system

Abstract

A system is provided.SOLUTION: A system comprising: means for receiving a mail body or a URL; means for generating a request for analyzing the mail body or the URL; means for transmitting the request to a AI model to analyze a possibility of scam; and means for notifying a user of a warning based on a result of the analysis.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] Fraudulent activities spread via the Internet and email are becoming increasingly sophisticated, making it difficult for users to identify these fraudulent activities on their own. This issue is a major concern in preventing damage caused by fraudulent emails and websites. The problem that this invention aims to solve is to automatically detect these sophisticated fraudulent activities and warn users, thereby reducing the risk of fraud and fraudulent use. [Means for solving the problem]

[0005] The present invention provides a system for an email application or web browsing application, including a means for a terminal to receive an email body or URL, a means for the terminal to generate a request for analyzing the email body or URL, a means for transmitting the request to an AI model and analyzing the possibility of fraud, and a means for issuing a warning to a user based on the analysis results. The AI ​​model has the function of analyzing the content of the email body using natural language processing technology, detecting suspicious phrases, links, and sender information, and generating a fraud likelihood score. The content of the warning message displayed to the user is determined based on this score, allowing the user to detect the risk of fraud in advance and take appropriate action.

[0006] "Terminal" refers to an electronic device for running an email application or a web browsing application, such as a computer or smartphone operated by a user.

[0007] The "mail body" refers to the content of an email, and is the text that contains the message or information that the sender wants to convey to the recipient of the email.

[0008] "URL" stands for Uniform Resource Locator and refers to an address used to identify a web page or resource on the Internet.

[0009] An "analysis request" is an instruction generated by a device to analyze an email body or URL, including a data structure and commands sent to an AI model.

[0010] An "AI model" refers to a system that uses artificial intelligence technology to automatically perform specific tasks, and in this invention in particular is a model that uses natural language processing technology to analyze email content and URLs.

[0011] "Natural language processing technology" refers to technology that allows computers to understand, interpret, and generate natural language, and is a computational technology used for text analysis and semantic understanding.

[0012] "Suspicious phrases" refer to text or expressions that are likely to be associated with fraudulent activity, typically keywords or phrases commonly found in phishing or spam emails.

[0013] The "score" refers to a numerical evaluation value generated by an AI model, which quantitatively indicates the likelihood that the subject of the analysis is fraudulent.

[0014] A "warning message" is a notification that the terminal displays to the user, and is a message that contains content that warns or calls the user's attention based on the analysis results. [Brief explanation of the drawings]

[0015] [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

[0016] 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.

[0017] First, the terms used in the following description will be explained.

[0018] 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).

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 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.

[0026] 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).

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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."

[0036] The present invention provides a system for automatically detecting fraudulent emails and websites in email applications and web browsing applications, and for issuing warnings to users. The processing of a program for implementing this system is specifically explained below in natural language.

[0037] System Overview

[0038] The system is embedded within email or web browsing applications and analyzes incoming emails and visited URLs to determine potential fraudulent activity using AI models running on the device.

[0039] Example of an email application

[0040] 1. Receiving emails and notifications

[0041] When a user starts an email application, the server receives new emails and transfers them to the terminal.

[0042] The terminal notifies the user that new mail has arrived.

[0043] 2. Generate email analysis request

[0044] When a user opens a new email, the device extracts the email body and header information, including the sender's email address, subject, body content, links, and attachment information.

[0045] The device generates an analysis request to the AI ​​model based on this information.

[0046] 3. Analysis using AI models

[0047] The device sends the analysis request generated to the AI ​​model.

[0048] The AI ​​model analyzes the email body and links, and uses natural language processing techniques to examine the content, comparing it with known phishing and spam email characteristics to detect suspicious phrases and links.

[0049] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[0050] 4. Receiving analysis results and generating alerts

[0051] The device receives the analysis results from the AI ​​model.

[0052] Based on the analysis results, the device will issue a warning to the user. For example, if the analysis results indicate a high probability of fraud, the device will display a warning message saying, "This email may be fraudulent."

[0053] 5. User Response

[0054] Users should acknowledge the warning and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0055] Example of a web browsing application

[0056] 1. Accessing the URL

[0057] A user clicks on a new URL in a web browsing application.

[0058] The device extracts the URL.

[0059] 2. URL parsing request generation

[0060] The device generates an analysis request to send the URL information to the AI ​​model.

[0061] 3. Analysis using AI models

[0062] The device sends the analysis request generated to the AI ​​model.

[0063] The AI ​​model analyzes the URL destination (web page content) and compares it with known phishing sites and malicious domains.

[0064] 4. Receiving analysis results and generating alerts

[0065] If the device receives the analysis results from the AI ​​model and determines that there is a high possibility of fraud, the device will display a warning saying, "This site may be fraudulent."

[0066] 5. User Response

[0067] Users should check the warning and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0068] In this way, the present invention is a system that automatically detects fraud risks in email and web browsing applications and warns users, preventing fraud damage before it occurs. In particular, by using an AI model, it achieves more accurate analysis than conventional simple filtering.

[0069] The processing flow will be explained below.

[0070] Step 1:

[0071] The server receives new emails and forwards them to the device, which causes the email application to recognize that new emails have arrived.

[0072] Step 2:

[0073] The device will notify the user of new emails. A notification will pop up to let you know that new email has arrived.

[0074] Step 3:

[0075] The user clicks to open a new email, which starts the process of displaying the email contents.

[0076] Step 4:

[0077] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[0078] Step 5:

[0079] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[0080] Step 6:

[0081] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[0082] Step 7:

[0083] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[0084] Step 8:

[0085] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[0086] Step 9:

[0087] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[0088] Step 10:

[0089] The device receives the analysis results and score and generates a warning to the user based on the results. For example, if there is a high possibility of fraud, it generates a warning message such as "This email may be fraudulent."

[0090] Step 11:

[0091] The device displays a warning message to the user. A message is displayed on the user interface to inform the user of the danger.

[0092] Step 12:

[0093] Users can prevent themselves from falling victim to fraud by checking the warning message and taking action such as deleting the email or not clicking on the link.

[0094] In this way, by performing specific processing at each step, a system is realized that can reliably detect fraudulent emails and warn users.

[0095] Example 1

[0096] 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."

[0097] Conventional email and web browsing applications have difficulty detecting fraud risks and providing appropriate warnings to users. This puts users at high risk of being exposed to phishing emails and fraudulent websites. The present invention aims to reduce such risks and enable users to safely communicate electronically and browse the web.

[0098] 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.

[0099] In this invention, the server includes means for a terminal to receive data related to electronic communications or websites, means for the terminal to generate a request for analyzing the data, means for transmitting the request to a machine learning model to analyze the possibility of fraud, means for notifying a user of a warning based on the analysis result, and means for guiding the user's operation based on the warning, thereby enabling the user to quickly recognize the risk of fraud and take appropriate measures.

[0100] "Device" refers to a hardware device used for electronic communication or website access, including, for example, a computer, smartphone, or tablet.

[0101] "Electronic Communications" means information such as text, images, links, etc. sent or received via email or messaging applications.

[0102] "Website" means a collection of web pages publicly available on the Internet and is a resource of information accessible to users.

[0103] "Data" refers to any information received from an electronic communication or website, including text, images, links, and metadata.

[0104] A "machine learning model" is an algorithm that analyzes large amounts of data, learns patterns and features from that data, and predicts and classifies new data.

[0105] "Natural language processing technology" is a general term for technologies that allow computers to understand, generate, and analyze human language, and includes text analysis, translation, and dialogue systems.

[0106] "Analysis Request" refers to a packet of information sent to a machine learning model to analyze data.

[0107] "Fraud" refers to malicious acts such as phishing emails and fraudulent websites that attempt to illegally obtain users' personal information or money.

[0108] "Warning" refers to a message that alerts the user based on the analysis results and provides information to help avoid risky behavior.

[0109] "Guiding user operations" means showing the user appropriate actions to take based on the warning, such as deleting the email or avoiding clicking on links.

[0110] The present invention provides a system for automatically detecting fraudulent emails and fraudulent websites in email applications and web browsing applications, and for issuing a warning to the user. A detailed description of how to implement the system of the present invention is provided below.

[0111] Hardware and software used

[0112] Hardware: Devices (computers, smartphones, tablets, etc.)

[0113] software:

[0114] Email application

[0115] Web browsing applications

[0116] Machine learning models (e.g., the BERT model, which is specialized for natural language processing)

[0117] System Overview

[0118] The system is embedded within email or web browsing applications and uses machine learning models running on the device to analyze emails received and website URLs visited in real time to identify potential fraudulent activity.

[0119] Example from an email application

[0120] Receiving emails and notifications

[0121] The server receives the user's email and forwards new emails to the terminal.

[0122] The device will notify the user of new emails by sounding an alert and displaying a pop-up message.

[0123] Generate email analysis request

[0124] When a user opens an email application and clicks on a new email, the device extracts the email body and header information.

