion engine to analyze the collected data and determine the user's emotional state. Here, the analysis is performed using emotion recognition technology (e.g., the OpenFace library). The input is facial expressions, voice, and text data sent from the device, and the output is an evaluation result of the user's emotional state. Specifically, the emotion engine processes the data and analyzes the emotional state in real time.

[0355] Step 8:

[0356] If the server detects inappropriate or deceptive content based on the analysis results of the generative AI model and emotion engine, it blocks access to that content. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is an instruction to block access and the generation of a warning page. Specifically, the server executes the access control logic and generates a warning page if necessary.

[0357] Step 9:

[0358] If the server detects content that may cause mental stress, it combines the results of the emotion engine to block access and send a warning page to the terminal. At the same time, it suggests appropriate content based on the user's emotional state. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is a list of suggested content. Specifically, the server runs the content suggestion algorithm to generate content appropriate for the user.

[0359] Step 10:

[0360] The terminal receives the analysis results, warning pages, and suggested content from the server and displays them to the user. The input is the warning page and suggested content sent from the server, and the output is the screen displayed to the user. Specifically, the terminal updates the screen according to instructions from the server and provides the necessary information to the user.

[0361] (Application example 2)

[0362] 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."

[0363] With the spread of the Internet, users have more opportunities to access various websites, many of which contain inappropriate or fraudulent content. Access to content that may cause mental stress is also increasing. Encountering such content can harm users' mental health, so a system that automatically filters this content and protects users' health is needed.

[0364] 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 a website access request from an Internet-connected device, means for acquiring content from a specified URL, means for analyzing the text and images of the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for recognizing emotions from the user's facial expression or voice, means for determining whether the recognized emotion is a stress state, and means for suggesting appropriate content according to the stress state. This not only protects users from inappropriate or fraudulent content when browsing websites, but also makes it possible to provide appropriate content to reduce mental stress.

[0365] "Internet-connected device" refers to any electronic device that can connect to the Internet.

[0366] "Website Access Request" refers to a request by a user to access a particular website.

[0367] "Specified URL" refers to the specific web address that the user attempted to access.

[0368] "Content" refers to information such as text data and image data displayed on a website.

[0369] A "generative model" refers to algorithms or software for generating and analyzing data based on artificial intelligence techniques.

[0370] "Means for analyzing text and images" refers to methods for evaluating and analyzing text data and image data within the Content.

[0371] "Inappropriate Content" refers to information on a website that contains material that is deemed harmful to users.

[0372] "Deceptive content" refers to false information or pages created with the intent to deceive users.

[0373] "Blocking measures" refers to methods of controlling users' access to content that is deemed inappropriate or fraudulent.

[0374] "Warning Page" refers to a notification page that informs users that certain content is inappropriate or deceptive.

[0375] "Means for recognizing emotions from a user's facial expression or voice" refers to technology that analyzes and determines a user's emotional state from their facial expression or voice.

[0376] A "stressed state" refers to a state in which the user feels mental tension or anxiety.

[0377] The "means for suggesting appropriate content" refers to a method for recommending content that has a relaxing effect based on the user's emotional state.

[0378] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[0379] System Configuration

[0380] Server Configuration

[0381] The server requires the following hardware and software:

[0382] Hardware: A server with a fast processor and large memory capacity.

[0383] software:

[0384] Generative model: OpenAI GPT-4

[0385] Computer Vision Technology: Google Vision API

[0386] Emotion engine: Affectiva SDK

[0387] Programming languages ​​and frameworks: Python, Django, Nginx

[0388] User's device

[0389] The user's device must have the following features:

[0390] Camera and microphone

[0391] Internet connection function

[0392] Dedicated applications

[0393] Program processing

[0394] Server Processing

[0395] The server first receives a URL access request sent from the user's device. It then makes an HTTP request to the specified URL to retrieve the webpage content. The retrieved content is separated into text data and image data and analyzed using a generative model (GPT-4) and computer vision technology (Google Vision API). If inappropriate or fraudulent content is detected, the server blocks access to the content and sends a warning page to the user's device.

[0396] The server also uses an emotion engine (Affectiva SDK) to recognize the user's emotions in real time. It determines the user's emotional state based on facial expressions, voice, and text input, and if it determines that the user is particularly stressed, it suggests content that will have a relaxing effect.

[0397] Terminal handling

[0398] The user's device sends a request to the server when they attempt to access a specific URL, and displays appropriate content or a warning page based on the analysis results.The camera and microphone are used to provide data to the emotion engine, which monitors the user's emotional state in real time.

[0399] Specific examples

[0400] Example 1: Accessing inappropriate content

[0401] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0402] Example 2: Visiting a fraudulent website

[0403] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and displays a warning page on the device. In this way, access to the fraudulent website is prevented.

[0404] Example 3: Visiting a stressful news site

[0405] A user may attempt to access a news site, but the content of that site may cause mental stress. The server analyzes the text and image content and determines that it may cause mental stress. Furthermore, the emotion engine monitors the user's emotional state. If it determines that the user is in a stressful state, the server blocks the access and sends a warning page to the user's device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[0406] Prompt Sentence Examples

[0407] "Emotion-aware filtering app. A user is about to visit a new news site. Check if the site contains harmful content and display a warning page if necessary. Also, if the user is under stress, suggest relaxation music."

[0408] The above is a specific description of the embodiment of the present invention. This system protects users from inappropriate or fraudulent content and reduces mental stress.

[0409] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0410] Step 1:

[0411] When a user tries to access a specific URL, the device generates a website access request and sends it to the server. The input is the URL specified by the user, and the output is the access request sent to the server.

[0412] Step 2:

[0413] The server receives a URL access request sent from the user's terminal. The input is the access request, and the output is a confirmation of the request. Based on this request, the server makes an HTTP request to the specified URL.

[0414] Step 3:

[0415] The server uses an HTTP request to retrieve the content of a specified URL. The input is the specified URL, and the output is the retrieved web page content (text data and image data).

[0416] Step 4:

[0417] The server separates the retrieved content into text data and image data. The input of this step is the retrieved web page content, and the output is the separated text data and image data. The server analyzes the text content using a generative model (GPT-4) and analyzes the image content using computer vision technology (Google Vision API).

[0418] Step 5:

[0419] The server determines whether content is inappropriate or fraudulent based on the analysis results of the generative model. The input is the analyzed text data and image data, and the output is the judgment result. If inappropriate or fraudulent content is detected, the server proceeds to the next step based on the judgment result.

[0420] Step 6:

[0421] If the server determines that the content is inappropriate or fraudulent, it blocks access to it and sends a warning page to the terminal. The input of this step is the result of the determination, and the output is a block instruction and the sending of a warning page.

[0422] Step 7:

[0423] The terminal receives the warning page sent from the server and displays it to the user. The input is the warning page, and the output is the display of a warning message to the user. This warning prevents the user from accessing inappropriate content or fraudulent sites.

[0424] Step 8:

[0425] The server uses an emotion engine (Affectiva SDK) to recognize emotions from the user's facial expressions or voice in real time. The input is facial expression data or voice data sent from the device, and the output is the user's emotional state.

[0426] Step 9:

[0427] The server determines whether the recognized emotion is a stress state. The input is the analysis result of the emotion engine, and the output is the stress state determination result.

[0428] Step 10:

[0429] If the server determines that the user is in a stressful state, it proposes appropriate content. The input to this step is the result of the stress state determination, and the output is a recommendation of content that has a relaxing effect. This appropriate content proposal is sent to the user's device.

[0430] The above is the specific processing flow of the system that realizes this application example. This system protects users from inappropriate or fraudulent content when browsing websites, thereby reducing mental stress.

[0431] 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.

[0432] 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.

[0433] 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.

[0434] [Second embodiment]

[0435] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0436] 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.

[0437] 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).

[0438] 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.

[0439] 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.

[0440] 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).

[0441] 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.

[0442] 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.

[0443] 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.

[0444] 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.

[0445] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0446] 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."

[0447] The present invention is a filtering system that receives website access requests from internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress.

[0448] Program processing

[0449] Server Processing

[0450] The server receives a URL access request sent from the user's device. After receiving the request, the server makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by the generative model.

[0451] The generative model uses natural language processing (NLP) technology to analyze the text content of a page and evaluate whether it contains harmful keywords or expressions, while also using computer vision technology to evaluate image data for inappropriate images.

[0452] If the analysis detects inappropriate or fraudulent content, the server blocks access to the content and sends a warning page to the user's device explaining the reason for the block and the importance of safe web browsing.

[0453] During the analysis, information that may cause psychological stress is also detected, and if such information is found, the server blocks access to it as well and sends a warning page.

[0454] Terminal handling

[0455] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0456] User Interface

[0457] Users can surf the internet using a regular browser or a dedicated application. If the website they are trying to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous content. It also provides advance warnings about information that may cause mental stress, protecting users' mental health.

[0458] Specific examples

[0459] Example 1: Accessing inappropriate content

[0460] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0461] Example 2: Visiting a fraudulent website

[0462] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[0463] Example 3: Visiting a stressful news site

[0464] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. The server blocks the access and sends a warning page. The terminal displays the warning page to the user, and their mental health is protected.

[0465] The above is a specific embodiment of the present invention.

[0466] The processing flow will be explained below.

[0467] Step 1:

[0468] A user attempts to access a specific URL using a browser or application.

[0469] Step 2:

[0470] The device sends an access request to the server for the specified URL.

[0471] Step 3:

[0472] The server logs URL access requests received from the device.

[0473] Step 4:

[0474] The server makes an HTTP request to the specified URL to retrieve the web page content.

[0475] Step 5:

[0476] The server separates the content of the acquired web page into text data and image data.

[0477] Step 6:

[0478] The server analyzes the text data using the generative model.

[0479] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[0480] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[0481] Step 7:

[0482] The server analyzes the image data using the generative model.

[0483] Computer vision techniques are used to evaluate the content of the image.

[0484] Detect inappropriate images and specific patterns.

[0485] Step 8:

[0486] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[0487] If there is a problem, the server will block access to the URL.

[0488] Step 9:

[0489] If the server blocks inappropriate or deceptive content, it generates a warning page.

[0490] The warning page will include information about why your access was blocked and the importance of safe internet use.

[0491] Step 10:

[0492] The server sends a warning page to the device.

[0493] Step 11:

[0494] The terminal displays the received warning page to the user.

[0495] Step 12:

[0496] The server analyzes content that may cause mental stress.

[0497] Detects trigger words and images to determine if they contain stress-related information.

[0498] Step 13:

[0499] If the server detects stress-related content, it will also block access to it.

[0500] In this case, a warning page is also generated and sent to the terminal.

[0501] Step 14:

[0502] The device will again display a warning page to the user to protect their mental health.

[0503] Through the above steps, the filtering system of the present invention provides an environment in which users can use the Internet safely and securely.

[0504] Example 1

[0505] 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."

[0506] The Internet contains inappropriate and fraudulent content that is harmful to many users, including minors and the elderly, and can also cause psychological stress. Therefore, a filtering system is needed to ensure that users can use the Internet safely and comfortably. However, existing filtering systems lack the ability to analyze text and images, making them unable to effectively block harmful content.

[0507] 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.

[0508] In this invention, the server includes means for receiving a website access request sent from a terminal, means for acquiring the content of the requested URL, means for separating the acquired content into text data and image data, means for analyzing the text data using a generative AI model to detect harmful keywords and expressions, means for analyzing the image data using computer vision technology to detect inappropriate images, means for determining inappropriate or fraudulent content based on the analysis results of the generative AI model, and means for blocking inappropriate or fraudulent content and sending a warning page to the terminal. This makes it possible to effectively block inappropriate or fraudulent content on the Internet and provide an environment in which users can use the Internet safely and comfortably.

[0509] A "terminal" is a device such as a computer or smartphone that allows a user to connect to the Internet.

[0510] A "server" is a computer system that provides information and performs processing on the Internet.

[0511] A "website access request" is a request that a user sends from a terminal to a server to access a specific website.

[0512] A "URL" is an address on the Internet that points to a specific web page.

[0513] "Content" is a general term for information displayed on a web page, and includes text data and image data.

[0514] A "generative AI model" is a statistical model for generating and analyzing data using artificial intelligence techniques.

[0515] "Natural language processing technology" is a technology that enables computers to understand and process human language.

[0516] "Computer vision technology" is a technology that allows computers to analyze and recognize images and videos.

[0517] A "warning page" is a web page that blocks access to inappropriate or fraudulent content and notifies users of this.

[0518] "Harmful keywords and expressions" are words or phrases that may threaten a user's mental health or safe use of the Internet.

[0519] An "inappropriate image" is an image that is perceived by a user as psychologically harmful or unpleasant.

[0520] "Fraudulent content" is information intended to deceive users and cause them financial harm.

[0521] "Mental stress" is the emotional strain that causes a user's mental health to deteriorate.

[0522] There is a lot of inappropriate or fraudulent content on the Internet that can threaten users' safety and mental health. The present invention provides an advanced filtering system to protect users from such harmful content.

[0523] Server hardware and software configuration

[0524] The server uses a high-performance computer system and has the following software configuration:

[0525] 1. Web server software: Use Apache HTTP Server or Nginx.

[0526] 2. Python programming environment: Uses the Python language and related libraries to perform natural language processing, image analysis, etc.

[0527] 3. Generative AI models: Use Hugging Face's Transformers model or OpenAI's GPT-3.

[0528] 4. Natural Language Processing techniques: Use pre-trained models such as BERT and RoBERTa.

[0529] 5. Computer vision technology: Image analysis is performed using TensorFlow and Detectron2.

[0530] System Operation Overview

[0531] The system of the present invention receives a website access request sent from a user's device and retrieves the content of the specified URL based on the request. The retrieved content is separated into text data and image data, each of which is analyzed using a generative AI model. If inappropriate or fraudulent content is detected based on the analysis results, the server blocks access to the content and sends a warning page to the user's device. In addition, content that may cause mental stress may also be detected, and access to such information is also restricted.

[0532] Specific examples

[0533] Example 1: Accessing inappropriate content

[0534] When a minor user attempts to access a website containing adult content using a browser, an access request is sent from the device to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is inappropriate, blocks the access. A warning page is displayed on the device, preventing access to the adult content.

[0535] Example 2: Visiting a fraudulent website

[0536] An elderly person clicks on a link in an email and attempts to access a financial fraud website. In this case, the device sends an access request to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is fraudulent, blocks the access. A warning page is displayed on the device, preventing access to the fraudulent website.

[0537] Example 3: Visiting a stressful news site

[0538] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server will analyze the text and image content and determine that it may cause mental stress. The server will block the access and send a warning page, which will be displayed on the device, protecting the user's mental health.

[0539] Prompt Sentence Examples

[0540] "Please explain in natural language how you analyze the URL of the website a user is attempting to access, determine if the URL contains inappropriate or deceptive content, and block access if necessary. Also, please specify the names of any specific hardware or software used."

[0541] As a result, by using the system of the present invention, users are protected from harmful content on the Internet, enabling safe and comfortable web browsing.

[0542] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0543] Step 1:

[0544] The server receives a website access request sent from the user's terminal.

[0545] Input: An HTTP GET request containing the URL of the website the user wants to access.

[0546] What happens: The server receives this request using web server software (e.g., Apache HTTP Server or Nginx).

[0547] Output: The requested URL.

[0548] Step 2:

[0549] The server makes an HTTP request to the received URL to retrieve the web page content.

[0550] Input: The requested URL.

[0551] What happens: The server uses Python's requests library to download HTML data from the specified URL. It then executes requests.get(url) to get the HTML content of the web page.

[0552] Output: The retrieved HTML data.

[0553] Step 3:

[0554] The server separates the acquired content into text data and image data.

[0555] Input: The retrieved HTML data.

[0556] What happens: The server uses the BeautifulSoup library to parse the HTML data and extract text and image data. It runs soup = BeautifulSoup(html_data, 'html.parser') and separates the data using functions like text_data = soup.get_text() and image_urls = [img['src'] for img in soup.find_all('img')] .

[0557] Output: Separated text and image data.

[0558] Step 4:

[0559] The server uses a generative AI model to analyze the extracted text data and detect harmful keywords and expressions.

[0560] Input: Separated text data.

[0561] Specific operation: The server uses Hugging Face's Transformers library to analyze the data by calling a generative AI model (e.g., BERT, RoBERTa). It then executes nlp_pipeline = pipeline('sentiment-analysis') and runs result = nlp_pipeline(text_data) to detect harmful keywords.

[0562] Output: A list of harmful keywords and expressions contained in the text data.

[0563] Step 5:

[0564] The server uses computer vision technology to analyze the extracted image data and detect inappropriate images.

[0565] Input: Segmented image data.

[0566] Specific operation: The server performs image analysis using TensorFlow and Detectron2. It downloads an image, inputs it into the model, executes detectron2_predictor = DefaultPredictor(cfg) and outputs = detectron2_predictor(image), and evaluates the image.

[0567] Output: A list of inappropriate images.

[0568] Step 6:

[0569] The server uses the analysis to determine whether the URL contains inappropriate or deceptive content.

[0570] Input: Analysis results of text data and image data.

[0571] What it does: The server evaluates the results of the generative AI model and determines whether they contain inappropriate or deceptive content, for example, by using conditional branching such as if 'toxic' in text_analysis_result or 'unsafe' in image_analysis_result.

[0572] Output: Access blocking decision result (whether to block or not).

[0573] Step 7:

[0574] If the server determines that the content is inappropriate or fraudulent, it will send a warning page to the user's device.

[0575] Input: Block decision result.

[0576] Specific operation: The server generates a warning page using HTML and CSS and sends it to the terminal as an HTTP response. response_html = " <h1> Warning< / h1> This site is blocked. " and send it as return response_html.

[0577] Output: A warning page that is displayed on the user's device.

[0578] Step 8:

[0579] The terminal displays a warning page to the user.

[0580] Input: The HTML of the warning page received from the server.

[0581] Specific operation: The user's device renders the received HTML in the browser and displays the warning content.

[0582] Output: The warning page that the user sees.

[0583] (Application example 1)

[0584] 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."

[0585] As the Internet becomes more widespread, users are increasingly at risk of accessing inappropriate content or fraudulent websites. Furthermore, there is a large amount of information that can potentially cause psychological stress, making it difficult for users to use the Internet safely and comfortably. Therefore, there is a need for a system that can detect dangerous content in real time and protect users.

[0586] 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.

[0587] In this invention, the server includes means for receiving a website access request from an Internet connection device, means for acquiring data from a specified URL, means for analyzing text and images in the data using a generative model, means for determining whether the data is inappropriate or fraudulent based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent data and displaying a warning page, and means for scanning data in real time and monitoring user Internet usage, thereby enabling users to use the Internet safely and comfortably.

[0588] An "Internet connection device" is a device for connecting to the Internet and capable of sending access requests to websites.

[0589] A "website access request" is a request sent from an Internet-connected device when a user attempts to access a specific website on the Internet.

[0590] A "specified URL" refers to the address of a particular web page that a user wishes to access.

[0591] "Data" means the content on the Website, including text and image data.

[0592] A "generative model" is a model based on artificial intelligence techniques used to analyze text and image data.

[0593] "Analysis results" refers to the analysis results of data obtained by a generative model, which are used to determine whether the data is inappropriate or fraudulent.

[0594] "Inappropriate Content" means content that is deemed harmful or offensive to users.

[0595] "Deceptive data" is content determined to be intended to deceive or mislead users.

[0596] A "warning page" is a page that is displayed when inappropriate or fraudulent data is detected to notify the user and block access.

[0597] "Real-time scanning" refers to the process of analyzing data as soon as it is acquired.

[0598] "Internet usage monitoring measures" means measures that continuously track and analyze users' Internet activity to detect inappropriate or fraudulent data.

[0599] Server Processing

[0600] The server receives a website access request sent from the user's Internet connection device. Based on the received access request, the server makes an HTTP request to the specified URL to obtain the web page data. This data is separated into text data and image data and analyzed using a generative model.

[0601] Generative models use natural language processing (NLP) techniques to analyze the text content of web pages, for example, to assess whether the text contains harmful keywords or inappropriate language, and simultaneously use computer vision techniques to analyze image data and assess whether it contains inappropriate images.

[0602] If the analysis detects inappropriate or fraudulent data, the server blocks access to it and sends a warning page to the user's internet connection, explaining the reason for the block and the importance of safe web browsing.

[0603] Additionally, the server detects any information that may cause psychological stress, and if such information is included, the server will similarly block access and send a warning page.

[0604] Terminal handling

[0605] When a user attempts to access a specific URL, the user's Internet connection device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate data or a warning page and displays it to the user. If access to inappropriate data or a fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0606] User Interface

[0607] Users can use a regular browser or dedicated application to access the Internet. If the website they are about to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous data. Advance warnings are also provided for information that may cause mental stress, protecting users' mental health.

[0608] Hardware and software used

[0609] The system uses the following hardware and software:

[0610] Hardware: Internet-connected devices (smartphones, tablets, PCs, etc.)

[0611] Software: Server-side HTTP request processing frameworks (e.g., Python's requests library), natural language processing models (e.g., Hugging Face's transformers library), computer vision models (e.g., OpenCV)

[0612] Specific examples

[0613] If a user attempts to access a news site, but it is determined that the content of the news site may cause mental stress, the following prompt sentence is presented to the generative model for analysis:

[0614] "Does this news article contain content that may cause emotional stress to users?"

[0615] The generative model generates an answer to this prompt and decides whether to block access based on the answer, allowing users to always use the Internet safely and with peace of mind.

[0616] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0617] Step 1:

[0618] Receives website access requests from internet-connected devices

[0619] The server receives a website access request sent from the user's Internet connection device. As input, this request contains a URL. The server takes this URL and proceeds to the next step.

[0620] Step 2:

[0621] Gets the data of the specified URL

[0622] The server makes an HTTP request to the received URL to retrieve the web page content of that URL. The URL is given as input, and the text and image data of the web page are obtained as output. The server separates these data for the next step.

[0623] Step 3:

[0624] Analyzing data using generative models

[0625] The server analyzes the acquired text data and image data using a generative model. The text data and image data are given as input, and the analysis results are obtained as output. Specifically, the text data is analyzed using a natural language processing (NLP) model, and the image data is analyzed using a computer vision model.

[0626] Step 4:

[0627] Identifying inappropriate or fraudulent data

[0628] The server determines whether data is inappropriate or fraudulent based on the analysis results of the generative model. The analysis results are given as input, and the output is a judgment result on whether the data is inappropriate or not. Specifically, the NLP model detects harmful keywords and expressions in text, and the computer vision model detects inappropriate content in images.

[0629] Step 5:

[0630] Blocking data and displaying a warning page

[0631] If the server detects inappropriate or fraudulent data, it blocks access to that data. The input is the verdict, and the output is a blocked access flag and a warning page that explains the inappropriate content and the importance of safe web browsing.

[0632] Step 6:

[0633] Sending and displaying a warning page

[0634] The server sends the generated warning page to the user's Internet connection device. The warning page is given as input, and the warning page is displayed on the user's terminal as output. The terminal receives this warning page and displays it to the user.

[0635] Step 7:

[0636] Detecting and limiting data that may cause mental stress

[0637] During the analysis process, the server detects information that may cause mental stress. Text data and image data are input, and a judgment result regarding mental stress is obtained as output. If the information is determined to cause mental stress, it similarly blocks access and sends a warning page.

[0638] Specific actions

[0639] The natural language processing model uses pre-trained AI models (e.g., Hugging Face's transformers library) to analyze text data and generate answers to prompts such as, "Does this news article contain content that would cause mental stress to the user?"

[0640] Computer vision technology uses OpenCV and other image recognition models to scan image data in real time to detect whether it contains inappropriate images.

[0641] These specific processes enable users to use the Internet safely and comfortably.

[0642] 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.

[0643] The present invention provides a filtering system that receives a website access request from an internet-connected device, retrieves the content of the specified URL, analyzes the text and images of the content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health state in real time and suggest appropriate content according to the user's specific emotional state.

[0644] Program processing

[0645] Server Processing

[0646] The server receives a URL access request sent from the user's device. The server then makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by a generative model. The generative model uses natural language processing (NLP) technology to analyze the text content and detect harmful keywords and context. At the same time, computer vision technology is used to detect inappropriate images in the image data.

[0647] Emotion engine processing

[0648] The server recognizes the user's emotions using an emotion engine. The emotion engine analyzes the user's device's camera, microphone, and text input data to determine the user's emotional state from their facial expressions, voice, and text input. Based on the results, the server grasps the user's current emotional state in real time.

[0649] Filtering and Content Suggestions

[0650] If inappropriate or fraudulent content is detected based on the analysis results of the generative model, the server blocks access to the content and sends a warning page to the user's device. On the other hand, if content that may cause mental stress is detected, the server blocks access and displays a warning page in combination with the results of the emotion engine.

[0651] Furthermore, the server can suggest content that takes into account the user's mental health based on the results of the emotion engine. For example, if a user is under stress, it can suggest relaxing content, soothing music, or videos for relaxation.

[0652] Terminal handling

[0653] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0654] The device also uses a camera and microphone to provide data to the emotion engine, monitoring the user's emotional state in real time.

[0655] User Interface

[0656] Users surf the Internet using a regular browser or a dedicated application. If the website they are attempting to access is deemed inappropriate or fraudulent, a warning page is displayed, protecting the user from dangerous content. The system also provides advance warnings about information that may cause mental stress, maintaining the user's mental health. Furthermore, an emotion engine monitors the user's emotional state in real time and suggests relaxing content as needed.

[0657] Specific examples

[0658] Example 1: Accessing inappropriate content

[0659] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0660] Example 2: Visiting a fraudulent website

[0661] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[0662] Example 3: Visiting a stressful news site

[0663] If a user attempts to access a news site, but the content of the site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. Furthermore, an emotion engine monitors the user's emotional state. If the server determines that the user is in a stressful state, the server blocks the access and sends a warning page to the device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[0664] The above is a specific embodiment of the present invention.

[0665] The processing flow will be explained below.

[0666] Step 1:

[0667] A user attempts to access a specific URL using a browser or application.

[0668] Step 2:

[0669] The device sends an access request to the server for the specified URL.

[0670] Step 3:

[0671] The server logs URL access requests received from the device.

[0672] Step 4:

[0673] The server makes an HTTP request to the specified URL to retrieve the web page content.

[0674] Step 5:

[0675] The server separates the content of the acquired web page into text data and image data.

[0676] Step 6:

[0677] The server analyzes the text data using the generative model.

[0678] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[0679] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[0680] Step 7:

[0681] The server analyzes the image data using the generative model.

[0682] It uses computer vision techniques to evaluate the content of images and detect inappropriate images or specific patterns.

[0683] Step 8:

[0684] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[0685] If there is a problem, the server will block access to the URL.

[0686] Step 9:

[0687] If the server blocks inappropriate or deceptive content, it generates a warning page.

[0688] The warning page will include information about why your access was blocked and the importance of safe internet use.

[0689] Step 10:

[0690] The server sends a warning page to the device.

[0691] Step 11:

[0692] The terminal displays the received warning page to the user.

[0693] Step 12:

[0694] The server analyzes content that may cause mental stress.

[0695] Detects trigger words and images to determine if they contain stress-related information.

[0696] Step 13:

[0697] If the server detects stress-related content, it will also block access to it.

[0698] In this case, a warning page is also generated and sent to the terminal.

[0699] Step 14:

[0700] The device will again display a warning page to the user to protect their mental health.

[0701] Emotion engine processing

[0702] Step 15:

[0703] The server uses an emotion engine to recognize the user's emotional state.

[0704] It collects and analyzes data from the device's camera, microphone, and text input, and determines the user's emotional state from their facial expressions, voice, and text input.

[0705] Step 16:

[0706] If the emotion engine determines that the user is stressed, the server uses this information to make decisions about suggesting more appropriate content.

[0707] Step 17:

[0708] The server selects content to reduce stress based on the results of the emotion engine.

[0709] It includes relaxing videos, music, articles, etc.

[0710] Step 18:

[0711] The server generates a page for proposing the selected stress reduction content to the user.

[0712] Step 19:

[0713] The server generates a proposal page and sends it to the terminal.

[0714] Step 20:

[0715] The terminal displays the received proposal page to the user.

[0716] It is expected that users will view the suggested content and reduce their mental stress.

[0717] The above is the flow of processing in a specific embodiment of the present invention.

[0718] Example 2

[0719] 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."

[0720] There is a lot of inappropriate or fraudulent content on the Internet, and accessing it poses a significant risk to users. Furthermore, involuntary access to content that causes mental stress can have a negative impact on users' mental health. Conventional filtering systems have had difficulty completely eliminating such risks. The present invention aims to solve these problems, protect users from harmful content on the Internet, and maintain their mental health.

[0721] 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.

[0722] In this invention, the server includes means for receiving a website access request from an Internet-connected device, means for retrieving content from a specified URL, means for analyzing text and images in the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for collecting data to recognize a user's emotional state, and means for analyzing the collected data using an emotion engine to determine the user's emotional state. This protects users from inappropriate or fraudulent content and from content that may cause mental stress, thereby enabling them to maintain their mental health.

[0723] An "Internet-connected device" is a device that has the ability to connect to the Internet and is capable of web browsing and data communication. Examples include smartphones, tablets, and personal computers.

[0724] A "website access request" is an HTTP request sent when a user attempts to access a particular web page using an Internet-connected device.