[0125] Based on the information extracted by the device, an analysis request is generated for the machine learning model.

[0126] Analysis using machine learning models

[0127] Sends device-generated analysis requests to machine learning models.

[0128] Machine learning models analyze email text and links, and use natural language processing techniques to inspect the content.

[0129] The model matches the characteristics of known phishing and spam emails to detect suspicious phrases and links.

[0130] The analysis generates a score indicating the likelihood of fraud.

[0131] Receiving analysis results and generating alerts

[0132] The device receives the analysis results and issues a warning to the user. For example, if it determines that the email is likely to be fraudulent, it displays a warning message saying, "This email may be fraudulent."

[0133] User response

[0134] Users acknowledge the warning and take action to reduce the risk of fraud, such as deleting the email or not clicking on the link.

[0135] Example of a web browsing application

[0136] URL access

[0137] When a user clicks on a new URL in a web browsing application, the device extracts the URL.

[0138] URL parsing request generation

[0139] The device generates an analysis request to send the URL information to the machine learning model.

[0140] Analysis using machine learning models

[0141] Sends device-generated analysis requests to machine learning models.

[0142] Machine learning models analyze the URL destination (web page content) and compare it with known phishing sites and malicious domains.

[0143] Receiving analysis results and generating alerts

[0144] If the device receives the analysis results and determines that there is a high possibility of fraud, it will display a warning message saying, "This site may be fraudulent."

[0145] User response

[0146] Users can acknowledge the warning and take action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0147] Examples of specific examples and prompts

[0148] Example of an email application

[0149] When a user opens a new email containing an "Unsubscribe" link, a machine learning model identifies the link as a phishing risk and displays a warning.

[0150] Prompt Sentence Examples

[0151] Please check the safety of this email before clicking any links, as it is likely a scam.

[0152] Example of a web browsing application

[0153] When a user attempts to visit a specific online shopping site, a machine learning model identifies the site as a known fraudulent site and displays a warning.

[0154] Prompt Sentence Examples

[0155] This site may be a scam. Please consider other safe ways to purchase.

[0156] This invention realizes a system that can instantly detect fraudulent behavior during email and web browsing, allowing users to avoid the risks of phishing emails and fraudulent sites. In particular, by using machine learning models, it is possible to perform more accurate analysis than conventional methods.

[0157] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0158] (Processing flow in email application)

[0159] Step 1:

[0160] The server receives the user's email and forwards new emails to the device.

[0161] Input: User's email account information, new emails

[0162] Output: New email notification to device

[0163] Specific operation: Records emails from various mail servers (IMAP / POP3) in the receiving log and forwards them to the mail application.

[0164] Step 2:

[0165] The device will notify the user that new email has arrived.

[0166] Input: New email

[0167] Output: New mail notification to user

[0168] Specific operation: On the smartphone, the message "You have new email" will appear in the notification bar and a notification sound will be heard on the device itself.

[0169] Step 3:

[0170] The user opens their email application and clicks on new email.

[0171] The device extracts the email body and header information.

[0172] Input: New emails opened by the user, including the email body and header information

[0173] Output: Extracted email body and header information

[0174] What happens: The sender address, subject, body, links, and attachment information are separated for analysis.

[0175] Step 4:

[0176] Based on the information extracted by the device, an analysis request is generated for the AI ​​model.

[0177] Input: Extracted email body and header information

[0178] Output: Analysis request to the AI ​​model

[0179] Specific operation: The analysis request is structured in JSON format or other format and prepared for sending to the machine learning model.

[0180] Step 5:

[0181] The device generates an analysis request and sends it to the AI ​​model.

[0182] Input: Analysis request to the AI ​​model

[0183] Output: Sending data to the AI ​​model

[0184] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[0185] Step 6:

[0186] The AI ​​model analyzes the email body and links, and uses natural language processing technology to inspect the content.

[0187] Input: Analysis request email body and link

[0188] Output: Analysis results (fraud likelihood score)

[0189] What it does: Contextual analysis, spam filtering, and matching links to known phishing phrases.

[0190] Step 7:

[0191] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[0192] Input: Parsed email body and links

[0193] Output: Analysis results with fraud probability score

[0194] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[0195] Step 8:

[0196] The device receives the analysis results from the AI ​​model.

[0197] Input: Analysis results with fraud likelihood score

[0198] Output: A warning notice to the user

[0199] Specific Actions: Based on the estimated risk, a warning message is prepared and displayed in the user interface.

[0200] Step 9:

[0201] The device will display a warning to the user.

[0202] Input: warning message

[0203] Output: A warning notice to alert the user

[0204] Specific behavior: If the email is rated as "likely fraudulent," a pop-up message will appear saying "This email may be fraudulent."

[0205] Step 10:

[0206] Users should acknowledge the warning and take action to avoid the risk of fraud.

[0207] Input: warning message

[0208] Output: User response (delete email, avoid clicking links)

[0209] Specific action: The user clicks the "Delete" button to delete the email.

[0210] (Processing flow in a web browsing application)

[0211] Step 1:

[0212] A user clicks on a new URL in their web browsing application.

[0213] Input: User clicks on a URL

[0214] Output: Extracted URLs

[0215] What it does: The browser captures the click event and records the URL being accessed.

[0216] Step 2:

[0217] The device will extract the URL.

[0218] Input: Clicked URL

[0219] Output: URL for analysis

[0220] What it does: Saves the URL as structured data.

[0221] Step 3:

[0222] The device generates an analysis request to send the URL information to the AI ​​model.

[0223] Input: A structured URL to parse

[0224] Output: Analysis request to the AI ​​model

[0225] Specific operation: Structure the analysis request in JSON format or similar.

[0226] Step 4:

[0227] The device generates an analysis request and sends it to the AI ​​model.

[0228] Input: Analysis request to the AI ​​model

[0229] Output: Sending data to the AI ​​model

[0230] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[0231] Step 5:

[0232] The AI ​​model analyzes the URL destination (webpage content) and compares it with known phishing sites and malicious domains.

[0233] Input: URL of the analysis request

[0234] Output: Analysis results

[0235] What it does: Scrape the content of web pages and compare it to a database of known phishing sites.

[0236] Step 6:

[0237] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[0238] Input: scraped web page content

[0239] Output: Analysis results with fraud probability score

[0240] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[0241] Step 7:

[0242] The device receives the analysis results and notifies the user with a warning.

[0243] Input: Analysis results with fraud likelihood score

[0244] Output: A warning notice to the user

[0245] Specific behavior: If there is a high possibility of fraud, a warning message will be displayed in a pop-up saying, "This site may be fraudulent."

[0246] Step 8:

[0247] Users should acknowledge the warning and take action to avoid the risk of fraud.

[0248] Input: warning message

[0249] Output: User response (stops visiting the site)

[0250] Specific Action: The user closes the browser tab or stops visiting the site.

[0251] (Application example 1)

[0252] 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."

[0253] In today's world, as internet-based communications become increasingly common, the number of victims of fraudulent emails and phishing sites is rapidly increasing. These fraudulent activities often result in the leakage of personal information and financial losses, posing a significant risk to many users. Current technology makes it difficult to detect these fraudulent activities in advance and implement countermeasures, creating a demand for systems with advanced analytical capabilities. The purpose of this invention is to solve these problems and enable users to use email and websites safely.

[0254] 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.

[0255] In this invention, the server includes: a terminal receiving an email body or URL; a terminal generating a request for analyzing the email body or URL; a terminal sending the request to an AI model to analyze the possibility of fraud; a terminal issuing a warning to the user based on the analysis results; a smartphone installed on the smartphone for analyzing emails and websites in real time; and a terminal displaying a warning if the possibility of fraud is high based on the analysis results. This enables sophisticated analysis of fraudulent emails and phishing sites and displaying a warning to the user in real time.

[0256] A "terminal" is an electronic device such as a smartphone, tablet, or PC that is operated by a user.

[0257] "Email body" refers to the body of the message written by the sender of the email, which may contain text and links.

[0258] "URL" stands for Uniform Resource Locator and indicates the address of a web page on the Internet.

[0259] An "analysis request" is request data generated to allow an AI model to determine the possibility of fraud based on the content of an email body or URL.

[0260] An "AI model" is an analytical engine that utilizes artificial intelligence technology and is designed to detect fraudulent emails and phishing sites.

[0261] "Natural language processing" is a technique in which AI models understand and analyze human language, and is used to scrutinize the content of emails and web pages.

[0262] A "warning message" is a notification message that is displayed to warn the user of possible fraud based on the analysis results.

[0263] "Real-time" refers to the instantaneous reception and analysis of data, allowing users to obtain results immediately.

[0264] This invention is a system that automatically detects fraudulent emails and websites in email and web browsing applications and issues a warning to the user. This system is mainly composed of a terminal and a server, and the terminal receives emails and URLs, analyzes their contents, and notifies the user.