[0725] A "specified URL" is the address of a particular web page that a user wishes to access, written in Uniform Resource Locator (URL) format.

[0726] A "generative model" is an AI model that is trained on large datasets and used to analyze text and images, allowing it to detect specific patterns and anomalies.

[0727] "Text analytics" is the process of analyzing text data using generative models, specifically using natural language processing techniques to understand and classify the content of the text.

[0728] "Image analysis" is the process of analyzing image data using generative models. It refers to the use of computer vision techniques to detect features and anomalies in images.

[0729] "Inappropriate Content" refers to content that is inappropriate for minors, or that contains violent, sexual, or discriminatory material, and that may have a harmful effect on users.

[0730] "Fraudulent content" refers to web content created with the intention of deceiving users, and which may result in financial loss or the leakage of personal information.

[0731] A "warning page" is a page that is displayed when the web page that a user is attempting to access is determined to be inappropriate or fraudulent, and serves to warn the user.

[0732] An "emotion engine" is a device or software that analyzes a user's emotional state and determines the emotional state based on the user's facial expressions, voice, text input, etc.

[0733] "Mental stress" refers to the psychological burden or state of tension caused by external stimuli or information.

[0734] "Content suggestion" is the process of recommending suitable content, such as videos or music with a relaxing effect, taking into account the user's current emotional state and mental health.

[0735] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[0736] The system's hardware includes internet-connected devices (e.g., smartphones, tablets, PCs, etc.), and its software includes generative AI models (e.g., the BERT model using NLP techniques and the YOLOv5 model using computer vision techniques) and emotion engines (e.g., the OpenFace library).

[0737] When a user attempts to access a specific URL using an Internet-connected device (hereinafter referred to as a terminal), the access request is sent to a server. The server makes an HTTP request to the specified URL and retrieves the web page content. The retrieved content is in HTML format, and web scraping technology (such as Beautiful Soup or Selenium) is used to separate it into text data and image data.

[0738] Text data is analyzed using generative AI models, which leverage NLP techniques to detect harmful keywords and contexts. A specific example is the BERT model, which uses the Transformers library. Similarly, image data is analyzed using computer vision techniques, such as YOLOv5 and OpenCV, to detect inappropriate images.

[0739] To recognize the user's emotional state, the emotion engine collects data from the device's camera, microphone, and text input data. The emotion engine (e.g., the OpenFace library) analyzes the collected data and determines the user's current emotional state from their facial expressions and voice.

[0740] If the server detects inappropriate or fraudulent content based on the analysis results of the generative AI model and the emotion engine, it blocks access to the content and sends a warning page to the device. If it detects content that may cause mental stress, it combines the results of the emotion engine to block access and display a warning page. Furthermore, it suggests appropriate content (e.g., videos or music with a relaxing effect) based on the user's emotional state.

[0741] As a concrete example, in the case of access to inappropriate content, an underage user attempts to access a website containing adult content. The access request from the device is sent to the server, which retrieves and analyzes the content of the URL. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0742] An example prompt might be, "Using a generative AI model, analyze the content of the specified URL to detect inappropriate or deceptive content. Also, consider the results of the user's sentiment engine to suggest appropriate content for the user."

[0743] The above is a detailed description of a specific embodiment of the present invention.

[0744] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0745] Step 1:

[0746] When a user attempts to access a specific URL, the device sends the access request to the server. The input is the URL entered by the user into the device, and the output is an HTTP request to the server. Specifically, a web browser or a dedicated application sends the URL to the server.

[0747] Step 2:

[0748] Based on the received URL access request, the server makes an HTTP request to the specified URL and retrieves the web page content. The input is the URL sent from the terminal, and the output is the web page content in HTML format. Specifically, the server communicates with the web server using the HTTP protocol and downloads the web page.

[0749] Step 3:

[0750] The server separates the retrieved HTML content into text data and image data. This process uses web scraping technology (e.g., Beautiful Soup or Selenium). The input is the web page content in HTML format, and the output is text data and image data. Specifically, the HTML parser analyzes the HTML tags and extracts the text and images.

[0751] Step 4:

[0752] The server analyzes the text data using a generative AI model. Here, natural language processing technology (e.g., the BERT model) is used to analyze the text content and detect harmful keywords and context. The input is the separated text data, and the output is an evaluation score for the text as the analysis result. Specifically, the text data is input into the generative model, and natural language processing is performed to evaluate the harmfulness.

[0753] Step 5:

[0754] The server analyzes image data using a generative AI model. Here, computer vision techniques (e.g., YOLOv5 or OpenCV) are used to detect inappropriate images. The input is the separated image data, and the output is an evaluation score for the image as the analysis result. Specifically, the image data is input into the generative model, and a computer vision algorithm is applied to evaluate its harmfulness.

[0755] Step 6:

[0756] The device collects camera, microphone, and text input data to recognize the user's emotional state. The input is the user's facial expression, voice, and text input data, and the output is to send this data to the server. Specific operations include the process in which the device's sensors capture data and send it to the server.

[0757] Step 7:

[0758] The server uses an emotion engine to analyze the collected data and determine the user's emotional state. Here, the analysis is performed using emotion recognition technology (e.g., the OpenFace library). The input is facial expressions, voice, and text data sent from the device, and the output is an evaluation result of the user's emotional state. Specifically, the emotion engine processes the data and analyzes the emotional state in real time.

[0759] Step 8:

[0760] If the server detects inappropriate or deceptive content based on the analysis results of the generative AI model and emotion engine, it blocks access to that content. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is an instruction to block access and the generation of a warning page. Specifically, the server executes the access control logic and generates a warning page if necessary.

[0761] Step 9:

[0762] If the server detects content that may cause mental stress, it combines the results of the emotion engine to block access and send a warning page to the terminal. At the same time, it suggests appropriate content based on the user's emotional state. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is a list of suggested content. Specifically, the server runs the content suggestion algorithm to generate content appropriate for the user.

[0763] Step 10:

[0764] The terminal receives the analysis results, warning pages, and suggested content from the server and displays them to the user. The input is the warning page and suggested content sent from the server, and the output is the screen displayed to the user. Specifically, the terminal updates the screen according to instructions from the server and provides the necessary information to the user.

[0765] (Application example 2)

[0766] 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."

[0767] With the spread of the Internet, users have more opportunities to access various websites, many of which contain inappropriate or fraudulent content. Access to content that may cause mental stress is also increasing. Encountering such content can harm users' mental health, so a system that automatically filters this content and protects users' health is needed.

[0768] 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 a website access request from an Internet-connected device, means for acquiring content from a specified URL, means for analyzing the text and images of the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for recognizing emotions from the user's facial expression or voice, means for determining whether the recognized emotion is a stress state, and means for suggesting appropriate content according to the stress state. This not only protects users from inappropriate or fraudulent content when browsing websites, but also makes it possible to provide appropriate content to reduce mental stress.

[0769] "Internet-connected device" refers to any electronic device that can connect to the Internet.

[0770] "Website Access Request" refers to a request by a user to access a particular website.

[0771] "Specified URL" refers to the specific web address that the user attempted to access.

[0772] "Content" refers to information such as text data and image data displayed on a website.

[0773] A "generative model" refers to algorithms or software for generating and analyzing data based on artificial intelligence techniques.

[0774] "Means for analyzing text and images" refers to methods for evaluating and analyzing text data and image data within the Content.

[0775] "Inappropriate Content" refers to information on a website that contains material that is deemed harmful to users.

[0776] "Deceptive content" refers to false information or pages created with the intent to deceive users.

[0777] "Blocking measures" refers to methods of controlling users' access to content that is deemed inappropriate or fraudulent.

[0778] "Warning Page" refers to a notification page that informs users that certain content is inappropriate or deceptive.

[0779] "Means for recognizing emotions from a user's facial expression or voice" refers to technology that analyzes and determines a user's emotional state from their facial expression or voice.

[0780] A "stressed state" refers to a state in which the user feels mental tension or anxiety.

[0781] The "means for suggesting appropriate content" refers to a method for recommending content that has a relaxing effect based on the user's emotional state.

[0782] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[0783] System Configuration

[0784] Server Configuration

[0785] The server requires the following hardware and software:

[0786] Hardware: A server with a fast processor and large memory capacity.

[0787] software:

[0788] Generative model: OpenAI GPT-4

[0789] Computer Vision Technology: Google Vision API

[0790] Emotion engine: Affectiva SDK

[0791] Programming languages ​​and frameworks: Python, Django, Nginx

[0792] User's device

[0793] The user's device must have the following features:

[0794] Camera and microphone

[0795] Internet connection function

[0796] Dedicated applications

[0797] Program processing

[0798] Server Processing

[0799] The server first receives a URL access request sent from the user's device. It then makes an HTTP request to the specified URL to retrieve the webpage content. The retrieved content is separated into text data and image data and analyzed using a generative model (GPT-4) and computer vision technology (Google Vision API). If inappropriate or fraudulent content is detected, the server blocks access to the content and sends a warning page to the user's device.

[0800] The server also uses an emotion engine (Affectiva SDK) to recognize the user's emotions in real time. It determines the user's emotional state based on facial expressions, voice, and text input, and if it determines that the user is particularly stressed, it suggests content that will have a relaxing effect.

[0801] Terminal handling

[0802] The user's device sends a request to the server when they attempt to access a specific URL, and displays appropriate content or a warning page based on the analysis results.The camera and microphone are used to provide data to the emotion engine, which monitors the user's emotional state in real time.

[0803] Specific examples

[0804] Example 1: Accessing inappropriate content

[0805] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0806] Example 2: Visiting a fraudulent website

[0807] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and displays a warning page on the device. In this way, access to the fraudulent website is prevented.

[0808] Example 3: Visiting a stressful news site

[0809] A user may attempt to access a news site, but the content of that site may cause mental stress. The server analyzes the text and image content and determines that it may cause mental stress. Furthermore, the emotion engine monitors the user's emotional state. If it determines that the user is in a stressful state, the server blocks the access and sends a warning page to the user's device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[0810] Prompt Sentence Examples

[0811] "Emotion-aware filtering app. A user is about to visit a new news site. Check if the site contains harmful content and display a warning page if necessary. Also, if the user is under stress, suggest relaxation music."

[0812] The above is a specific description of the embodiment of the present invention. This system protects users from inappropriate or fraudulent content and reduces mental stress.

[0813] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0814] Step 1:

[0815] When a user tries to access a specific URL, the device generates a website access request and sends it to the server. The input is the URL specified by the user, and the output is the access request sent to the server.

[0816] Step 2:

[0817] The server receives a URL access request sent from the user's terminal. The input is the access request, and the output is a confirmation of the request. Based on this request, the server makes an HTTP request to the specified URL.

[0818] Step 3:

[0819] The server uses an HTTP request to retrieve the content of a specified URL. The input is the specified URL, and the output is the retrieved web page content (text data and image data).

[0820] Step 4:

[0821] The server separates the retrieved content into text data and image data. The input of this step is the retrieved web page content, and the output is the separated text data and image data. The server analyzes the text content using a generative model (GPT-4) and analyzes the image content using computer vision technology (Google Vision API).

[0822] Step 5:

[0823] The server determines whether content is inappropriate or fraudulent based on the analysis results of the generative model. The input is the analyzed text data and image data, and the output is the judgment result. If inappropriate or fraudulent content is detected, the server proceeds to the next step based on the judgment result.

[0824] Step 6:

[0825] If the server determines that the content is inappropriate or fraudulent, it blocks access to it and sends a warning page to the terminal. The input of this step is the result of the determination, and the output is a block instruction and the sending of a warning page.

[0826] Step 7:

[0827] The terminal receives the warning page sent from the server and displays it to the user. The input is the warning page, and the output is the display of a warning message to the user. This warning prevents the user from accessing inappropriate content or fraudulent sites.

[0828] Step 8:

[0829] The server uses an emotion engine (Affectiva SDK) to recognize emotions from the user's facial expressions or voice in real time. The input is facial expression data or voice data sent from the device, and the output is the user's emotional state.

[0830] Step 9:

[0831] The server determines whether the recognized emotion is a stress state. The input is the analysis result of the emotion engine, and the output is the stress state determination result.

[0832] Step 10:

[0833] If the server determines that the user is in a stressful state, it proposes appropriate content. The input to this step is the result of the stress state determination, and the output is a recommendation of content that has a relaxing effect. This appropriate content proposal is sent to the user's device.

[0834] The above is the specific processing flow of the system that realizes this application example. This system protects users from inappropriate or fraudulent content when browsing websites, thereby reducing mental stress.

[0835] 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.

[0836] 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.

[0837] 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.

[0838] [Third embodiment]

[0839] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0840] 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.

[0841] 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).

[0842] 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.

[0843] 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.

[0844] 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).

[0845] 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.

[0846] 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.

[0847] 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.

[0848] 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.

[0849] 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.

[0850] 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."

[0851] The present invention is a filtering system that receives website access requests from internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress.

[0852] Program processing

[0853] Server Processing

[0854] The server receives a URL access request sent from the user's device. After receiving the request, the server makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by the generative model.

[0855] The generative model uses natural language processing (NLP) technology to analyze the text content of a page and evaluate whether it contains harmful keywords or expressions, while also using computer vision technology to evaluate image data for inappropriate images.

[0856] If the analysis detects inappropriate or fraudulent content, the server blocks access to the content and sends a warning page to the user's device explaining the reason for the block and the importance of safe web browsing.

[0857] During the analysis, information that may cause psychological stress is also detected, and if such information is found, the server blocks access to it as well and sends a warning page.

[0858] Terminal handling

[0859] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[0860] User Interface

[0861] Users can surf the internet using a regular browser or a dedicated application. If the website they are trying to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous content. It also provides advance warnings about information that may cause mental stress, protecting users' mental health.

[0862] Specific examples

[0863] Example 1: Accessing inappropriate content

[0864] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[0865] Example 2: Visiting a fraudulent website

[0866] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[0867] Example 3: Visiting a stressful news site

[0868] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. The server blocks the access and sends a warning page. The terminal displays the warning page to the user, and their mental health is protected.

[0869] The above is a specific embodiment of the present invention.

[0870] The processing flow will be explained below.

[0871] Step 1:

[0872] A user attempts to access a specific URL using a browser or application.

[0873] Step 2:

[0874] The device sends an access request to the server for the specified URL.

[0875] Step 3:

[0876] The server logs URL access requests received from the device.

[0877] Step 4:

[0878] The server makes an HTTP request to the specified URL to retrieve the web page content.

[0879] Step 5:

[0880] The server separates the content of the acquired web page into text data and image data.

[0881] Step 6:

[0882] The server analyzes the text data using the generative model.

[0883] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[0884] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[0885] Step 7:

[0886] The server analyzes the image data using the generative model.

[0887] Computer vision techniques are used to evaluate the content of the image.

[0888] Detect inappropriate images and specific patterns.

[0889] Step 8:

[0890] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[0891] If there is a problem, the server will block access to the URL.

[0892] Step 9:

[0893] If the server blocks inappropriate or deceptive content, it generates a warning page.

[0894] The warning page will include information about why your access was blocked and the importance of safe internet use.

[0895] Step 10:

[0896] The server sends a warning page to the device.

[0897] Step 11:

[0898] The terminal displays the received warning page to the user.

[0899] Step 12:

[0900] The server analyzes content that may cause mental stress.

[0901] Detects trigger words and images to determine if they contain stress-related information.

[0902] Step 13:

[0903] If the server detects stress-related content, it will also block access to it.

[0904] In this case, a warning page is also generated and sent to the terminal.

[0905] Step 14:

[0906] The device will again display a warning page to the user to protect their mental health.

[0907] Through the above steps, the filtering system of the present invention provides an environment in which users can use the Internet safely and securely.

[0908] Example 1

[0909] 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."

[0910] The Internet contains inappropriate and fraudulent content that is harmful to many users, including minors and the elderly, and can also cause psychological stress. Therefore, a filtering system is needed to ensure that users can use the Internet safely and comfortably. However, existing filtering systems lack the ability to analyze text and images, making them unable to effectively block harmful content.

[0911] 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.

[0912] In this invention, the server includes means for receiving a website access request sent from a terminal, means for acquiring the content of the requested URL, means for separating the acquired content into text data and image data, means for analyzing the text data using a generative AI model to detect harmful keywords and expressions, means for analyzing the image data using computer vision technology to detect inappropriate images, means for determining inappropriate or fraudulent content based on the analysis results of the generative AI model, and means for blocking inappropriate or fraudulent content and sending a warning page to the terminal. This makes it possible to effectively block inappropriate or fraudulent content on the Internet and provide an environment in which users can use the Internet safely and comfortably.

[0913] A "terminal" is a device such as a computer or smartphone that allows a user to connect to the Internet.

[0914] A "server" is a computer system that provides information and performs processing on the Internet.

[0915] A "website access request" is a request that a user sends from a terminal to a server to access a specific website.

[0916] A "URL" is an address on the Internet that points to a specific web page.

[0917] "Content" is a general term for information displayed on a web page, and includes text data and image data.

[0918] A "generative AI model" is a statistical model for generating and analyzing data using artificial intelligence techniques.

[0919] "Natural language processing technology" is a technology that enables computers to understand and process human language.

[0920] "Computer vision technology" is a technology that allows computers to analyze and recognize images and videos.

[0921] A "warning page" is a web page that blocks access to inappropriate or fraudulent content and notifies users of this.

[0922] "Harmful keywords and expressions" are words or phrases that may threaten a user's mental health or safe use of the Internet.

[0923] An "inappropriate image" is an image that is perceived by a user as psychologically harmful or unpleasant.

[0924] "Fraudulent content" is information intended to deceive users and cause them financial harm.

[0925] "Mental stress" is the emotional strain that causes a user's mental health to deteriorate.

[0926] There is a lot of inappropriate or fraudulent content on the Internet that can threaten users' safety and mental health. The present invention provides an advanced filtering system to protect users from such harmful content.

[0927] Server hardware and software configuration

[0928] The server uses a high-performance computer system and has the following software configuration:

[0929] 1. Web server software: Use Apache HTTP Server or Nginx.

[0930] 2. Python programming environment: Uses the Python language and related libraries to perform natural language processing, image analysis, etc.

[0931] 3. Generative AI models: Use Hugging Face's Transformers model or OpenAI's GPT-3.

[0932] 4. Natural Language Processing techniques: Use pre-trained models such as BERT and RoBERTa.

[0933] 5. Computer vision technology: Image analysis is performed using TensorFlow and Detectron2.

[0934] System Operation Overview

[0935] The system of the present invention receives a website access request sent from a user's device and retrieves the content of the specified URL based on the request. The retrieved content is separated into text data and image data, each of which is analyzed using a generative AI model. If inappropriate or fraudulent content is detected based on the analysis results, the server blocks access to the content and sends a warning page to the user's device. In addition, content that may cause mental stress may also be detected, and access to such information is also restricted.

[0936] Specific examples

[0937] Example 1: Accessing inappropriate content

[0938] When a minor user attempts to access a website containing adult content using a browser, an access request is sent from the device to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is inappropriate, blocks the access. A warning page is displayed on the device, preventing access to the adult content.

[0939] Example 2: Visiting a fraudulent website

[0940] An elderly person clicks on a link in an email and attempts to access a financial fraud website. In this case, the device sends an access request to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is fraudulent, blocks the access. A warning page is displayed on the device, preventing access to the fraudulent website.

[0941] Example 3: Visiting a stressful news site

[0942] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server will analyze the text and image content and determine that it may cause mental stress. The server will block the access and send a warning page, which will be displayed on the device, protecting the user's mental health.

[0943] Prompt Sentence Examples

[0944] "Please explain in natural language how you analyze the URL of the website a user is attempting to access, determine if the URL contains inappropriate or deceptive content, and block access if necessary. Also, please specify the names of any specific hardware or software used."

[0945] As a result, by using the system of the present invention, users are protected from harmful content on the Internet, enabling safe and comfortable web browsing.

[0946] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0947] Step 1:

[0948] The server receives a website access request sent from the user's terminal.

[0949] Input: An HTTP GET request containing the URL of the website the user wants to access.

[0950] What happens: The server receives this request using web server software (e.g., Apache HTTP Server or Nginx).

[0951] Output: The requested URL.

[0952] Step 2:

[0953] The server makes an HTTP request to the received URL to retrieve the web page content.

[0954] Input: The requested URL.

[0955] What happens: The server uses Python's requests library to download HTML data from the specified URL. It then executes requests.get(url) to get the HTML content of the web page.

[0956] Output: The retrieved HTML data.

[0957] Step 3:

[0958] The server separates the acquired content into text data and image data.

[0959] Input: The retrieved HTML data.

[0960] What happens: The server uses the BeautifulSoup library to parse the HTML data and extract text and image data. It runs soup = BeautifulSoup(html_data, 'html.parser') and separates the data using functions like text_data = soup.get_text() and image_urls = [img['src'] for img in soup.find_all('img')] .

[0961] Output: Separated text and image data.

[0962] Step 4:

[0963] The server uses a generative AI model to analyze the extracted text data and detect harmful keywords and expressions.

[0964] Input: Separated text data.

[0965] Specific operation: The server uses Hugging Face's Transformers library to analyze the data by calling a generative AI model (e.g., BERT, RoBERTa). It then executes nlp_pipeline = pipeline('sentiment-analysis') and runs result = nlp_pipeline(text_data) to detect harmful keywords.

[0966] Output: A list of harmful keywords and expressions contained in the text data.

[0967] Step 5:

[0968] The server uses computer vision technology to analyze the extracted image data and detect inappropriate images.

[0969] Input: Segmented image data.

[0970] Specific operation: The server performs image analysis using TensorFlow and Detectron2. It downloads an image, inputs it into the model, executes detectron2_predictor = DefaultPredictor(cfg) and outputs = detectron2_predictor(image), and evaluates the image.

[0971] Output: A list of inappropriate images.

[0972] Step 6:

[0973] The server uses the analysis to determine whether the URL contains inappropriate or deceptive content.

[0974] Input: Analysis results of text data and image data.

[0975] What it does: The server evaluates the results of the generative AI model and determines whether they contain inappropriate or deceptive content, for example, by using conditional branching such as if 'toxic' in text_analysis_result or 'unsafe' in image_analysis_result.

[0976] Output: Access blocking decision result (whether to block or not).

[0977] Step 7:

[0978] If the server determines that the content is inappropriate or fraudulent, it will send a warning page to the user's device.

[0979] Input: Block decision result.

[0980] Specific operation: The server generates a warning page using HTML and CSS and sends it to the terminal as an HTTP response. response_html = " <h1> Warning< / h1> This site is blocked. " and send it as return response_html.

[0981] Output: A warning page that is displayed on the user's device.

[0982] Step 8:

[0983] The terminal displays a warning page to the user.

[0984] Input: The HTML of the warning page received from the server.

[0985] Specific operation: The user's device renders the received HTML in the browser and displays the warning content.

[0986] Output: The warning page that the user sees.

[0987] (Application example 1)

[0988] 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."

[0989] As the Internet becomes more widespread, users are increasingly at risk of accessing inappropriate content or fraudulent websites. Furthermore, there is a large amount of information that can potentially cause psychological stress, making it difficult for users to use the Internet safely and comfortably. Therefore, there is a need for a system that can detect dangerous content in real time and protect users.

[0990] 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.

[0991] In this invention, the server includes means for receiving a website access request from an Internet connection device, means for acquiring data from a specified URL, means for analyzing text and images in the data using a generative model, means for determining whether the data is inappropriate or fraudulent based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent data and displaying a warning page, and means for scanning data in real time and monitoring user Internet usage, thereby enabling users to use the Internet safely and comfortably.

[0992] An "Internet connection device" is a device for connecting to the Internet and capable of sending access requests to websites.

[0993] A "website access request" is a request sent from an Internet-connected device when a user attempts to access a specific website on the Internet.

[0994] A "specified URL" refers to the address of a particular web page that a user wishes to access.

[0995] "Data" means the content on the Website, including text and image data.

[0996] A "generative model" is a model based on artificial intelligence techniques used to analyze text and image data.

[0997] "Analysis results" refers to the analysis results of data obtained by a generative model, which are used to determine whether the data is inappropriate or fraudulent.

[0998] "Inappropriate Content" means content that is deemed harmful or offensive to users.

[0999] "Deceptive data" is content determined to be intended to deceive or mislead users.

[1000] A "warning page" is a page that is displayed when inappropriate or fraudulent data is detected to notify the user and block access.

[1001] "Real-time scanning" refers to the process of analyzing data as soon as it is acquired.

[1002] "Internet usage monitoring measures" means measures that continuously track and analyze users' Internet activity to detect inappropriate or fraudulent data.

[1003] Server Processing

[1004] The server receives a website access request sent from the user's Internet connection device. Based on the received access request, the server makes an HTTP request to the specified URL to obtain the web page data. This data is separated into text data and image data and analyzed using a generative model.

[1005] Generative models use natural language processing (NLP) techniques to analyze the text content of web pages, for example, to assess whether the text contains harmful keywords or inappropriate language, and simultaneously use computer vision techniques to analyze image data and assess whether it contains inappropriate images.

[1006] If the analysis detects inappropriate or fraudulent data, the server blocks access to it and sends a warning page to the user's internet connection, explaining the reason for the block and the importance of safe web browsing.

[1007] Additionally, the server detects any information that may cause psychological stress, and if such information is included, the server will similarly block access and send a warning page.

[1008] Terminal handling

[1009] When a user attempts to access a specific URL, the user's Internet connection device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate data or a warning page and displays it to the user. If access to inappropriate data or a fraudulent website is blocked, the terminal displays a warning page to warn the user.

[1010] User Interface

[1011] Users can use a regular browser or dedicated application to access the Internet. If the website they are about to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous data. Advance warnings are also provided for information that may cause mental stress, protecting users' mental health.

[1012] Hardware and software used

[1013] The system uses the following hardware and software:

[1014] Hardware: Internet-connected devices (smartphones, tablets, PCs, etc.)

[1015] Software: Server-side HTTP request processing frameworks (e.g., Python's requests library), natural language processing models (e.g., Hugging Face's transformers library), computer vision models (e.g., OpenCV)

[1016] Specific examples

[1017] If a user attempts to access a news site, but it is determined that the content of the news site may cause mental stress, the following prompt sentence is presented to the generative model for analysis:

[1018] "Does this news article contain content that may cause emotional stress to users?"

[1019] The generative model generates an answer to this prompt and decides whether to block access based on the answer, allowing users to always use the Internet safely and with peace of mind.

[1020] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1021] Step 1:

[1022] Receives website access requests from internet-connected devices

[1023] The server receives a website access request sent from the user's Internet connection device. As input, this request contains a URL. The server takes this URL and proceeds to the next step.

[1024] Step 2:

[1025] Gets the data of the specified URL

[1026] The server makes an HTTP request to the received URL to retrieve the web page content of that URL. The URL is given as input, and the text and image data of the web page are obtained as output. The server separates these data for the next step.

[1027] Step 3:

[1028] Analyzing data using generative models

[1029] The server analyzes the acquired text data and image data using a generative model. The text data and image data are given as input, and the analysis results are obtained as output. Specifically, the text data is analyzed using a natural language processing (NLP) model, and the image data is analyzed using a computer vision model.

[1030] Step 4:

[1031] Identifying inappropriate or fraudulent data

[1032] The server determines whether data is inappropriate or fraudulent based on the analysis results of the generative model. The analysis results are given as input, and the output is a judgment result on whether the data is inappropriate or not. Specifically, the NLP model detects harmful keywords and expressions in text, and the computer vision model detects inappropriate content in images.

[1033] Step 5:

[1034] Blocking data and displaying a warning page

[1035] If the server detects inappropriate or fraudulent data, it blocks access to that data. The input is the verdict, and the output is a blocked access flag and a warning page that explains the inappropriate content and the importance of safe web browsing.

[1036] Step 6:

[1037] Sending and displaying a warning page

[1038] The server sends the generated warning page to the user's Internet connection device. The warning page is given as input, and the warning page is displayed on the user's terminal as output. The terminal receives this warning page and displays it to the user.

[1039] Step 7:

[1040] Detecting and limiting data that may cause mental stress

[1041] During the analysis process, the server detects information that may cause mental stress. Text data and image data are input, and a judgment result regarding mental stress is obtained as output. If the information is determined to cause mental stress, it similarly blocks access and sends a warning page.

[1042] Specific actions

[1043] The natural language processing model uses pre-trained AI models (e.g., Hugging Face's transformers library) to analyze text data and generate answers to prompts such as, "Does this news article contain content that would cause mental stress to the user?"

[1044] Computer vision technology uses OpenCV and other image recognition models to scan image data in real time to detect whether it contains inappropriate images.

[1045] These specific processes enable users to use the Internet safely and comfortably.

[1046] 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.

[1047] The present invention provides a filtering system that receives a website access request from an internet-connected device, retrieves the content of the specified URL, analyzes the text and images of the content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health state in real time and suggest appropriate content according to the user's specific emotional state.

[1048] Program processing

[1049] Server Processing

[1050] The server receives a URL access request sent from the user's device. The server then makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by a generative model. The generative model uses natural language processing (NLP) technology to analyze the text content and detect harmful keywords and context. At the same time, computer vision technology is used to detect inappropriate images in the image data.

[1051] Emotion engine processing

[1052] The server recognizes the user's emotions using an emotion engine. The emotion engine analyzes the user's device's camera, microphone, and text input data to determine the user's emotional state from their facial expressions, voice, and text input. Based on the results, the server grasps the user's current emotional state in real time.