[0265] Program processing explanation

[0266] The server first receives the email body or URL sent from the device. The email body includes the sender address, subject, body text, links, etc. The server generates a request to analyze this content and sends it to an AI model. This AI model uses natural language processing (NLP) technology and is designed to detect potential fraud and phishing. Specifically, it analyzes the content by comparing it with keywords and phrases commonly found in fraudulent emails, as well as information about the domains to which the links lead.

[0267] As a result of the analysis, the AI ​​model assigns a score indicating the likelihood of fraud and sends it back to the server. The server generates and notifies the user of a warning message based on this score. For example, if the score is high, a warning such as "This email may be fraudulent" or "This site may be fraudulent" will be displayed.

[0268] The system of the present invention mainly uses the following hardware and software. The terminal is a general computing device such as a smartphone or tablet. The server can be a cloud server or a dedicated server in a data center. In addition, the AI ​​model uses a deep learning framework such as TensorFlow or PyTorch.

[0269] Specific examples

[0270] For example, when a user receives an email disguised as a message from a bank, the body of the email often contains keywords such as "transfer" and "password change." When the user opens the email, the device sends the email body to a server, where the AI ​​model analyzes these keywords. If the analysis returns a score of "very likely to be fraudulent," the user is shown a warning message saying, "This email may be fraudulent."

[0271] Prompt Sentence Examples

[0272] 1. Enter the sender, subject, and body of the email you received and assess its likelihood of fraud.

[0273] 2. "Analyze the URL destination to determine whether it is phishing."

[0274] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0275] Step 1:

[0276] The device receives the email or URL.

[0277] What it does: The device receives new emails or website URLs via applications the user has opened. For emails, information such as the sender, subject, body, and links is extracted.

[0278] Input: Received email body or URL

[0279] Output: Extracted email and URL information

[0280] Step 2:

[0281] The device generates an analysis request.

[0282] What it does: Based on the information in the received email or URL, it creates an analysis request to send to the AI ​​model. This request includes the text of the email body and the URL information.

[0283] Input: Extracted email and URL information

[0284] Output: Analysis request (email body text, URL information, etc.)

[0285] Step 3:

[0286] The device sends an analysis request to the server.

[0287] Specific operation: The generated analysis request is sent to the server, which receives it and prepares to pass the request to the AI ​​model.

[0288] Input: Parse request

[0289] Output: The parsing request sent to the server

[0290] Step 4:

[0291] The server sends an analysis request to the AI ​​model and begins the analysis.

[0292] How it works: When the server receives an analysis request, it passes it to the AI ​​model, which uses natural language processing techniques to analyze the content of the email body and URLs to detect potential fraud.

[0293] Input: Parse request

[0294] Output: Analysis results (e.g., fraud probability score)

[0295] Step 5:

[0296] The server receives the analysis results and sends them to the terminal.

[0297] How it works: The server receives the analysis results returned by the AI ​​model and sends them to the device, which includes a score indicating the likelihood of fraud.

[0298] Input: Analysis results (e.g., fraud likelihood score)

[0299] Output: Analysis results sent to the device

[0300] Step 6:

[0301] The device notifies the user of a warning based on the analysis results.

[0302] Specific operation: When the device receives the analysis results, if there is a high possibility of fraud, it will display a warning message to the user. The warning message will include content such as "This email may be fraudulent" or "This site may be fraudulent."

[0303] Input: Analysis result (fraud likelihood score)

[0304] Output: The warning message displayed to the user.

[0305] 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.

[0306] This invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. It also incorporates an emotion engine to understand the user's emotional state and take appropriate action. Below, we will explain in detail the processing of the program for implementing this system.

[0307] System Overview

[0308] This system is embedded in email or web browsing applications, analyzes incoming emails and accessed URLs, and determines whether fraud is likely. It also uses an emotion engine to recognize the user's emotional state and adjust the display method and content of warning messages accordingly.

[0309] Example of an email application

[0310] 1. Receiving emails and notifications

[0311] The server receives new emails and forwards them to the terminal.

[0312] The terminal notifies the user that new mail has arrived.

[0313] 2. Generate email analysis request

[0314] The user clicks to open a new email.

[0315] The terminal extracts the email body and email header.

[0316] The device generates an analysis request to the AI ​​model based on the information extracted.

[0317] 3. Analysis using AI models

[0318] The device sends the analysis request generated to the AI ​​model.

[0319] The AI ​​model analyzes the email text and links, and uses natural language processing techniques to detect suspicious phrases and links.

[0320] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[0321] 4. Emotion Recognition by Emotion Engine

[0322] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[0323] An emotion engine analyzes this data to determine the user's emotional state.

[0324] 5. Integration of analysis results and emotion recognition results

[0325] The device receives the analysis results from the AI ​​model and the emotion recognition results from the emotion engine.

[0326] Based on these results, the terminal determines the content and display method of a warning message to the user.

[0327] 6. Displaying a warning message

[0328] The device will display a warning message to the user. For example, if the user is nervous, it will display a friendly message such as "This email may be fraudulent. Please proceed with caution."

[0329] If the user is calm, a simple message such as "This email may be fraudulent" will be displayed.

[0330] 7. User Response

[0331] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0332] Example of a web browsing application

[0333] 1. Accessing the URL

[0334] A user clicks on a new URL in a web browsing application.

[0335] The device extracts the URL.

[0336] 2. URL parsing request generation

[0337] The device generates an analysis request to the AI ​​model based on the URL information.

[0338] 3. Analysis using AI models

[0339] The device sends the analysis request generated to the AI ​​model.

[0340] An AI model analyzes URL destinations and compares them with known phishing sites and malicious domains.

[0341] 4. Emotion Recognition by Emotion Engine

[0342] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[0343] An emotion engine analyzes this data to determine the user's emotional state.

[0344] 5. Integration of analysis results and emotion recognition results

[0345] The device receives the analysis results and emotion recognition results from the AI ​​model and determines the content and display method of the warning message to the user based on these.

[0346] 6. Displaying a warning message

[0347] The device displays a warning message to the user, for example, if the user is surprised, it displays a descriptive message such as "This site may be fraudulent. Please investigate further."

[0348] If the user is calm, a simple message such as "This site may be a scam" will be displayed.

[0349] 7. User Response

[0350] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0351] Thus, the present invention is a system that recognizes a user's emotional state in email and web browsing applications and adjusts the content and display of warning messages accordingly, allowing users to more effectively avoid the risk of fraud.

[0352] The processing flow will be explained below.

[0353] Step 1:

[0354] The server receives new emails and forwards them to the device, which then recognizes that new emails have been received.

[0355] Step 2:

[0356] The device will notify the user of new mail. A notification will pop up to let you know that new mail has arrived.

[0357] Step 3:

[0358] The user clicks to open a new email, which starts the process of displaying the email contents.

[0359] Step 4:

[0360] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[0361] Step 5:

[0362] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[0363] Step 6:

[0364] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[0365] Step 7:

[0366] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[0367] Step 8:

[0368] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[0369] Step 9:

[0370] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[0371] Step 10:

[0372] The device receives the analysis results from the AI ​​model and generates a warning for the user based on the results.

[0373] Step 11:

[0374] The device activates an emotion engine to recognize the user's emotions, and collects the user's facial expressions and voice using the device's camera or microphone.

[0375] Step 12:

[0376] The emotion engine analyzes the collected data to determine the user's current emotional state, categorizing it into emotion categories such as nervous, angry, surprised, or calm.

[0377] Step 13:

[0378] The device combines the analysis results of the AI ​​model with the emotion recognition results of the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it will select a friendly and reassuring message.

[0379] Step 14:

[0380] The device will then display appropriate warning messages to the user, adjusting the tone and details of the messages to match the user's emotional state.

[0381] Step 15:

[0382] Users should review the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0383] In this way, by performing specific processing at each step, a system is realized that not only reliably detects fraudulent emails and warns users, but also uses an emotion engine to take appropriate action taking into account the user's emotional state.

[0384] Example 2

[0385] 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."

[0386] In today's digital communications, damage caused by fraudulent emails and phishing sites is increasing. In particular, these fraudulent activities use sophisticated techniques, making them difficult to detect with conventional security systems. Furthermore, since warning messages are provided uniformly without taking into account the user's emotional state, many users ignore the warnings or fail to take appropriate action, which is a problem. Therefore, there is a need for a system that provides more appropriate warnings by taking into account the user's emotional state.

[0387] 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.

[0388] In this invention, the terminal includes means for receiving an email body or a URL, means for generating a request for analyzing the email body or the URL, means for sending a request to a generative AI model and analyzing the possibility of fraud, means for detecting the user's facial expressions and voice using a camera or microphone and activating an emotion engine that determines the user's emotional state, means for notifying the user of a warning based on the analysis result and the emotional state, means for determining the content and display method of the warning message based on the analysis result and the emotion recognition result, and means for displaying a warning message customized according to the user's emotional state. This makes it possible to effectively notify the user of the risk of fraud and encourage appropriate action while taking the user's emotional state into consideration.