[1053] Filtering and Content Suggestions

[1054] If inappropriate or fraudulent content is detected based on the analysis results of the generative model, the server blocks access to the content and sends a warning page to the user's device. On the other hand, if content that may cause mental stress is detected, the server blocks access and displays a warning page in combination with the results of the emotion engine.

[1055] Furthermore, the server can suggest content that takes into account the user's mental health based on the results of the emotion engine. For example, if a user is under stress, it can suggest relaxing content, soothing music, or videos for relaxation.

[1056] Terminal handling

[1057] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[1058] The device also uses a camera and microphone to provide data to the emotion engine, monitoring the user's emotional state in real time.

[1059] User Interface

[1060] Users surf the Internet using a regular browser or a dedicated application. If the website they are attempting to access is deemed inappropriate or fraudulent, a warning page is displayed, protecting the user from dangerous content. The system also provides advance warnings about information that may cause mental stress, maintaining the user's mental health. Furthermore, an emotion engine monitors the user's emotional state in real time and suggests relaxing content as needed.

[1061] Specific examples

[1062] Example 1: Accessing inappropriate content

[1063] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1064] Example 2: Visiting a fraudulent website

[1065] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[1066] Example 3: Visiting a stressful news site

[1067] If a user attempts to access a news site, but the content of the site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. Furthermore, an emotion engine monitors the user's emotional state. If the server determines that the user is in a stressful state, the server blocks the access and sends a warning page to the device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[1068] The above is a specific embodiment of the present invention.

[1069] The processing flow will be explained below.

[1070] Step 1:

[1071] A user attempts to access a specific URL using a browser or application.

[1072] Step 2:

[1073] The device sends an access request to the server for the specified URL.

[1074] Step 3:

[1075] The server logs URL access requests received from the device.

[1076] Step 4:

[1077] The server makes an HTTP request to the specified URL to retrieve the web page content.

[1078] Step 5:

[1079] The server separates the content of the acquired web page into text data and image data.

[1080] Step 6:

[1081] The server analyzes the text data using the generative model.

[1082] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[1083] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[1084] Step 7:

[1085] The server analyzes the image data using the generative model.

[1086] It uses computer vision techniques to evaluate the content of images and detect inappropriate images or specific patterns.

[1087] Step 8:

[1088] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[1089] If there is a problem, the server will block access to the URL.

[1090] Step 9:

[1091] If the server blocks inappropriate or deceptive content, it generates a warning page.

[1092] The warning page will include information about why your access was blocked and the importance of safe internet use.

[1093] Step 10:

[1094] The server sends a warning page to the device.

[1095] Step 11:

[1096] The terminal displays the received warning page to the user.

[1097] Step 12:

[1098] The server analyzes content that may cause mental stress.

[1099] Detects trigger words and images to determine if they contain stress-related information.

[1100] Step 13:

[1101] If the server detects stress-related content, it will also block access to it.

[1102] In this case, a warning page is also generated and sent to the terminal.

[1103] Step 14:

[1104] The device will again display a warning page to the user to protect their mental health.

[1105] Emotion engine processing

[1106] Step 15:

[1107] The server uses an emotion engine to recognize the user's emotional state.

[1108] It collects and analyzes data from the device's camera, microphone, and text input, and determines the user's emotional state from their facial expressions, voice, and text input.

[1109] Step 16:

[1110] If the emotion engine determines that the user is stressed, the server uses this information to make decisions about suggesting more appropriate content.

[1111] Step 17:

[1112] The server selects content to reduce stress based on the results of the emotion engine.

[1113] It includes relaxing videos, music, articles, etc.

[1114] Step 18:

[1115] The server generates a page for proposing the selected stress reduction content to the user.

[1116] Step 19:

[1117] The server generates a proposal page and sends it to the terminal.

[1118] Step 20:

[1119] The terminal displays the received proposal page to the user.

[1120] It is expected that users will view the suggested content and reduce their mental stress.

[1121] The above is the flow of processing in a specific embodiment of the present invention.

[1122] Example 2

[1123] 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."

[1124] There is a lot of inappropriate or fraudulent content on the Internet, and accessing it poses a significant risk to users. Furthermore, involuntary access to content that causes mental stress can have a negative impact on users' mental health. Conventional filtering systems have had difficulty completely eliminating such risks. The present invention aims to solve these problems, protect users from harmful content on the Internet, and maintain their mental health.

[1125] 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.

[1126] In this invention, the server includes means for receiving a website access request from an Internet-connected device, means for retrieving content from a specified URL, means for analyzing text and images in the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for collecting data to recognize a user's emotional state, and means for analyzing the collected data using an emotion engine to determine the user's emotional state. This protects users from inappropriate or fraudulent content and from content that may cause mental stress, thereby enabling them to maintain their mental health.

[1127] An "Internet-connected device" is a device that has the ability to connect to the Internet and is capable of web browsing and data communication. Examples include smartphones, tablets, and personal computers.

[1128] A "website access request" is an HTTP request sent when a user attempts to access a particular web page using an Internet-connected device.

[1129] A "specified URL" is the address of a particular web page that a user wishes to access, written in Uniform Resource Locator (URL) format.

[1130] A "generative model" is an AI model that is trained on large datasets and used to analyze text and images, allowing it to detect specific patterns and anomalies.

[1131] "Text analytics" is the process of analyzing text data using generative models, specifically using natural language processing techniques to understand and classify the content of the text.

[1132] "Image analysis" is the process of analyzing image data using generative models. It refers to the use of computer vision techniques to detect features and anomalies in images.

[1133] "Inappropriate Content" refers to content that is inappropriate for minors, or that contains violent, sexual, or discriminatory material, and that may have a harmful effect on users.

[1134] "Fraudulent content" refers to web content created with the intention of deceiving users, and which may result in financial loss or the leakage of personal information.

[1135] A "warning page" is a page that is displayed when the web page that a user is attempting to access is determined to be inappropriate or fraudulent, and serves to warn the user.

[1136] An "emotion engine" is a device or software that analyzes a user's emotional state and determines the emotional state based on the user's facial expressions, voice, text input, etc.

[1137] "Mental stress" refers to the psychological burden or state of tension caused by external stimuli or information.

[1138] "Content suggestion" is the process of recommending suitable content, such as videos or music with a relaxing effect, taking into account the user's current emotional state and mental health.

[1139] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[1140] The system's hardware includes internet-connected devices (e.g., smartphones, tablets, PCs, etc.), and its software includes generative AI models (e.g., the BERT model using NLP techniques and the YOLOv5 model using computer vision techniques) and emotion engines (e.g., the OpenFace library).

[1141] When a user attempts to access a specific URL using an Internet-connected device (hereinafter referred to as a terminal), the access request is sent to a server. The server makes an HTTP request to the specified URL and retrieves the web page content. The retrieved content is in HTML format, and web scraping technology (such as Beautiful Soup or Selenium) is used to separate it into text data and image data.

[1142] Text data is analyzed using generative AI models, which leverage NLP techniques to detect harmful keywords and contexts. A specific example is the BERT model, which uses the Transformers library. Similarly, image data is analyzed using computer vision techniques, such as YOLOv5 and OpenCV, to detect inappropriate images.

[1143] To recognize the user's emotional state, the emotion engine collects data from the device's camera, microphone, and text input data. The emotion engine (e.g., the OpenFace library) analyzes the collected data and determines the user's current emotional state from their facial expressions and voice.

[1144] If the server detects inappropriate or fraudulent content based on the analysis results of the generative AI model and the emotion engine, it blocks access to the content and sends a warning page to the device. If it detects content that may cause mental stress, it combines the results of the emotion engine to block access and display a warning page. Furthermore, it suggests appropriate content (e.g., videos or music with a relaxing effect) based on the user's emotional state.

[1145] As a concrete example, in the case of access to inappropriate content, an underage user attempts to access a website containing adult content. The access request from the device is sent to the server, which retrieves and analyzes the content of the URL. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1146] An example prompt might be, "Using a generative AI model, analyze the content of the specified URL to detect inappropriate or deceptive content. Also, consider the results of the user's sentiment engine to suggest appropriate content for the user."

[1147] The above is a detailed description of a specific embodiment of the present invention.

[1148] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1149] Step 1:

[1150] When a user attempts to access a specific URL, the device sends the access request to the server. The input is the URL entered by the user into the device, and the output is an HTTP request to the server. Specifically, a web browser or a dedicated application sends the URL to the server.

[1151] Step 2:

[1152] Based on the received URL access request, the server makes an HTTP request to the specified URL and retrieves the web page content. The input is the URL sent from the terminal, and the output is the web page content in HTML format. Specifically, the server communicates with the web server using the HTTP protocol and downloads the web page.

[1153] Step 3:

[1154] The server separates the retrieved HTML content into text data and image data. This process uses web scraping technology (e.g., Beautiful Soup or Selenium). The input is the web page content in HTML format, and the output is text data and image data. Specifically, the HTML parser analyzes the HTML tags and extracts the text and images.

[1155] Step 4:

[1156] The server analyzes the text data using a generative AI model. Here, natural language processing technology (e.g., the BERT model) is used to analyze the text content and detect harmful keywords and context. The input is the separated text data, and the output is an evaluation score for the text as the analysis result. Specifically, the text data is input into the generative model, and natural language processing is performed to evaluate the harmfulness.

[1157] Step 5:

[1158] The server analyzes image data using a generative AI model. Here, computer vision techniques (e.g., YOLOv5 or OpenCV) are used to detect inappropriate images. The input is the separated image data, and the output is an evaluation score for the image as the analysis result. Specifically, the image data is input into the generative model, and a computer vision algorithm is applied to evaluate its harmfulness.

[1159] Step 6:

[1160] The device collects camera, microphone, and text input data to recognize the user's emotional state. The input is the user's facial expression, voice, and text input data, and the output is to send this data to the server. Specific operations include the process in which the device's sensors capture data and send it to the server.

[1161] Step 7:

[1162] The server uses an emotion engine to analyze the collected data and determine the user's emotional state. Here, the analysis is performed using emotion recognition technology (e.g., the OpenFace library). The input is facial expressions, voice, and text data sent from the device, and the output is an evaluation result of the user's emotional state. Specifically, the emotion engine processes the data and analyzes the emotional state in real time.

[1163] Step 8:

[1164] If the server detects inappropriate or deceptive content based on the analysis results of the generative AI model and emotion engine, it blocks access to that content. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is an instruction to block access and the generation of a warning page. Specifically, the server executes the access control logic and generates a warning page if necessary.

[1165] Step 9:

[1166] If the server detects content that may cause mental stress, it combines the results of the emotion engine to block access and send a warning page to the terminal. At the same time, it suggests appropriate content based on the user's emotional state. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is a list of suggested content. Specifically, the server runs the content suggestion algorithm to generate content appropriate for the user.

[1167] Step 10:

[1168] The terminal receives the analysis results, warning pages, and suggested content from the server and displays them to the user. The input is the warning page and suggested content sent from the server, and the output is the screen displayed to the user. Specifically, the terminal updates the screen according to instructions from the server and provides the necessary information to the user.

[1169] (Application example 2)

[1170] 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."

[1171] With the spread of the Internet, users have more opportunities to access various websites, many of which contain inappropriate or fraudulent content. Access to content that may cause mental stress is also increasing. Encountering such content can harm users' mental health, so a system that automatically filters this content and protects users' health is needed.

[1172] 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 a website access request from an Internet-connected device, means for acquiring content from a specified URL, means for analyzing the text and images of the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for recognizing emotions from the user's facial expression or voice, means for determining whether the recognized emotion is a stress state, and means for suggesting appropriate content according to the stress state. This not only protects users from inappropriate or fraudulent content when browsing websites, but also makes it possible to provide appropriate content to reduce mental stress.

[1173] "Internet-connected device" refers to any electronic device that can connect to the Internet.

[1174] "Website Access Request" refers to a request by a user to access a particular website.

[1175] "Specified URL" refers to the specific web address that the user attempted to access.

[1176] "Content" refers to information such as text data and image data displayed on a website.

[1177] A "generative model" refers to algorithms or software for generating and analyzing data based on artificial intelligence techniques.

[1178] "Means for analyzing text and images" refers to methods for evaluating and analyzing text data and image data within the Content.

[1179] "Inappropriate Content" refers to information on a website that contains material that is deemed harmful to users.

[1180] "Deceptive content" refers to false information or pages created with the intent to deceive users.

[1181] "Blocking measures" refers to methods of controlling users' access to content that is deemed inappropriate or fraudulent.

[1182] "Warning Page" refers to a notification page that informs users that certain content is inappropriate or deceptive.

[1183] "Means for recognizing emotions from a user's facial expression or voice" refers to technology that analyzes and determines a user's emotional state from their facial expression or voice.

[1184] A "stressed state" refers to a state in which the user feels mental tension or anxiety.

[1185] The "means for suggesting appropriate content" refers to a method for recommending content that has a relaxing effect based on the user's emotional state.

[1186] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[1187] System Configuration

[1188] Server Configuration

[1189] The server requires the following hardware and software:

[1190] Hardware: A server with a fast processor and large memory capacity.

[1191] software:

[1192] Generative model: OpenAI GPT-4

[1193] Computer Vision Technology: Google Vision API

[1194] Emotion engine: Affectiva SDK

[1195] Programming languages ​​and frameworks: Python, Django, Nginx

[1196] User's device

[1197] The user's device must have the following features:

[1198] Camera and microphone

[1199] Internet connection function

[1200] Dedicated applications

[1201] Program processing

[1202] Server Processing

[1203] The server first receives a URL access request sent from the user's device. It then makes an HTTP request to the specified URL to retrieve the webpage content. The retrieved content is separated into text data and image data and analyzed using a generative model (GPT-4) and computer vision technology (Google Vision API). If inappropriate or fraudulent content is detected, the server blocks access to the content and sends a warning page to the user's device.

[1204] The server also uses an emotion engine (Affectiva SDK) to recognize the user's emotions in real time. It determines the user's emotional state based on facial expressions, voice, and text input, and if it determines that the user is particularly stressed, it suggests content that will have a relaxing effect.

[1205] Terminal handling

[1206] The user's device sends a request to the server when they attempt to access a specific URL, and displays appropriate content or a warning page based on the analysis results.The camera and microphone are used to provide data to the emotion engine, which monitors the user's emotional state in real time.

[1207] Specific examples

[1208] Example 1: Accessing inappropriate content

[1209] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1210] Example 2: Visiting a fraudulent website

[1211] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and displays a warning page on the device. In this way, access to the fraudulent website is prevented.

[1212] Example 3: Visiting a stressful news site

[1213] A user may attempt to access a news site, but the content of that site may cause mental stress. The server analyzes the text and image content and determines that it may cause mental stress. Furthermore, the emotion engine monitors the user's emotional state. If it determines that the user is in a stressful state, the server blocks the access and sends a warning page to the user's device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[1214] Prompt Sentence Examples

[1215] "Emotion-aware filtering app. A user is about to visit a new news site. Check if the site contains harmful content and display a warning page if necessary. Also, if the user is under stress, suggest relaxation music."