[0389] "Terminal" refers to a computing device used by a user, on which an email application or a web browsing application runs.

[0390] The "mail body" refers to the main content of an email sent or received by a user, and is the part that contains the written message or content.

[0391] "URL" refers to the address that specifies the destination of a link to a web page or resource.

[0392] "Means for generating a request" refers to the function for constructing and generating the request data required to analyze the email body or URL.

[0393] A "generative AI model" refers to an artificial intelligence model that uses machine learning technology to analyze email text and URLs to assess the likelihood of fraud.

[0394] "Means for analyzing" refers to the ability of the generative AI model to assess the likelihood of fraud based on the generated request.

[0395] An "emotion engine" refers to a system that uses a camera or microphone to capture a user's facial expressions and voice, and analyzes that data to determine the user's emotional state.

[0396] "Means for notifying warnings" refers to the ability to display warnings to users based on the likelihood of fraud or their emotional state.

[0397] "Means for determining the content and display method of the warning message" refers to a function that takes into consideration the analysis results and emotion recognition results and determines what kind of warning message to display to the user and how.

[0398] A "customized alert message" refers to an alert message whose content and presentation are tailored to the user's current emotional state.

[0399] The present invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. Furthermore, by combining it with an emotion engine, the system grasps the user's emotional state and takes appropriate action. Detailed embodiments for implementing this system are described below.

[0400] The system is integrated into email or web browsing applications running on a device. The device requires a camera, microphone, display, and internet connection for its hardware. The software includes email client software, a web browser, a generative AI model for natural language processing (NLP), and an emotion engine for analyzing the user's emotional state.

[0401] First, the server receives a new email and forwards it to the device. The device displays a notification of the new email to the user. When the user opens the email, the device extracts the email body and headers. Based on this, the device generates an analysis request for the AI ​​model and sends it to the generative AI model.

[0402] The generative AI model analyzes the email body and links, using natural language processing to detect suspicious phrases and links, and assigns a score indicating the likelihood of fraud. The analysis results are then sent back to the device.

[0403] In parallel, the device activates an emotion engine that uses the camera and microphone to detect the user's facial expressions and voice, and determines the user's emotional state based on the collected data.

[0404] The device integrates the analysis results from the AI ​​model and the emotion recognition results from the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it may display "This email may be fraudulent. Please proceed with caution," but if the user is calm, it may simply display "This email may be fraudulent." The device displays an appropriate warning message to the user, who can then confirm the message and take appropriate action to avoid the risk of fraud.

[0405] It works similarly in web browsing applications. When a user clicks on a new URL, the device extracts the URL and generates an analysis request from the AI ​​model. The generative AI model analyzes the URL's destination and compares it with known phishing sites and malicious domains. The emotion engine determines the user's emotional state and, based on the analysis and emotion recognition results, determines the content and display method of the warning message. The device then displays an appropriate warning to the user and prevents them from accessing the dangerous site.

[0406] A specific example of this is the prompt:

[0407] Email applications:

[0408] "This email may be fraudulent. Please review it carefully and verify with other sources before proceeding."

[0409] Web browsing applications:

[0410] "This site may be a scam. Please avoid visiting."

[0411] This system provides warning messages that take into account the user's emotional state, allowing for more effective prevention of fraud risks.

[0412] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0413] Step 1: Receiving emails and notifications

[0414] The server receives new emails. The input is the new email and the output is the received email data. Specifically, the server receives new emails using the SMTP protocol and saves them in the user's mailbox. The server then transfers this email data to the terminal. The terminal displays a pop-up notification of new emails to the user. The input is the email data received from the server and the output is the notification to the user.

[0415] Step 2: Generate an email analysis request

[0416] The user clicks to open a new email. The input is the user's click action, and the output is the email body and email header data. The device extracts the email body and email header. The device obtains this data using the email client's internal API and generates an analysis request for the generative AI model. The input is the extracted email data, and the output is the analysis request.

[0417] Step 3: Analysis by AI model

[0418] The device sends the analysis request it generated to the generative AI model. The input is the analysis request, and the output is the analysis result. Specifically, data is sent using an HTTP POST request. The generative AI model analyzes the email body and links, and uses natural language processing technology to detect suspicious phrases and links. The output is an analysis result with a score for the likelihood of fraud. The analysis result is returned to the device in JSON format.

[0419] Step 4: Emotion Recognition with the Emotion Engine

[0420] The device starts the emotion engine and detects the user's facial expressions and voice using a camera or microphone. The input is real-time data of the user's facial expressions and voice, and the output is the analyzed emotional state. The emotion engine analyzes this data and determines the user's emotional state. The output is the user's emotional state (e.g., nervous, surprised, calm, etc.).

[0421] Step 5: Integrating analysis results with emotion recognition results

[0422] The device receives the analysis results from the generative AI model and the emotion recognition results from the emotion engine. The inputs are the analysis results and emotion recognition results, and the output is the integrated information used for notifications. Specifically, the device runs an algorithm based on the score of the analysis results and the emotion recognition results to determine the content and display method of the warning message to the user.

[0423] Step 6: Displaying warning messages

[0424] The terminal displays a customized warning message to the user. The input is the integrated information, and the output is a warning message to the user. The terminal uses a widget in an email client or web browser to display a pop-up message on the screen. For example, if the user is nervous, it displays "This email may be fraudulent. Please proceed with caution," and if the user is calm, it displays a simple message saying "This email may be fraudulent."

[0425] Step 7: User Action

[0426] The user checks the warning message and takes appropriate action to avoid the risk of fraud. The input is the warning message, and the output is the user's action. Specifically, the user may take action such as deleting the email or not clicking on the link. For example, if the user sees the warning message and selects "Delete fraudulent email," the email will be deleted.

[0427] (Application example 2)

[0428] 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."

[0429] Conventional email and web browsing applications are ineffective at detecting fraudulent emails and websites, and warnings to users are provided in a generic format that is not adapted to the user's emotional state. As a result, users are at increased risk of falling victim to fraud because they are unable to take appropriate action. Therefore, the present invention aims to improve the accuracy of fraud detection and provide warnings that are adapted to the user's emotional state, encouraging the user to take more effective action.

[0430] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0431] In this invention, the server includes, in an email or web browsing application, a means for a terminal to receive an email body or URL, a means for the terminal to generate a request for analyzing the email body or URL, a means for the terminal to send the request to an AI model to analyze the possibility of fraud, a means for the terminal to detect a user's emotional state using a camera and microphone, a means for an emotion engine to analyze the user's emotional state, and a means for the terminal to notify a user of a warning based on the analysis results of the AI ​​model and the emotion analysis results of the emotion engine. This enables highly accurate detection of fraudulent activities and the provision of a warning message adapted to the user's emotional state.

[0432] A "terminal" is an electronic device operated by a user, and is a device for receiving emails and browsing the web.

[0433] The "mail body" is the content of the email, and is information that includes text, images, links, and the like.

[0434] "URL" is a Uniform Resource Locator that indicates the address of a web page, and is an identifier for specifying the destination of browsing access.

[0435] An "AI model" is an analytical system that uses artificial intelligence algorithms and is a model trained to perform a specific task.

[0436] An "analysis request" is a command sent from a device to an AI model that contains the information necessary to analyze specific data.

[0437] "Fraudulence" is a probability or score that indicates whether an email or website is attempting to commit fraud.

[0438] The "emotional state" indicates the psychological state of the user, and is an emotion judged from facial expressions, voice, etc.

[0439] An "emotion engine" is software that analyzes a user's emotional state from their facial expressions and voice.

[0440] A "warning message" is a notification message that alerts users to risks and caution, and is in the form of text or audio to notify them of possible fraudulent activity.

[0441] A "camera" is a device that captures images and videos and is used to detect a user's facial expression.

[0442] A "microphone" is a device that records sound and is used to detect the user's voice.

[0443] This invention realizes a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and then analyzes the user's emotional state and responds appropriately. The specific hardware and software that make up the system, as well as the actual processing procedures, are described below.

[0444] System Configuration

[0445] The system consists of the following major hardware and software components:

[0446] Device: An electronic device operated by a user, such as a smartphone, tablet, or computer.

[0447] Camera: A built-in camera used to detect the user's facial expressions.

[0448] Microphone: A microphone built into the device, used to detect the user's voice.

[0449] AI model: An artificial intelligence model that analyzes the content of emails and URLs to determine the likelihood of fraud.

[0450] Emotion engine: Software for analyzing the user's emotional state from their facial expressions and voice.

[0451] Natural language processing technology: Used to analyze email text and URLs.

[0452] System operation explanation

[0453] The device receives an email or URL: When a user receives a new email or clicks on a new website URL, the device recognizes this.

[0454] Generating an analysis request: The device generates an analysis request to the AI ​​model based on the received email body and URL. This request includes detailed information about the email content and URL.

[0455] Performing fraud analysis: The device sends an analysis request to the AI ​​model, which uses natural language processing techniques to analyze the content of emails and URLs and calculates a score indicating the likelihood of fraud.

[0456] Emotional state detection: The device activates the camera and microphone to collect the user's facial expressions and voice in real time, and the emotion engine then analyzes this data to determine the user's emotional state.

[0457] Integrated display of warning messages: The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to adjust the content and display method of the warning message displayed to the user. For example, if the user is surprised, a more detailed warning message will be displayed, and if the user is calm, a simple message will be displayed.

[0458] Examples and prompts

[0459] For example, if a user receives an email that reads, "Congratulations! You have won a $1000 gift card. Click here to claim.", the device will generate the following prompt:

[0460] Given the following email content and header, detect if it's a phishing email and provide a confidence score. Email content: "Congratulations! You have won a $1000 gift card. Click here to claim." Email header: "From: no-reply@fakesite.com, Subject: Claim your prize"

[0461] This prompt is sent to an AI model that analyzes the email for its potential fraudulent nature. Based on the analysis results and the user's emotional state, a warning message such as "This email may be fraudulent. Please investigate further" is displayed.

[0462] In this way, the system can achieve highly accurate detection of fraudulent activity and present warning messages that adapt to the user's emotional state.

[0463] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0464] Step 1:

[0465] The device receives a new email or URL. By receiving an email or clicking a URL, the device obtains this information. The input is the new email or the accessed URL, and the output is the email body or URL to be analyzed.

[0466] Step 2:

[0467] The device generates a request to analyze the email body and URL. Specifically, it generates a prompt message containing detailed information about the email content and URL. The input is the email body and URL, and the output is an analysis request to send to the AI ​​model.

[0468] Step 3:

[0469] The analysis request generated by the device is sent to an AI model to analyze the likelihood of fraud. The AI ​​model uses natural language processing techniques to analyze the content of emails and URLs and generate a score indicating fraud. The input is the analysis request, and the output is a fraud likelihood score.

[0470] Step 4:

[0471] The device detects the user's emotional state using a camera and microphone. Specifically, the camera captures the user's facial expressions and the microphone records their voice. The emotion engine analyzes this data and determines the user's emotions. The input is camera images and voice data, and the output is the user's emotional state.

[0472] Step 5:

[0473] The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to determine the most appropriate content and display method of the warning message. The input is the fraud probability score and emotional state, and the output is an optimized warning message.

[0474] Step 6:

[0475] The terminal displays a warning message to the user. The content of the message is a warning about the risk of fraud, and appropriate wording is used depending on the user's emotional state. The input is the optimized warning message, and the output is the warning information that the user confirms.

[0476] Step 7:

[0477] The user checks the warning message and takes appropriate action to avoid the risk of fraud. For example, the user may delete the email, not click on the link, or restrict access to the site. The input is the warning message, and the output is the user's appropriate response.

[0478] 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.

[0479] 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.

[0480] 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.

[0481] [Second embodiment]

[0482] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0483] 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.

[0484] 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).

[0485] 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.

[0486] 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.

[0487] 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).

[0488] 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.

[0489] 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.

[0490] 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.

[0491] 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.

[0492] 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.

[0493] 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."

[0494] The present invention provides a system for automatically detecting fraudulent emails and websites in email applications and web browsing applications, and for issuing warnings to users. The processing of a program for implementing this system is specifically explained below in natural language.

[0495] System Overview

[0496] The system is embedded within email or web browsing applications and analyzes incoming emails and visited URLs to determine potential fraudulent activity using AI models running on the device.

[0497] Example of an email application

[0498] 1. Receiving emails and notifications

[0499] When a user starts an email application, the server receives new emails and transfers them to the terminal.

[0500] The terminal notifies the user that new mail has arrived.

[0501] 2. Generate email analysis request

[0502] When a user opens a new email, the device extracts the email body and header information, including the sender's email address, subject, body content, links, and attachment information.

[0503] The device generates an analysis request to the AI ​​model based on this information.

[0504] 3. Analysis using AI models

[0505] The device sends the analysis request generated to the AI ​​model.

[0506] The AI ​​model analyzes the email body and links, and uses natural language processing techniques to examine the content, comparing it with known phishing and spam email characteristics to detect suspicious phrases and links.

[0507] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[0508] 4. Receiving analysis results and generating alerts

[0509] The device receives the analysis results from the AI ​​model.

[0510] Based on the analysis results, the device will issue a warning to the user. For example, if the analysis results indicate a high probability of fraud, the device will display a warning message saying, "This email may be fraudulent."

[0511] 5. User Response

[0512] Users should acknowledge the warning and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0513] Example of a web browsing application

[0514] 1. Accessing the URL

[0515] A user clicks on a new URL in a web browsing application.

[0516] The device extracts the URL.

[0517] 2. URL parsing request generation

[0518] The device generates an analysis request to send the URL information to the AI ​​model.

[0519] 3. Analysis using AI models

[0520] The device sends the analysis request generated to the AI ​​model.

[0521] The AI ​​model analyzes the URL destination (web page content) and compares it with known phishing sites and malicious domains.

[0522] 4. Receiving analysis results and generating alerts

[0523] If the device receives the analysis results from the AI ​​model and determines that there is a high possibility of fraud, the device will display a warning saying, "This site may be fraudulent."

[0524] 5. User Response

[0525] Users should check the warning and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0526] In this way, the present invention is a system that automatically detects fraud risks in email and web browsing applications and warns users, preventing fraud damage before it occurs. In particular, by using an AI model, it achieves more accurate analysis than conventional simple filtering.

[0527] The processing flow will be explained below.

[0528] Step 1:

[0529] The server receives new emails and forwards them to the device, which causes the email application to recognize that new emails have arrived.

[0530] Step 2:

[0531] The device will notify the user of new emails. A notification will pop up to let you know that new email has arrived.

[0532] Step 3:

[0533] The user clicks to open a new email, which starts the process of displaying the email contents.

[0534] Step 4:

[0535] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[0536] Step 5:

[0537] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[0538] Step 6:

[0539] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[0540] Step 7:

[0541] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[0542] Step 8:

[0543] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[0544] Step 9:

[0545] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[0546] Step 10:

[0547] The device receives the analysis results and score and generates a warning to the user based on the results. For example, if there is a high possibility of fraud, it generates a warning message such as "This email may be fraudulent."

[0548] Step 11:

[0549] The device displays a warning message to the user. A message is displayed on the user interface to inform the user of the danger.

[0550] Step 12:

[0551] Users can prevent themselves from falling victim to fraud by checking the warning message and taking action such as deleting the email or not clicking on the link.

[0552] In this way, by performing specific processing at each step, a system is realized that can reliably detect fraudulent emails and warn users.

[0553] Example 1

[0554] 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."

[0555] Conventional email and web browsing applications have difficulty detecting fraud risks and providing appropriate warnings to users. This puts users at high risk of being exposed to phishing emails and fraudulent websites. The present invention aims to reduce such risks and enable users to safely communicate electronically and browse the web.

[0556] 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.

[0557] In this invention, the server includes means for a terminal to receive data related to electronic communications or websites, means for the terminal to generate a request for analyzing the data, means for transmitting the request to a machine learning model to analyze the possibility of fraud, means for notifying a user of a warning based on the analysis result, and means for guiding the user's operation based on the warning, thereby enabling the user to quickly recognize the risk of fraud and take appropriate measures.

[0558] "Device" refers to a hardware device used for electronic communication or website access, including, for example, a computer, smartphone, or tablet.

[0559] "Electronic Communications" means information such as text, images, links, etc. sent or received via email or messaging applications.

[0560] "Website" means a collection of web pages publicly available on the Internet and is a resource of information accessible to users.

[0561] "Data" refers to any information received from an electronic communication or website, including text, images, links, and metadata.

[0562] A "machine learning model" is an algorithm that analyzes large amounts of data, learns patterns and features from that data, and predicts and classifies new data.

[0563] "Natural language processing technology" is a general term for technologies that allow computers to understand, generate, and analyze human language, and includes text analysis, translation, and dialogue systems.

[0564] "Analysis Request" refers to a packet of information sent to a machine learning model to analyze data.

[0565] "Fraud" refers to malicious acts such as phishing emails and fraudulent websites that attempt to illegally obtain users' personal information or money.

[0566] "Warning" refers to a message that alerts the user based on the analysis results and provides information to help avoid risky behavior.

[0567] "Guiding user operations" means showing the user appropriate actions to take based on the warning, such as deleting the email or avoiding clicking on links.

[0568] The present invention provides a system for automatically detecting fraudulent emails and fraudulent websites in email applications and web browsing applications, and for issuing a warning to the user. A detailed description of how to implement the system of the present invention is provided below.

[0569] Hardware and software used

[0570] Hardware: Devices (computers, smartphones, tablets, etc.)