[1216] The above is a specific description of the embodiment of the present invention. This system protects users from inappropriate or fraudulent content and reduces mental stress.

[1217] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1218] Step 1:

[1219] When a user tries to access a specific URL, the device generates a website access request and sends it to the server. The input is the URL specified by the user, and the output is the access request sent to the server.

[1220] Step 2:

[1221] The server receives a URL access request sent from the user's terminal. The input is the access request, and the output is a confirmation of the request. Based on this request, the server makes an HTTP request to the specified URL.

[1222] Step 3:

[1223] The server uses an HTTP request to retrieve the content of a specified URL. The input is the specified URL, and the output is the retrieved web page content (text data and image data).

[1224] Step 4:

[1225] The server separates the retrieved content into text data and image data. The input of this step is the retrieved web page content, and the output is the separated text data and image data. The server analyzes the text content using a generative model (GPT-4) and analyzes the image content using computer vision technology (Google Vision API).

[1226] Step 5:

[1227] The server determines whether content is inappropriate or fraudulent based on the analysis results of the generative model. The input is the analyzed text data and image data, and the output is the judgment result. If inappropriate or fraudulent content is detected, the server proceeds to the next step based on the judgment result.

[1228] Step 6:

[1229] If the server determines that the content is inappropriate or fraudulent, it blocks access to it and sends a warning page to the terminal. The input of this step is the result of the determination, and the output is a block instruction and the sending of a warning page.

[1230] Step 7:

[1231] The terminal receives the warning page sent from the server and displays it to the user. The input is the warning page, and the output is the display of a warning message to the user. This warning prevents the user from accessing inappropriate content or fraudulent sites.

[1232] Step 8:

[1233] The server uses an emotion engine (Affectiva SDK) to recognize emotions from the user's facial expressions or voice in real time. The input is facial expression data or voice data sent from the device, and the output is the user's emotional state.

[1234] Step 9:

[1235] The server determines whether the recognized emotion is a stress state. The input is the analysis result of the emotion engine, and the output is the stress state determination result.

[1236] Step 10:

[1237] If the server determines that the user is in a stressful state, it proposes appropriate content. The input to this step is the result of the stress state determination, and the output is a recommendation of content that has a relaxing effect. This appropriate content proposal is sent to the user's device.

[1238] The above is the specific processing flow of the system that realizes this application example. This system protects users from inappropriate or fraudulent content when browsing websites, thereby reducing mental stress.

[1239] 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.

[1240] 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.

[1241] 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.

[1242] [Fourth embodiment]

[1243] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1244] 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.

[1245] 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).

[1246] 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.

[1247] 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.

[1248] 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).

[1249] 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.

[1250] 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.

[1251] 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.

[1252] 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.

[1253] 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.

[1254] 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.

[1255] 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."

[1256] The present invention is a filtering system that receives website access requests from internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress.

[1257] Program processing

[1258] Server Processing

[1259] The server receives a URL access request sent from the user's device. After receiving the request, the server makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by the generative model.

[1260] The generative model uses natural language processing (NLP) technology to analyze the text content of a page and evaluate whether it contains harmful keywords or expressions, while also using computer vision technology to evaluate image data for inappropriate images.

[1261] If the analysis detects inappropriate or fraudulent content, the server blocks access to the content and sends a warning page to the user's device explaining the reason for the block and the importance of safe web browsing.

[1262] During the analysis, information that may cause psychological stress is also detected, and if such information is found, the server blocks access to it as well and sends a warning page.

[1263] Terminal handling

[1264] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[1265] User Interface

[1266] Users can surf the internet using a regular browser or a dedicated application. If the website they are trying to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous content. It also provides advance warnings about information that may cause mental stress, protecting users' mental health.

[1267] Specific examples

[1268] Example 1: Accessing inappropriate content

[1269] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1270] Example 2: Visiting a fraudulent website

[1271] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[1272] Example 3: Visiting a stressful news site

[1273] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. The server blocks the access and sends a warning page. The terminal displays the warning page to the user, and their mental health is protected.

[1274] The above is a specific embodiment of the present invention.

[1275] The processing flow will be explained below.

[1276] Step 1:

[1277] A user attempts to access a specific URL using a browser or application.

[1278] Step 2:

[1279] The device sends an access request to the server for the specified URL.

[1280] Step 3:

[1281] The server logs URL access requests received from the device.

[1282] Step 4:

[1283] The server makes an HTTP request to the specified URL to retrieve the web page content.

[1284] Step 5:

[1285] The server separates the content of the acquired web page into text data and image data.

[1286] Step 6:

[1287] The server analyzes the text data using the generative model.

[1288] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[1289] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[1290] Step 7:

[1291] The server analyzes the image data using the generative model.

[1292] Computer vision techniques are used to evaluate the content of the image.

[1293] Detect inappropriate images and specific patterns.

[1294] Step 8:

[1295] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[1296] If there is a problem, the server will block access to the URL.

[1297] Step 9:

[1298] If the server blocks inappropriate or deceptive content, it generates a warning page.

[1299] The warning page will include information about why your access was blocked and the importance of safe internet use.

[1300] Step 10:

[1301] The server sends a warning page to the device.

[1302] Step 11:

[1303] The terminal displays the received warning page to the user.

[1304] Step 12:

[1305] The server analyzes content that may cause mental stress.

[1306] Detects trigger words and images to determine if they contain stress-related information.

[1307] Step 13:

[1308] If the server detects stress-related content, it will also block access to it.

[1309] In this case, a warning page is also generated and sent to the terminal.

[1310] Step 14:

[1311] The device will again display a warning page to the user to protect their mental health.

[1312] Through the above steps, the filtering system of the present invention provides an environment in which users can use the Internet safely and securely.

[1313] Example 1

[1314] 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."

[1315] The Internet contains inappropriate and fraudulent content that is harmful to many users, including minors and the elderly, and can also cause psychological stress. Therefore, a filtering system is needed to ensure that users can use the Internet safely and comfortably. However, existing filtering systems lack the ability to analyze text and images, making them unable to effectively block harmful content.

[1316] 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.

[1317] In this invention, the server includes means for receiving a website access request sent from a terminal, means for acquiring the content of the requested URL, means for separating the acquired content into text data and image data, means for analyzing the text data using a generative AI model to detect harmful keywords and expressions, means for analyzing the image data using computer vision technology to detect inappropriate images, means for determining inappropriate or fraudulent content based on the analysis results of the generative AI model, and means for blocking inappropriate or fraudulent content and sending a warning page to the terminal. This makes it possible to effectively block inappropriate or fraudulent content on the Internet and provide an environment in which users can use the Internet safely and comfortably.

[1318] A "terminal" is a device such as a computer or smartphone that allows a user to connect to the Internet.

[1319] A "server" is a computer system that provides information and performs processing on the Internet.

[1320] A "website access request" is a request that a user sends from a terminal to a server to access a specific website.

[1321] A "URL" is an address on the Internet that points to a specific web page.

[1322] "Content" is a general term for information displayed on a web page, and includes text data and image data.

[1323] A "generative AI model" is a statistical model for generating and analyzing data using artificial intelligence techniques.

[1324] "Natural language processing technology" is a technology that enables computers to understand and process human language.

[1325] "Computer vision technology" is a technology that allows computers to analyze and recognize images and videos.

[1326] A "warning page" is a web page that blocks access to inappropriate or fraudulent content and notifies users of this.

[1327] "Harmful keywords and expressions" are words or phrases that may threaten a user's mental health or safe use of the Internet.

[1328] An "inappropriate image" is an image that is perceived by a user as psychologically harmful or unpleasant.

[1329] "Fraudulent content" is information intended to deceive users and cause them financial harm.

[1330] "Mental stress" is the emotional strain that causes a user's mental health to deteriorate.

[1331] There is a lot of inappropriate or fraudulent content on the Internet that can threaten users' safety and mental health. The present invention provides an advanced filtering system to protect users from such harmful content.

[1332] Server hardware and software configuration

[1333] The server uses a high-performance computer system and has the following software configuration:

[1334] 1. Web server software: Use Apache HTTP Server or Nginx.

[1335] 2. Python programming environment: Uses the Python language and related libraries to perform natural language processing, image analysis, etc.

[1336] 3. Generative AI models: Use Hugging Face's Transformers model or OpenAI's GPT-3.

[1337] 4. Natural Language Processing techniques: Use pre-trained models such as BERT and RoBERTa.

[1338] 5. Computer vision technology: Image analysis is performed using TensorFlow and Detectron2.

[1339] System Operation Overview

[1340] The system of the present invention receives a website access request sent from a user's device and retrieves the content of the specified URL based on the request. The retrieved content is separated into text data and image data, each of which is analyzed using a generative AI model. If inappropriate or fraudulent content is detected based on the analysis results, the server blocks access to the content and sends a warning page to the user's device. In addition, content that may cause mental stress may also be detected, and access to such information is also restricted.

[1341] Specific examples

[1342] Example 1: Accessing inappropriate content

[1343] When a minor user attempts to access a website containing adult content using a browser, an access request is sent from the device to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is inappropriate, blocks the access. A warning page is displayed on the device, preventing access to the adult content.

[1344] Example 2: Visiting a fraudulent website

[1345] An elderly person clicks on a link in an email and attempts to access a financial fraud website. In this case, the device sends an access request to the server. The server retrieves the URL's content and, if the generative AI model determines that the content is fraudulent, blocks the access. A warning page is displayed on the device, preventing access to the fraudulent website.

[1346] Example 3: Visiting a stressful news site

[1347] If a user attempts to access a news site, but the content of the news site is likely to cause mental stress, the server will analyze the text and image content and determine that it may cause mental stress. The server will block the access and send a warning page, which will be displayed on the device, protecting the user's mental health.

[1348] Prompt Sentence Examples

[1349] "Please explain in natural language how you analyze the URL of the website a user is attempting to access, determine if the URL contains inappropriate or deceptive content, and block access if necessary. Also, please specify the names of any specific hardware or software used."

[1350] As a result, by using the system of the present invention, users are protected from harmful content on the Internet, enabling safe and comfortable web browsing.

[1351] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1352] Step 1:

[1353] The server receives a website access request sent from the user's terminal.

[1354] Input: An HTTP GET request containing the URL of the website the user wants to access.

[1355] What happens: The server receives this request using web server software (e.g., Apache HTTP Server or Nginx).

[1356] Output: The requested URL.

[1357] Step 2:

[1358] The server makes an HTTP request to the received URL to retrieve the web page content.

[1359] Input: The requested URL.

[1360] What happens: The server uses Python's requests library to download HTML data from the specified URL. It then executes requests.get(url) to get the HTML content of the web page.

[1361] Output: The retrieved HTML data.

[1362] Step 3:

[1363] The server separates the acquired content into text data and image data.

[1364] Input: The retrieved HTML data.

[1365] What happens: The server uses the BeautifulSoup library to parse the HTML data and extract text and image data. It runs soup = BeautifulSoup(html_data, 'html.parser') and separates the data using functions like text_data = soup.get_text() and image_urls = [img['src'] for img in soup.find_all('img')] .

[1366] Output: Separated text and image data.

[1367] Step 4:

[1368] The server uses a generative AI model to analyze the extracted text data and detect harmful keywords and expressions.

[1369] Input: Separated text data.

[1370] Specific operation: The server uses Hugging Face's Transformers library to analyze the data by calling a generative AI model (e.g., BERT, RoBERTa). It then executes nlp_pipeline = pipeline('sentiment-analysis') and runs result = nlp_pipeline(text_data) to detect harmful keywords.

[1371] Output: A list of harmful keywords and expressions contained in the text data.

[1372] Step 5:

[1373] The server uses computer vision technology to analyze the extracted image data and detect inappropriate images.

[1374] Input: Segmented image data.

[1375] Specific operation: The server performs image analysis using TensorFlow and Detectron2. It downloads an image, inputs it into the model, executes detectron2_predictor = DefaultPredictor(cfg) and outputs = detectron2_predictor(image), and evaluates the image.

[1376] Output: A list of inappropriate images.

[1377] Step 6:

[1378] The server uses the analysis to determine whether the URL contains inappropriate or deceptive content.

[1379] Input: Analysis results of text data and image data.

[1380] What it does: The server evaluates the results of the generative AI model and determines whether they contain inappropriate or deceptive content, for example, by using conditional branching such as if 'toxic' in text_analysis_result or 'unsafe' in image_analysis_result.

[1381] Output: Access blocking decision result (whether to block or not).

[1382] Step 7:

[1383] If the server determines that the content is inappropriate or fraudulent, it will send a warning page to the user's device.

[1384] Input: Block decision result.

[1385] Specific operation: The server generates a warning page using HTML and CSS and sends it to the terminal as an HTTP response. response_html = " <h1> Warning< / h1> This site is blocked. " and send it as return response_html.

[1386] Output: A warning page that is displayed on the user's device.

[1387] Step 8:

[1388] The terminal displays a warning page to the user.

[1389] Input: The HTML of the warning page received from the server.

[1390] Specific operation: The user's device renders the received HTML in the browser and displays the warning content.

[1391] Output: The warning page that the user sees.

[1392] (Application example 1)

[1393] 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."

[1394] As the Internet becomes more widespread, users are increasingly at risk of accessing inappropriate content or fraudulent websites. Furthermore, there is a large amount of information that can potentially cause psychological stress, making it difficult for users to use the Internet safely and comfortably. Therefore, there is a need for a system that can detect dangerous content in real time and protect users.

[1395] 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.

[1396] In this invention, the server includes means for receiving a website access request from an Internet connection device, means for acquiring data from a specified URL, means for analyzing text and images in the data using a generative model, means for determining whether the data is inappropriate or fraudulent based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent data and displaying a warning page, and means for scanning data in real time and monitoring user Internet usage, thereby enabling users to use the Internet safely and comfortably.

[1397] An "Internet connection device" is a device for connecting to the Internet and capable of sending access requests to websites.

[1398] A "website access request" is a request sent from an Internet-connected device when a user attempts to access a specific website on the Internet.

[1399] A "specified URL" refers to the address of a particular web page that a user wishes to access.

[1400] "Data" means the content on the Website, including text and image data.

[1401] A "generative model" is a model based on artificial intelligence techniques used to analyze text and image data.

[1402] "Analysis results" refers to the analysis results of data obtained by a generative model, which are used to determine whether the data is inappropriate or fraudulent.

[1403] "Inappropriate Content" means content that is deemed harmful or offensive to users.

[1404] "Deceptive data" is content determined to be intended to deceive or mislead users.

[1405] A "warning page" is a page that is displayed when inappropriate or fraudulent data is detected to notify the user and block access.