[0571] software:

[0572] Email application

[0573] Web browsing applications

[0574] Machine learning models (e.g., the BERT model, which is specialized for natural language processing)

[0575] System Overview

[0576] The system is embedded within email or web browsing applications and uses machine learning models running on the device to analyze emails received and website URLs visited in real time to identify potential fraudulent activity.

[0577] Example from an email application

[0578] Receiving emails and notifications

[0579] The server receives the user's email and forwards new emails to the terminal.

[0580] The device will notify the user of new emails by sounding an alert and displaying a pop-up message.

[0581] Generate email analysis request

[0582] When a user opens an email application and clicks on a new email, the device extracts the email body and header information.

[0583] Based on the information extracted by the device, an analysis request is generated for the machine learning model.

[0584] Analysis using machine learning models

[0585] Sends device-generated analysis requests to machine learning models.

[0586] Machine learning models analyze email text and links, and use natural language processing techniques to inspect the content.

[0587] The model matches the characteristics of known phishing and spam emails to detect suspicious phrases and links.

[0588] The analysis generates a score indicating the likelihood of fraud.

[0589] Receiving analysis results and generating alerts

[0590] The device receives the analysis results and issues a warning to the user. For example, if it determines that the email is likely to be fraudulent, it displays a warning message saying, "This email may be fraudulent."

[0591] User response

[0592] Users acknowledge the warning and take action to reduce the risk of fraud, such as deleting the email or not clicking on the link.

[0593] Example of a web browsing application

[0594] URL access

[0595] When a user clicks on a new URL in a web browsing application, the device extracts the URL.

[0596] URL parsing request generation

[0597] The device generates an analysis request to send the URL information to the machine learning model.

[0598] Analysis using machine learning models

[0599] Sends device-generated analysis requests to machine learning models.

[0600] Machine learning models analyze the URL destination (web page content) and compare it with known phishing sites and malicious domains.

[0601] Receiving analysis results and generating alerts

[0602] If the device receives the analysis results and determines that there is a high possibility of fraud, it will display a warning message saying, "This site may be fraudulent."

[0603] User response

[0604] Users can acknowledge the warning and take action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0605] Examples of specific examples and prompts

[0606] Example of an email application

[0607] When a user opens a new email containing an "Unsubscribe" link, a machine learning model identifies the link as a phishing risk and displays a warning.

[0608] Prompt Sentence Examples

[0609] Please check the safety of this email before clicking any links, as it is likely a scam.

[0610] Example of a web browsing application

[0611] When a user attempts to visit a specific online shopping site, a machine learning model identifies the site as a known fraudulent site and displays a warning.

[0612] Prompt Sentence Examples

[0613] This site may be a scam. Please consider other safe ways to purchase.

[0614] This invention realizes a system that can instantly detect fraudulent behavior during email and web browsing, allowing users to avoid the risks of phishing emails and fraudulent sites. In particular, by using machine learning models, it is possible to perform more accurate analysis than conventional methods.

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

[0616] (Processing flow in email application)

[0617] Step 1:

[0618] The server receives the user's email and forwards new emails to the device.

[0619] Input: User's email account information, new emails

[0620] Output: New email notification to device

[0621] Specific operation: Records emails from various mail servers (IMAP / POP3) in the receiving log and forwards them to the mail application.

[0622] Step 2:

[0623] The device will notify the user that new email has arrived.

[0624] Input: New email

[0625] Output: New mail notification to user

[0626] Specific operation: On the smartphone, the message "You have new email" will appear in the notification bar and a notification sound will be heard on the device itself.

[0627] Step 3:

[0628] The user opens their email application and clicks on new email.

[0629] The device extracts the email body and header information.

[0630] Input: New emails opened by the user, including the email body and header information

[0631] Output: Extracted email body and header information

[0632] What happens: The sender address, subject, body, links, and attachment information are separated for analysis.

[0633] Step 4:

[0634] Based on the information extracted by the device, an analysis request is generated for the AI ​​model.

[0635] Input: Extracted email body and header information

[0636] Output: Analysis request to the AI ​​model

[0637] Specific operation: The analysis request is structured in JSON format or other format and prepared for sending to the machine learning model.

[0638] Step 5:

[0639] The device generates an analysis request and sends it to the AI ​​model.

[0640] Input: Analysis request to the AI ​​model

[0641] Output: Sending data to the AI ​​model

[0642] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[0643] Step 6:

[0644] The AI ​​model analyzes the email body and links, and uses natural language processing technology to inspect the content.

[0645] Input: Analysis request email body and link

[0646] Output: Analysis results (fraud likelihood score)

[0647] What it does: Contextual analysis, spam filtering, and matching links to known phishing phrases.

[0648] Step 7:

[0649] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[0650] Input: Parsed email body and links

[0651] Output: Analysis results with fraud probability score

[0652] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[0653] Step 8:

[0654] The device receives the analysis results from the AI ​​model.

[0655] Input: Analysis results with fraud likelihood score

[0656] Output: A warning notice to the user

[0657] Specific Actions: Based on the estimated risk, a warning message is prepared and displayed in the user interface.

[0658] Step 9:

[0659] The device will display a warning to the user.

[0660] Input: warning message

[0661] Output: A warning notice to alert the user

[0662] Specific behavior: If the email is rated as "likely fraudulent," a pop-up message will appear saying "This email may be fraudulent."

[0663] Step 10:

[0664] Users should acknowledge the warning and take action to avoid the risk of fraud.

[0665] Input: warning message

[0666] Output: User response (delete email, avoid clicking links)

[0667] Specific action: The user clicks the "Delete" button to delete the email.

[0668] (Processing flow in a web browsing application)

[0669] Step 1:

[0670] A user clicks on a new URL in their web browsing application.

[0671] Input: User clicks on a URL

[0672] Output: Extracted URLs

[0673] What it does: The browser captures the click event and records the URL being accessed.

[0674] Step 2:

[0675] The device will extract the URL.

[0676] Input: Clicked URL

[0677] Output: URL for analysis

[0678] What it does: Saves the URL as structured data.

[0679] Step 3:

[0680] The device generates an analysis request to send the URL information to the AI ​​model.

[0681] Input: A structured URL to parse

[0682] Output: Analysis request to the AI ​​model

[0683] Specific operation: Structure the analysis request in JSON format or similar.

[0684] Step 4:

[0685] The device generates an analysis request and sends it to the AI ​​model.

[0686] Input: Analysis request to the AI ​​model

[0687] Output: Sending data to the AI ​​model

[0688] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[0689] Step 5:

[0690] The AI ​​model analyzes the URL destination (webpage content) and compares it with known phishing sites and malicious domains.

[0691] Input: URL of the analysis request

[0692] Output: Analysis results

[0693] What it does: Scrape the content of web pages and compare it to a database of known phishing sites.

[0694] Step 6:

[0695] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[0696] Input: scraped web page content

[0697] Output: Analysis results with fraud probability score

[0698] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[0699] Step 7:

[0700] The device receives the analysis results and notifies the user with a warning.

[0701] Input: Analysis results with fraud likelihood score

[0702] Output: A warning notice to the user

[0703] Specific behavior: If there is a high possibility of fraud, a warning message will be displayed in a pop-up saying, "This site may be fraudulent."

[0704] Step 8:

[0705] Users should acknowledge the warning and take action to avoid the risk of fraud.

[0706] Input: warning message

[0707] Output: User response (stops visiting the site)

[0708] Specific Action: The user closes the browser tab or stops visiting the site.

[0709] (Application example 1)

[0710] 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."

[0711] In today's world, as internet-based communications become increasingly common, the number of victims of fraudulent emails and phishing sites is rapidly increasing. These fraudulent activities often result in the leakage of personal information and financial losses, posing a significant risk to many users. Current technology makes it difficult to detect these fraudulent activities in advance and implement countermeasures, creating a demand for systems with advanced analytical capabilities. The purpose of this invention is to solve these problems and enable users to use email and websites safely.

[0712] 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.

[0713] In this invention, the server includes: a terminal receiving an email body or URL; a terminal generating a request for analyzing the email body or URL; a terminal sending the request to an AI model to analyze the possibility of fraud; a terminal issuing a warning to the user based on the analysis results; a smartphone installed on the smartphone for analyzing emails and websites in real time; and a terminal displaying a warning if the possibility of fraud is high based on the analysis results. This enables sophisticated analysis of fraudulent emails and phishing sites and displaying a warning to the user in real time.

[0714] A "terminal" is an electronic device such as a smartphone, tablet, or PC that is operated by a user.

[0715] "Email body" refers to the body of the message written by the sender of the email, which may contain text and links.

[0716] "URL" stands for Uniform Resource Locator and indicates the address of a web page on the Internet.

[0717] An "analysis request" is request data generated to allow an AI model to determine the possibility of fraud based on the content of an email body or URL.

[0718] An "AI model" is an analytical engine that utilizes artificial intelligence technology and is designed to detect fraudulent emails and phishing sites.

[0719] "Natural language processing" is a technique in which AI models understand and analyze human language, and is used to scrutinize the content of emails and web pages.