[1406] "Real-time scanning" refers to the process of analyzing data as soon as it is acquired.

[1407] "Internet usage monitoring measures" means measures that continuously track and analyze users' Internet activity to detect inappropriate or fraudulent data.

[1408] Server Processing

[1409] The server receives a website access request sent from the user's Internet connection device. Based on the received access request, the server makes an HTTP request to the specified URL to obtain the web page data. This data is separated into text data and image data and analyzed using a generative model.

[1410] Generative models use natural language processing (NLP) techniques to analyze the text content of web pages, for example, to assess whether the text contains harmful keywords or inappropriate language, and simultaneously use computer vision techniques to analyze image data and assess whether it contains inappropriate images.

[1411] If the analysis detects inappropriate or fraudulent data, the server blocks access to it and sends a warning page to the user's internet connection, explaining the reason for the block and the importance of safe web browsing.

[1412] Additionally, the server detects any information that may cause psychological stress, and if such information is included, the server will similarly block access and send a warning page.

[1413] Terminal handling

[1414] When a user attempts to access a specific URL, the user's Internet connection device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate data or a warning page and displays it to the user. If access to inappropriate data or a fraudulent website is blocked, the terminal displays a warning page to warn the user.

[1415] User Interface

[1416] Users can use a regular browser or dedicated application to access the Internet. If the website they are about to access is deemed inappropriate or fraudulent, a warning page will be displayed, protecting users from dangerous data. Advance warnings are also provided for information that may cause mental stress, protecting users' mental health.

[1417] Hardware and software used

[1418] The system uses the following hardware and software:

[1419] Hardware: Internet-connected devices (smartphones, tablets, PCs, etc.)

[1420] Software: Server-side HTTP request processing frameworks (e.g., Python's requests library), natural language processing models (e.g., Hugging Face's transformers library), computer vision models (e.g., OpenCV)

[1421] Specific examples

[1422] If a user attempts to access a news site, but it is determined that the content of the news site may cause mental stress, the following prompt sentence is presented to the generative model for analysis:

[1423] "Does this news article contain content that may cause emotional stress to users?"

[1424] The generative model generates an answer to this prompt and decides whether to block access based on the answer, allowing users to always use the Internet safely and with peace of mind.

[1425] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1426] Step 1:

[1427] Receives website access requests from internet-connected devices

[1428] The server receives a website access request sent from the user's Internet connection device. As input, this request contains a URL. The server takes this URL and proceeds to the next step.

[1429] Step 2:

[1430] Gets the data of the specified URL

[1431] The server makes an HTTP request to the received URL to retrieve the web page content of that URL. The URL is given as input, and the text and image data of the web page are obtained as output. The server separates these data for the next step.

[1432] Step 3:

[1433] Analyzing data using generative models

[1434] The server analyzes the acquired text data and image data using a generative model. The text data and image data are given as input, and the analysis results are obtained as output. Specifically, the text data is analyzed using a natural language processing (NLP) model, and the image data is analyzed using a computer vision model.

[1435] Step 4:

[1436] Identifying inappropriate or fraudulent data

[1437] The server determines whether data is inappropriate or fraudulent based on the analysis results of the generative model. The analysis results are given as input, and the output is a judgment result on whether the data is inappropriate or not. Specifically, the NLP model detects harmful keywords and expressions in text, and the computer vision model detects inappropriate content in images.

[1438] Step 5:

[1439] Blocking data and displaying a warning page

[1440] If the server detects inappropriate or fraudulent data, it blocks access to that data. The input is the verdict, and the output is a blocked access flag and a warning page that explains the inappropriate content and the importance of safe web browsing.

[1441] Step 6:

[1442] Sending and displaying a warning page

[1443] The server sends the generated warning page to the user's Internet connection device. The warning page is given as input, and the warning page is displayed on the user's terminal as output. The terminal receives this warning page and displays it to the user.

[1444] Step 7:

[1445] Detecting and limiting data that may cause mental stress

[1446] During the analysis process, the server detects information that may cause mental stress. Text data and image data are input, and a judgment result regarding mental stress is obtained as output. If the information is determined to cause mental stress, it similarly blocks access and sends a warning page.

[1447] Specific actions

[1448] The natural language processing model uses pre-trained AI models (e.g., Hugging Face's transformers library) to analyze text data and generate answers to prompts such as, "Does this news article contain content that would cause mental stress to the user?"

[1449] Computer vision technology uses OpenCV and other image recognition models to scan image data in real time to detect whether it contains inappropriate images.

[1450] These specific processes enable users to use the Internet safely and comfortably.

[1451] 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.

[1452] The present invention provides a filtering system that receives a website access request from an internet-connected device, retrieves the content of the specified URL, analyzes the text and images of the content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health state in real time and suggest appropriate content according to the user's specific emotional state.

[1453] Program processing

[1454] Server Processing

[1455] The server receives a URL access request sent from the user's device. The server then makes an HTTP request to the specified URL to retrieve the web page content. The retrieved content is separated into text data and image data, and analyzed by a generative model. The generative model uses natural language processing (NLP) technology to analyze the text content and detect harmful keywords and context. At the same time, computer vision technology is used to detect inappropriate images in the image data.

[1456] Emotion engine processing

[1457] The server recognizes the user's emotions using an emotion engine. The emotion engine analyzes the user's device's camera, microphone, and text input data to determine the user's emotional state from their facial expressions, voice, and text input. Based on the results, the server grasps the user's current emotional state in real time.

[1458] Filtering and Content Suggestions

[1459] If inappropriate or fraudulent content is detected based on the analysis results of the generative model, the server blocks access to the content and sends a warning page to the user's device. On the other hand, if content that may cause mental stress is detected, the server blocks access and displays a warning page in combination with the results of the emotion engine.

[1460] Furthermore, the server can suggest content that takes into account the user's mental health based on the results of the emotion engine. For example, if a user is under stress, it can suggest relaxing content, soothing music, or videos for relaxation.

[1461] Terminal handling

[1462] When a user attempts to access a specific URL, their internet-connected device (terminal) sends the request to the server. Based on the analysis results from the server, the terminal receives appropriate content or a warning page and displays it to the user. If access to an inappropriate or fraudulent website is blocked, the terminal displays a warning page to warn the user.

[1463] The device also uses a camera and microphone to provide data to the emotion engine, monitoring the user's emotional state in real time.

[1464] User Interface

[1465] Users surf the Internet using a regular browser or a dedicated application. If the website they are attempting to access is deemed inappropriate or fraudulent, a warning page is displayed, protecting the user from dangerous content. The system also provides advance warnings about information that may cause mental stress, maintaining the user's mental health. Furthermore, an emotion engine monitors the user's emotional state in real time and suggests relaxing content as needed.

[1466] Specific examples

[1467] Example 1: Accessing inappropriate content

[1468] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1469] Example 2: Visiting a fraudulent website

[1470] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and sends a warning page. The device displays the warning page to the user, preventing access to the fraudulent website.

[1471] Example 3: Visiting a stressful news site

[1472] If a user attempts to access a news site, but the content of the site is likely to cause mental stress, the server analyzes the text and image content and determines that it may cause mental stress. Furthermore, an emotion engine monitors the user's emotional state. If the server determines that the user is in a stressful state, the server blocks the access and sends a warning page to the device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[1473] The above is a specific embodiment of the present invention.

[1474] The processing flow will be explained below.

[1475] Step 1:

[1476] A user attempts to access a specific URL using a browser or application.

[1477] Step 2:

[1478] The device sends an access request to the server for the specified URL.

[1479] Step 3:

[1480] The server logs URL access requests received from the device.

[1481] Step 4:

[1482] The server makes an HTTP request to the specified URL to retrieve the web page content.

[1483] Step 5:

[1484] The server separates the content of the acquired web page into text data and image data.

[1485] Step 6:

[1486] The server analyzes the text data using the generative model.

[1487] The generative model uses natural language processing (NLP) techniques to evaluate the text content within a page.

[1488] Detecting specific keywords and contexts to determine whether they may be inappropriate, deceptive, or potentially frustrating.

[1489] Step 7:

[1490] The server analyzes the image data using the generative model.

[1491] It uses computer vision techniques to evaluate the content of images and detect inappropriate images or specific patterns.

[1492] Step 8:

[1493] The server determines whether the content is inappropriate or deceptive based on the results of analyzing the generative model.

[1494] If there is a problem, the server will block access to the URL.

[1495] Step 9:

[1496] If the server blocks inappropriate or deceptive content, it generates a warning page.

[1497] The warning page will include information about why your access was blocked and the importance of safe internet use.

[1498] Step 10:

[1499] The server sends a warning page to the device.

[1500] Step 11:

[1501] The terminal displays the received warning page to the user.

[1502] Step 12:

[1503] The server analyzes content that may cause mental stress.

[1504] Detects trigger words and images to determine if they contain stress-related information.

[1505] Step 13:

[1506] If the server detects stress-related content, it will also block access to it.

[1507] In this case, a warning page is also generated and sent to the terminal.

[1508] Step 14:

[1509] The device will again display a warning page to the user to protect their mental health.

[1510] Emotion engine processing

[1511] Step 15:

[1512] The server uses an emotion engine to recognize the user's emotional state.

[1513] It collects and analyzes data from the device's camera, microphone, and text input, and determines the user's emotional state from their facial expressions, voice, and text input.

[1514] Step 16:

[1515] If the emotion engine determines that the user is stressed, the server uses this information to make decisions about suggesting more appropriate content.

[1516] Step 17:

[1517] The server selects content to reduce stress based on the results of the emotion engine.

[1518] It includes relaxing videos, music, articles, etc.

[1519] Step 18:

[1520] The server generates a page for proposing the selected stress reduction content to the user.

[1521] Step 19:

[1522] The server generates a proposal page and sends it to the terminal.

[1523] Step 20:

[1524] The terminal displays the received proposal page to the user.

[1525] It is expected that users will view the suggested content and reduce their mental stress.

[1526] The above is the flow of processing in a specific embodiment of the present invention.

[1527] Example 2

[1528] 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."

[1529] There is a lot of inappropriate or fraudulent content on the Internet, and accessing it poses a significant risk to users. Furthermore, involuntary access to content that causes mental stress can have a negative impact on users' mental health. Conventional filtering systems have had difficulty completely eliminating such risks. The present invention aims to solve these problems, protect users from harmful content on the Internet, and maintain their mental health.

[1530] 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.

[1531] In this invention, the server includes means for receiving a website access request from an Internet-connected device, means for retrieving content from a specified URL, means for analyzing text and images in the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for collecting data to recognize a user's emotional state, and means for analyzing the collected data using an emotion engine to determine the user's emotional state. This protects users from inappropriate or fraudulent content and from content that may cause mental stress, thereby enabling them to maintain their mental health.

[1532] An "Internet-connected device" is a device that has the ability to connect to the Internet and is capable of web browsing and data communication. Examples include smartphones, tablets, and personal computers.

[1533] A "website access request" is an HTTP request sent when a user attempts to access a particular web page using an Internet-connected device.

[1534] A "specified URL" is the address of a particular web page that a user wishes to access, written in Uniform Resource Locator (URL) format.

[1535] A "generative model" is an AI model that is trained on large datasets and used to analyze text and images, allowing it to detect specific patterns and anomalies.

[1536] "Text analytics" is the process of analyzing text data using generative models, specifically using natural language processing techniques to understand and classify the content of the text.

[1537] "Image analysis" is the process of analyzing image data using generative models. It refers to the use of computer vision techniques to detect features and anomalies in images.

[1538] "Inappropriate Content" refers to content that is inappropriate for minors, or that contains violent, sexual, or discriminatory material, and that may have a harmful effect on users.

[1539] "Fraudulent content" refers to web content created with the intention of deceiving users, and which may result in financial loss or the leakage of personal information.

[1540] A "warning page" is a page that is displayed when the web page that a user is attempting to access is determined to be inappropriate or fraudulent, and serves to warn the user.

[1541] An "emotion engine" is a device or software that analyzes a user's emotional state and determines the emotional state based on the user's facial expressions, voice, text input, etc.

[1542] "Mental stress" refers to the psychological burden or state of tension caused by external stimuli or information.

[1543] "Content suggestion" is the process of recommending suitable content, such as videos or music with a relaxing effect, taking into account the user's current emotional state and mental health.

[1544] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining an emotion engine that recognizes the user's emotions, the present invention can monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[1545] The system's hardware includes internet-connected devices (e.g., smartphones, tablets, PCs, etc.), and its software includes generative AI models (e.g., the BERT model using NLP techniques and the YOLOv5 model using computer vision techniques) and emotion engines (e.g., the OpenFace library).

[1546] When a user attempts to access a specific URL using an Internet-connected device (hereinafter referred to as a terminal), the access request is sent to a server. The server makes an HTTP request to the specified URL and retrieves the web page content. The retrieved content is in HTML format, and web scraping technology (such as Beautiful Soup or Selenium) is used to separate it into text data and image data.

[1547] Text data is analyzed using generative AI models, which leverage NLP techniques to detect harmful keywords and contexts. A specific example is the BERT model, which uses the Transformers library. Similarly, image data is analyzed using computer vision techniques, such as YOLOv5 and OpenCV, to detect inappropriate images.

[1548] To recognize the user's emotional state, the emotion engine collects data from the device's camera, microphone, and text input data. The emotion engine (e.g., the OpenFace library) analyzes the collected data and determines the user's current emotional state from their facial expressions and voice.

[1549] If the server detects inappropriate or fraudulent content based on the analysis results of the generative AI model and the emotion engine, it blocks access to the content and sends a warning page to the device. If it detects content that may cause mental stress, it combines the results of the emotion engine to block access and display a warning page. Furthermore, it suggests appropriate content (e.g., videos or music with a relaxing effect) based on the user's emotional state.

[1550] As a concrete example, in the case of access to inappropriate content, an underage user attempts to access a website containing adult content. The access request from the device is sent to the server, which retrieves and analyzes the content of the URL. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1551] An example prompt might be, "Using a generative AI model, analyze the content of the specified URL to detect inappropriate or deceptive content. Also, consider the results of the user's sentiment engine to suggest appropriate content for the user."

[1552] The above is a detailed description of a specific embodiment of the present invention.

[1553] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1554] Step 1:

[1555] When a user attempts to access a specific URL, the device sends the access request to the server. The input is the URL entered by the user into the device, and the output is an HTTP request to the server. Specifically, a web browser or a dedicated application sends the URL to the server.

[1556] Step 2:

[1557] Based on the received URL access request, the server makes an HTTP request to the specified URL and retrieves the web page content. The input is the URL sent from the terminal, and the output is the web page content in HTML format. Specifically, the server communicates with the web server using the HTTP protocol and downloads the web page.

[1558] Step 3:

[1559] The server separates the retrieved HTML content into text data and image data. This process uses web scraping technology (e.g., Beautiful Soup or Selenium). The input is the web page content in HTML format, and the output is text data and image data. Specifically, the HTML parser analyzes the HTML tags and extracts the text and images.