[0720] A "warning message" is a notification message that is displayed to warn the user of possible fraud based on the analysis results.

[0721] "Real-time" refers to the instantaneous reception and analysis of data, allowing users to obtain results immediately.

[0722] This invention is a system that automatically detects fraudulent emails and websites in email and web browsing applications and issues a warning to the user. This system is mainly composed of a terminal and a server, and the terminal receives emails and URLs, analyzes their contents, and notifies the user.

[0723] Program processing explanation

[0724] The server first receives the email body or URL sent from the device. The email body includes the sender address, subject, body text, links, etc. The server generates a request to analyze this content and sends it to an AI model. This AI model uses natural language processing (NLP) technology and is designed to detect potential fraud and phishing. Specifically, it analyzes the content by comparing it with keywords and phrases commonly found in fraudulent emails, as well as information about the domains to which the links lead.

[0725] As a result of the analysis, the AI ​​model assigns a score indicating the likelihood of fraud and sends it back to the server. The server generates and notifies the user of a warning message based on this score. For example, if the score is high, a warning such as "This email may be fraudulent" or "This site may be fraudulent" will be displayed.

[0726] The system of the present invention mainly uses the following hardware and software. The terminal is a general computing device such as a smartphone or tablet. The server can be a cloud server or a dedicated server in a data center. In addition, the AI ​​model uses a deep learning framework such as TensorFlow or PyTorch.

[0727] Specific examples

[0728] For example, when a user receives an email disguised as a message from a bank, the body of the email often contains keywords such as "transfer" and "password change." When the user opens the email, the device sends the email body to a server, where the AI ​​model analyzes these keywords. If the analysis returns a score of "very likely to be fraudulent," the user is shown a warning message saying, "This email may be fraudulent."

[0729] Prompt Sentence Examples

[0730] 1. Enter the sender, subject, and body of the email you received and assess its likelihood of fraud.

[0731] 2. "Analyze the URL destination to determine whether it is phishing."

[0732] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0733] Step 1:

[0734] The device receives the email or URL.

[0735] What it does: The device receives new emails or website URLs via applications the user has opened. For emails, information such as the sender, subject, body, and links is extracted.

[0736] Input: Received email body or URL

[0737] Output: Extracted email and URL information

[0738] Step 2:

[0739] The device generates an analysis request.

[0740] What it does: Based on the information in the received email or URL, it creates an analysis request to send to the AI ​​model. This request includes the text of the email body and the URL information.

[0741] Input: Extracted email and URL information

[0742] Output: Analysis request (email body text, URL information, etc.)

[0743] Step 3:

[0744] The device sends an analysis request to the server.

[0745] Specific operation: The generated analysis request is sent to the server, which receives it and prepares to pass the request to the AI ​​model.

[0746] Input: Parse request

[0747] Output: The parsing request sent to the server

[0748] Step 4:

[0749] The server sends an analysis request to the AI ​​model and begins the analysis.

[0750] How it works: When the server receives an analysis request, it passes it to the AI ​​model, which uses natural language processing techniques to analyze the content of the email body and URLs to detect potential fraud.

[0751] Input: Parse request

[0752] Output: Analysis results (e.g., fraud probability score)

[0753] Step 5:

[0754] The server receives the analysis results and sends them to the terminal.

[0755] How it works: The server receives the analysis results returned by the AI ​​model and sends them to the device, which includes a score indicating the likelihood of fraud.

[0756] Input: Analysis results (e.g., fraud likelihood score)

[0757] Output: Analysis results sent to the device

[0758] Step 6:

[0759] The device notifies the user of a warning based on the analysis results.

[0760] Specific operation: When the device receives the analysis results, if there is a high possibility of fraud, it will display a warning message to the user. The warning message will include content such as "This email may be fraudulent" or "This site may be fraudulent."

[0761] Input: Analysis result (fraud likelihood score)

[0762] Output: The warning message displayed to the user.

[0763] 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.

[0764] This invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. It also incorporates an emotion engine to understand the user's emotional state and take appropriate action. Below, we will explain in detail the processing of the program for implementing this system.

[0765] System Overview

[0766] This system is embedded in email or web browsing applications, analyzes incoming emails and accessed URLs, and determines whether fraud is likely. It also uses an emotion engine to recognize the user's emotional state and adjust the display method and content of warning messages accordingly.

[0767] Example of an email application

[0768] 1. Receiving emails and notifications

[0769] The server receives new emails and forwards them to the terminal.

[0770] The terminal notifies the user that new mail has arrived.

[0771] 2. Generate email analysis request

[0772] The user clicks to open a new email.

[0773] The terminal extracts the email body and email header.

[0774] The device generates an analysis request to the AI ​​model based on the information extracted.

[0775] 3. Analysis using AI models

[0776] The device sends the analysis request generated to the AI ​​model.

[0777] The AI ​​model analyzes the email text and links, and uses natural language processing techniques to detect suspicious phrases and links.

[0778] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[0779] 4. Emotion Recognition by Emotion Engine

[0780] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[0781] An emotion engine analyzes this data to determine the user's emotional state.

[0782] 5. Integration of analysis results and emotion recognition results

[0783] The device receives the analysis results from the AI ​​model and the emotion recognition results from the emotion engine.

[0784] Based on these results, the terminal determines the content and display method of a warning message to the user.

[0785] 6. Displaying a warning message

[0786] The device will display a warning message to the user. For example, if the user is nervous, it will display a friendly message such as "This email may be fraudulent. Please proceed with caution."

[0787] If the user is calm, a simple message such as "This email may be fraudulent" will be displayed.

[0788] 7. User Response

[0789] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0790] Example of a web browsing application

[0791] 1. Accessing the URL

[0792] A user clicks on a new URL in a web browsing application.

[0793] The device extracts the URL.

[0794] 2. URL parsing request generation

[0795] The device generates an analysis request to the AI ​​model based on the URL information.

[0796] 3. Analysis using AI models

[0797] The device sends the analysis request generated to the AI ​​model.

[0798] An AI model analyzes URL destinations and compares them with known phishing sites and malicious domains.

[0799] 4. Emotion Recognition by Emotion Engine

[0800] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[0801] An emotion engine analyzes this data to determine the user's emotional state.

[0802] 5. Integration of analysis results and emotion recognition results

[0803] The device receives the analysis results and emotion recognition results from the AI ​​model and determines the content and display method of the warning message to the user based on these.

[0804] 6. Displaying a warning message

[0805] The device displays a warning message to the user, for example, if the user is surprised, it displays a descriptive message such as "This site may be fraudulent. Please investigate further."

[0806] If the user is calm, a simple message such as "This site may be a scam" will be displayed.

[0807] 7. User Response

[0808] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0809] Thus, the present invention is a system that recognizes a user's emotional state in email and web browsing applications and adjusts the content and display of warning messages accordingly, allowing users to more effectively avoid the risk of fraud.

[0810] The processing flow will be explained below.

[0811] Step 1:

[0812] The server receives new emails and forwards them to the device, which then recognizes that new emails have been received.

[0813] Step 2:

[0814] The device will notify the user of new mail. A notification will pop up to let you know that new mail has arrived.

[0815] Step 3:

[0816] The user clicks to open a new email, which starts the process of displaying the email contents.

[0817] Step 4:

[0818] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[0819] Step 5:

[0820] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[0821] Step 6:

[0822] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[0823] Step 7:

[0824] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[0825] Step 8:

[0826] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[0827] Step 9:

[0828] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[0829] Step 10:

[0830] The device receives the analysis results from the AI ​​model and generates a warning for the user based on the results.

[0831] Step 11:

[0832] The device activates an emotion engine to recognize the user's emotions, and collects the user's facial expressions and voice using the device's camera or microphone.

[0833] Step 12:

[0834] The emotion engine analyzes the collected data to determine the user's current emotional state, categorizing it into emotion categories such as nervous, angry, surprised, or calm.

[0835] Step 13:

[0836] The device combines the analysis results of the AI ​​model with the emotion recognition results of the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it will select a friendly and reassuring message.

[0837] Step 14:

[0838] The device will then display appropriate warning messages to the user, adjusting the tone and details of the messages to match the user's emotional state.

[0839] Step 15:

[0840] Users should review the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0841] In this way, by performing specific processing at each step, a system is realized that not only reliably detects fraudulent emails and warns users, but also uses an emotion engine to take appropriate action taking into account the user's emotional state.

[0842] Example 2

[0843] 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."

[0844] In today's digital communications, damage caused by fraudulent emails and phishing sites is increasing. In particular, these fraudulent activities use sophisticated techniques, making them difficult to detect with conventional security systems. Furthermore, since warning messages are provided uniformly without taking into account the user's emotional state, many users ignore the warnings or fail to take appropriate action, which is a problem. Therefore, there is a need for a system that provides more appropriate warnings by taking into account the user's emotional state.

[0845] 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.