[1560] Step 4:

[1561] The server analyzes the text data using a generative AI model. Here, natural language processing technology (e.g., the BERT model) is used to analyze the text content and detect harmful keywords and context. The input is the separated text data, and the output is an evaluation score for the text as the analysis result. Specifically, the text data is input into the generative model, and natural language processing is performed to evaluate the harmfulness.

[1562] Step 5:

[1563] The server analyzes image data using a generative AI model. Here, computer vision techniques (e.g., YOLOv5 or OpenCV) are used to detect inappropriate images. The input is the separated image data, and the output is an evaluation score for the image as the analysis result. Specifically, the image data is input into the generative model, and a computer vision algorithm is applied to evaluate its harmfulness.

[1564] Step 6:

[1565] The device collects camera, microphone, and text input data to recognize the user's emotional state. The input is the user's facial expression, voice, and text input data, and the output is to send this data to the server. Specific operations include the process in which the device's sensors capture data and send it to the server.

[1566] Step 7:

[1567] The server uses an emotion engine to analyze the collected data and determine the user's emotional state. Here, the analysis is performed using emotion recognition technology (e.g., the OpenFace library). The input is facial expressions, voice, and text data sent from the device, and the output is an evaluation result of the user's emotional state. Specifically, the emotion engine processes the data and analyzes the emotional state in real time.

[1568] Step 8:

[1569] If the server detects inappropriate or deceptive content based on the analysis results of the generative AI model and emotion engine, it blocks access to that content. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is an instruction to block access and the generation of a warning page. Specifically, the server executes the access control logic and generates a warning page if necessary.

[1570] Step 9:

[1571] If the server detects content that may cause mental stress, it combines the results of the emotion engine to block access and send a warning page to the terminal. At the same time, it suggests appropriate content based on the user's emotional state. The input is the results of text and image analysis and the evaluation of the emotional state, and the output is a list of suggested content. Specifically, the server runs the content suggestion algorithm to generate content appropriate for the user.

[1572] Step 10:

[1573] The terminal receives the analysis results, warning pages, and suggested content from the server and displays them to the user. The input is the warning page and suggested content sent from the server, and the output is the screen displayed to the user. Specifically, the terminal updates the screen according to instructions from the server and provides the necessary information to the user.

[1574] (Application example 2)

[1575] 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."

[1576] With the spread of the Internet, users have more opportunities to access various websites, many of which contain inappropriate or fraudulent content. Access to content that may cause mental stress is also increasing. Encountering such content can harm users' mental health, so a system that automatically filters this content and protects users' health is needed.

[1577] 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 a website access request from an Internet-connected device, means for acquiring content from a specified URL, means for analyzing the text and images of the content using a generative model, means for determining inappropriate or fraudulent content based on the analysis results of the generative model, means for blocking the inappropriate or fraudulent content and displaying a warning page, means for recognizing emotions from the user's facial expression or voice, means for determining whether the recognized emotion is a stress state, and means for suggesting appropriate content according to the stress state. This not only protects users from inappropriate or fraudulent content when browsing websites, but also makes it possible to provide appropriate content to reduce mental stress.

[1578] "Internet-connected device" refers to any electronic device that can connect to the Internet.

[1579] "Website Access Request" refers to a request by a user to access a particular website.

[1580] "Specified URL" refers to the specific web address that the user attempted to access.

[1581] "Content" refers to information such as text data and image data displayed on a website.

[1582] A "generative model" refers to algorithms or software for generating and analyzing data based on artificial intelligence techniques.

[1583] "Means for analyzing text and images" refers to methods for evaluating and analyzing text data and image data within the Content.

[1584] "Inappropriate Content" refers to information on a website that contains material that is deemed harmful to users.

[1585] "Deceptive content" refers to false information or pages created with the intent to deceive users.

[1586] "Blocking measures" refers to methods of controlling users' access to content that is deemed inappropriate or fraudulent.

[1587] "Warning Page" refers to a notification page that informs users that certain content is inappropriate or deceptive.

[1588] "Means for recognizing emotions from a user's facial expression or voice" refers to technology that analyzes and determines a user's emotional state from their facial expression or voice.

[1589] A "stressed state" refers to a state in which the user feels mental tension or anxiety.

[1590] The "means for suggesting appropriate content" refers to a method for recommending content that has a relaxing effect based on the user's emotional state.

[1591] The present invention is a filtering system that receives website access requests from Internet-connected devices, retrieves the content of the specified URL, analyzes the text and images of that content using a generative model, blocks inappropriate or fraudulent content, and restricts access to content that may cause mental stress. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to monitor the user's mental health in real time and suggest appropriate content according to the user's specific emotional state.

[1592] System Configuration

[1593] Server Configuration

[1594] The server requires the following hardware and software:

[1595] Hardware: A server with a fast processor and large memory capacity.

[1596] software:

[1597] Generative model: OpenAI GPT-4

[1598] Computer Vision Technology: Google Vision API

[1599] Emotion engine: Affectiva SDK

[1600] Programming languages ​​and frameworks: Python, Django, Nginx

[1601] User's device

[1602] The user's device must have the following features:

[1603] Camera and microphone

[1604] Internet connection function

[1605] Dedicated applications

[1606] Program processing

[1607] Server Processing

[1608] The server first receives a URL access request sent from the user's device. It then makes an HTTP request to the specified URL to retrieve the webpage content. The retrieved content is separated into text data and image data and analyzed using a generative model (GPT-4) and computer vision technology (Google Vision API). If inappropriate or fraudulent content is detected, the server blocks access to the content and sends a warning page to the user's device.

[1609] The server also uses an emotion engine (Affectiva SDK) to recognize the user's emotions in real time. It determines the user's emotional state based on facial expressions, voice, and text input, and if it determines that the user is particularly stressed, it suggests content that will have a relaxing effect.

[1610] Terminal handling

[1611] The user's device sends a request to the server when they attempt to access a specific URL, and displays appropriate content or a warning page based on the analysis results.The camera and microphone are used to provide data to the emotion engine, which monitors the user's emotional state in real time.

[1612] Specific examples

[1613] Example 1: Accessing inappropriate content

[1614] An underage user attempts to access a website containing adult content using a browser. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is inappropriate, the server blocks the access and sends a warning page to the device. The device then displays the warning page to the user, preventing access to the adult content.

[1615] Example 2: Visiting a fraudulent website

[1616] An elderly person clicks on a link in an email and attempts to access a financial fraud website. The access request from the device is sent to the server, which retrieves and analyzes the URL content. If the generative model determines that the content is fraudulent, the server blocks the access and displays a warning page on the device. In this way, access to the fraudulent website is prevented.

[1617] Example 3: Visiting a stressful news site

[1618] A user may attempt to access a news site, but the content of that site may cause mental stress. The server analyzes the text and image content and determines that it may cause mental stress. Furthermore, the emotion engine monitors the user's emotional state. If it determines that the user is in a stressful state, the server blocks the access and sends a warning page to the user's device. The device displays the warning page to the user, protecting them from stress. At the same time, it suggests content with a relaxing effect to support the user's mental health.

[1619] Prompt Sentence Examples

[1620] "Emotion-aware filtering app. A user is about to visit a new news site. Check if the site contains harmful content and display a warning page if necessary. Also, if the user is under stress, suggest relaxation music."

[1621] The above is a specific description of the embodiment of the present invention. This system protects users from inappropriate or fraudulent content and reduces mental stress.

[1622] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1623] Step 1:

[1624] When a user tries to access a specific URL, the device generates a website access request and sends it to the server. The input is the URL specified by the user, and the output is the access request sent to the server.

[1625] Step 2:

[1626] The server receives a URL access request sent from the user's terminal. The input is the access request, and the output is a confirmation of the request. Based on this request, the server makes an HTTP request to the specified URL.

[1627] Step 3:

[1628] The server uses an HTTP request to retrieve the content of a specified URL. The input is the specified URL, and the output is the retrieved web page content (text data and image data).

[1629] Step 4:

[1630] The server separates the retrieved content into text data and image data. The input of this step is the retrieved web page content, and the output is the separated text data and image data. The server analyzes the text content using a generative model (GPT-4) and analyzes the image content using computer vision technology (Google Vision API).

[1631] Step 5:

[1632] The server determines whether content is inappropriate or fraudulent based on the analysis results of the generative model. The input is the analyzed text data and image data, and the output is the judgment result. If inappropriate or fraudulent content is detected, the server proceeds to the next step based on the judgment result.

[1633] Step 6:

[1634] If the server determines that the content is inappropriate or fraudulent, it blocks access to it and sends a warning page to the terminal. The input of this step is the result of the determination, and the output is a block instruction and the sending of a warning page.

[1635] Step 7:

[1636] The terminal receives the warning page sent from the server and displays it to the user. The input is the warning page, and the output is the display of a warning message to the user. This warning prevents the user from accessing inappropriate content or fraudulent sites.

[1637] Step 8:

[1638] The server uses an emotion engine (Affectiva SDK) to recognize emotions from the user's facial expressions or voice in real time. The input is facial expression data or voice data sent from the device, and the output is the user's emotional state.

[1639] Step 9:

[1640] The server determines whether the recognized emotion is a stress state. The input is the analysis result of the emotion engine, and the output is the stress state determination result.

[1641] Step 10:

[1642] If the server determines that the user is in a stressful state, it proposes appropriate content. The input to this step is the result of the stress state determination, and the output is a recommendation of content that has a relaxing effect. This appropriate content proposal is sent to the user's device.

[1643] The above is the specific processing flow of the system that realizes this application example. This system protects users from inappropriate or fraudulent content when browsing websites, thereby reducing mental stress.

[1644] 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.

[1645] 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.

[1646] 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.

[1647] 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.

[1648] 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.

[1649] 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.

[1650] 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).

[1651] 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.

[1652] 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."

[1653] 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.

[1654] 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).

[1655] 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.

[1656] 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.

[1657] 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.

[1658] 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.

[1659] 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.

[1660] 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.

[1661] 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.

[1662] 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.

[1663] 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.

[1664] 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.

[1665] The following is further disclosed regarding the above embodiment.

[1666] (Claim 1)

[1667] means for receiving a website access request from an internet connected device;

[1668] A means of retrieving the content of a specified URL;

[1669] means for analyzing the text and images of the content using a generative model;

[1670] a means for determining inappropriate or deceptive content based on the analysis results of the generative model;

[1671] A means to block inappropriate or deceptive content and display a warning page;

[1672] A system including:

[1673] (Claim 2)

[1674] 10. The system of claim 1, further comprising means for detecting content that may cause psychological stress and restricting access to that content.

[1675] (Claim 3)

[1676] 10. The system of claim 1, wherein the text analysis means uses natural language processing techniques and the image analysis means uses computer vision techniques.

[1677] "Example 1"

[1678] (Claim 1)

[1679] means for receiving a website access request sent from a terminal;

[1680] A means for retrieving the content of the requested URL;

[1681] A means for separating the acquired content into text data and image data;

[1682] A means of analyzing text data using a generative AI model to detect harmful keywords and expressions;

[1683] means for analyzing image data using computer vision technology to detect inappropriate images;

[1684] A means for determining inappropriate or deceptive content based on the analysis of the generative AI model; and

[1685] A means to block inappropriate or fraudulent content and send a warning page to the device;

[1686] A system including:

[1687] (Claim 2)

[1688] 10. The system of claim 1, further comprising means for detecting content that may cause psychological stress and restricting access to that content.

[1689] (Claim 3)

[1690] 10. The system of claim 1, wherein the analysis of the text data uses natural language processing techniques and the analysis of the image data uses computer vision techniques.

[1691] "Application Example 1"

[1692] (Claim 1)

[1693] means for receiving a website access request from an internet-connected device;

[1694] A means of retrieving data from a specified URL;

[1695] means for analyzing the text and images of the data using a generative model;

[1696] A means for determining inappropriate or fraudulent data based on the analysis results of the generative model;

[1697] A means to block inappropriate or fraudulent data and display a warning page;

[1698] A means to scan data in real time and monitor users' internet usage;

[1699] A system including:

[1700] (Claim 2)

[1701] 10. The system of claim 1, further comprising means for detecting data that may cause psychological stress and restricting access to the data.

[1702] (Claim 3)

[1703] 10. The system of claim 1, wherein the text analysis means uses natural language processing techniques and the image analysis means uses computer vision techniques to perform analysis in real time based on the access request.

[1704] "Example 2: Combining Emotion Engines"

[1705] (Claim 1)

[1706] means for receiving a website access request from an internet connected device;

[1707] A means of retrieving the content of a specified URL;

[1708] means for analyzing the text and images of the content using a generative model;

[1709] a means for determining inappropriate or deceptive content based on the analysis results of the generative model;

[1710] A means to block inappropriate or deceptive content and display a warning page;

[1711] means for collecting data for recognizing an emotional state of a user;

[1712] means for analyzing the collected data using an emotion engine to determine the user's emotional state;

[1713] A system including:

[1714] (Claim 2)

[1715] 10. The system of claim 1, further comprising means for detecting content that may cause mental stress and restricting access to the content, and means for suggesting content that takes into consideration the user's mental health.

[1716] (Claim 3)

[1717] 10. The system of claim 1, wherein the text analysis means uses natural language processing techniques and the image analysis means uses computer vision techniques.

[1718] "Application example 2 when combining emotion engines"

[1719] (Claim 1)

[1720] means for receiving a website access request from an internet connected device;

[1721] A means of retrieving the content of a specified URL;

[1722] means for analyzing the text and images of the content using a generative model;

[1723] a means for determining inappropriate or deceptive content based on the analysis results of the generative model;

[1724] A means to block inappropriate or deceptive content and display a warning page;

[1725] means for recognizing emotions from a user's facial expression or voice;

[1726] a means for determining whether the recognized emotion is a stress state;

[1727] A method to suggest appropriate content according to the stress level,

[1728] A system including:

[1729] (Claim 2)

[1730] 10. The system of claim 1, further comprising means for detecting content that may cause psychological stress and restricting access to that content.

[1731] (Claim 3)

[1732] 10. The system of claim 1, wherein the text analysis means uses natural language processing techniques, the image analysis means uses computer vision techniques, and the emotion recognition means uses techniques that analyze a user's facial expressions, voice, or text input data. [Explanation of symbols]

[1733] 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 a website access request from an internet connected device; A means of retrieving the content of a specified URL; means for analyzing the text and images of the content using a generative model; a means for determining inappropriate or deceptive content based on the analysis results of the generative model; A means to block inappropriate or deceptive content and display a warning page; A system including:

2. The system of claim 1 further comprising means for detecting content that may cause psychological stress and restricting access to that content.

3. The system of claim 1 , wherein the text analysis means uses natural language processing techniques and the image analysis means uses computer vision techniques.

Images