[0846] In this invention, the terminal includes means for receiving an email body or a URL, means for generating a request for analyzing the email body or the URL, means for sending a request to a generative AI model and analyzing the possibility of fraud, means for detecting the user's facial expressions and voice using a camera or microphone and activating an emotion engine that determines the user's emotional state, means for notifying the user of a warning based on the analysis result and the emotional state, means for determining the content and display method of the warning message based on the analysis result and the emotion recognition result, and means for displaying a warning message customized according to the user's emotional state. This makes it possible to effectively notify the user of the risk of fraud and encourage appropriate action while taking the user's emotional state into consideration.

[0847] "Terminal" refers to a computing device used by a user, on which an email application or a web browsing application runs.

[0848] The "mail body" refers to the main content of an email sent or received by a user, and is the part that contains the written message or content.

[0849] "URL" refers to the address that specifies the destination of a link to a web page or resource.

[0850] "Means for generating a request" refers to the function for constructing and generating the request data required to analyze the email body or URL.

[0851] A "generative AI model" refers to an artificial intelligence model that uses machine learning technology to analyze email text and URLs to assess the likelihood of fraud.

[0852] "Means for analyzing" refers to the ability of the generative AI model to assess the likelihood of fraud based on the generated request.

[0853] An "emotion engine" refers to a system that uses a camera or microphone to capture a user's facial expressions and voice, and analyzes that data to determine the user's emotional state.

[0854] "Means for notifying warnings" refers to the ability to display warnings to users based on the likelihood of fraud or their emotional state.

[0855] "Means for determining the content and display method of the warning message" refers to a function that takes into consideration the analysis results and emotion recognition results and determines what kind of warning message to display to the user and how.

[0856] A "customized alert message" refers to an alert message whose content and presentation are tailored to the user's current emotional state.

[0857] The present invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. Furthermore, by combining it with an emotion engine, the system grasps the user's emotional state and takes appropriate action. Detailed embodiments for implementing this system are described below.

[0858] The system is integrated into email or web browsing applications running on a device. The device requires a camera, microphone, display, and internet connection for its hardware. The software includes email client software, a web browser, a generative AI model for natural language processing (NLP), and an emotion engine for analyzing the user's emotional state.

[0859] First, the server receives a new email and forwards it to the device. The device displays a notification of the new email to the user. When the user opens the email, the device extracts the email body and headers. Based on this, the device generates an analysis request for the AI ​​model and sends it to the generative AI model.

[0860] The generative AI model analyzes the email body and links, using natural language processing to detect suspicious phrases and links, and assigns a score indicating the likelihood of fraud. The analysis results are then sent back to the device.

[0861] In parallel, the device activates an emotion engine that uses the camera and microphone to detect the user's facial expressions and voice, and determines the user's emotional state based on the collected data.

[0862] The device integrates the analysis results from the AI ​​model and the emotion recognition results from the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it may display "This email may be fraudulent. Please proceed with caution," but if the user is calm, it may simply display "This email may be fraudulent." The device displays an appropriate warning message to the user, who can then confirm the message and take appropriate action to avoid the risk of fraud.

[0863] It works similarly in web browsing applications. When a user clicks on a new URL, the device extracts the URL and generates an analysis request from the AI ​​model. The generative AI model analyzes the URL's destination and compares it with known phishing sites and malicious domains. The emotion engine determines the user's emotional state and, based on the analysis and emotion recognition results, determines the content and display method of the warning message. The device then displays an appropriate warning to the user and prevents them from accessing the dangerous site.

[0864] A specific example of this is the prompt:

[0865] Email applications:

[0866] "This email may be fraudulent. Please review it carefully and verify with other sources before proceeding."

[0867] Web browsing applications:

[0868] "This site may be a scam. Please avoid visiting."

[0869] This system provides warning messages that take into account the user's emotional state, allowing for more effective prevention of fraud risks.

[0870] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0871] Step 1: Receiving emails and notifications

[0872] The server receives new emails. The input is the new email and the output is the received email data. Specifically, the server receives new emails using the SMTP protocol and saves them in the user's mailbox. The server then transfers this email data to the terminal. The terminal displays a pop-up notification of new emails to the user. The input is the email data received from the server and the output is the notification to the user.

[0873] Step 2: Generate an email analysis request

[0874] The user clicks to open a new email. The input is the user's click action, and the output is the email body and email header data. The device extracts the email body and email header. The device obtains this data using the email client's internal API and generates an analysis request for the generative AI model. The input is the extracted email data, and the output is the analysis request.

[0875] Step 3: Analysis by AI model

[0876] The device sends the analysis request it generated to the generative AI model. The input is the analysis request, and the output is the analysis result. Specifically, data is sent using an HTTP POST request. The generative AI model analyzes the email body and links, and uses natural language processing technology to detect suspicious phrases and links. The output is an analysis result with a score for the likelihood of fraud. The analysis result is returned to the device in JSON format.

[0877] Step 4: Emotion Recognition with the Emotion Engine

[0878] The device starts the emotion engine and detects the user's facial expressions and voice using a camera or microphone. The input is real-time data of the user's facial expressions and voice, and the output is the analyzed emotional state. The emotion engine analyzes this data and determines the user's emotional state. The output is the user's emotional state (e.g., nervous, surprised, calm, etc.).

[0879] Step 5: Integrating analysis results with emotion recognition results

[0880] The device receives the analysis results from the generative AI model and the emotion recognition results from the emotion engine. The inputs are the analysis results and emotion recognition results, and the output is the integrated information used for notifications. Specifically, the device runs an algorithm based on the score of the analysis results and the emotion recognition results to determine the content and display method of the warning message to the user.

[0881] Step 6: Displaying warning messages

[0882] The terminal displays a customized warning message to the user. The input is the integrated information, and the output is a warning message to the user. The terminal uses a widget in an email client or web browser to display a pop-up message on the screen. For example, if the user is nervous, it displays "This email may be fraudulent. Please proceed with caution," and if the user is calm, it displays a simple message saying "This email may be fraudulent."

[0883] Step 7: User Action

[0884] The user checks the warning message and takes appropriate action to avoid the risk of fraud. The input is the warning message, and the output is the user's action. Specifically, the user may take action such as deleting the email or not clicking on the link. For example, if the user sees the warning message and selects "Delete fraudulent email," the email will be deleted.

[0885] (Application example 2)

[0886] 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."

[0887] Conventional email and web browsing applications are ineffective at detecting fraudulent emails and websites, and warnings to users are provided in a generic format that is not adapted to the user's emotional state. As a result, users are at increased risk of falling victim to fraud because they are unable to take appropriate action. Therefore, the present invention aims to improve the accuracy of fraud detection and provide warnings that are adapted to the user's emotional state, encouraging the user to take more effective action.

[0888] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0889] In this invention, the server includes, in an email or web browsing application, a means for a terminal to receive an email body or URL, a means for the terminal to generate a request for analyzing the email body or URL, a means for the terminal to send the request to an AI model to analyze the possibility of fraud, a means for the terminal to detect a user's emotional state using a camera and microphone, a means for an emotion engine to analyze the user's emotional state, and a means for the terminal to notify a user of a warning based on the analysis results of the AI ​​model and the emotion analysis results of the emotion engine. This enables highly accurate detection of fraudulent activities and the provision of a warning message adapted to the user's emotional state.

[0890] A "terminal" is an electronic device operated by a user, and is a device for receiving emails and browsing the web.

[0891] The "mail body" is the content of the email, and is information that includes text, images, links, and the like.

[0892] "URL" is a Uniform Resource Locator that indicates the address of a web page, and is an identifier for specifying the destination of browsing access.

[0893] An "AI model" is an analytical system that uses artificial intelligence algorithms and is a model trained to perform a specific task.

[0894] An "analysis request" is a command sent from a device to an AI model that contains the information necessary to analyze specific data.

[0895] "Fraudulence" is a probability or score that indicates whether an email or website is attempting to commit fraud.

[0896] The "emotional state" indicates the psychological state of the user, and is an emotion judged from facial expressions, voice, etc.

[0897] An "emotion engine" is software that analyzes a user's emotional state from their facial expressions and voice.

[0898] A "warning message" is a notification message that alerts users to risks and caution, and is in the form of text or audio to notify them of possible fraudulent activity.

[0899] A "camera" is a device that captures images and videos and is used to detect a user's facial expression.

[0900] A "microphone" is a device that records sound and is used to detect the user's voice.

[0901] This invention realizes a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and then analyzes the user's emotional state and responds appropriately. The specific hardware and software that make up the system, as well as the actual processing procedures, are described below.

[0902] System Configuration

[0903] The system consists of the following major hardware and software components:

[0904] Device: An electronic device operated by a user, such as a smartphone, tablet, or computer.

[0905] Camera: A built-in camera used to detect the user's facial expressions.

[0906] Microphone: A microphone built into the device, used to detect the user's voice.

[0907] AI model: An artificial intelligence model that analyzes the content of emails and URLs to determine the likelihood of fraud.

[0908] Emotion engine: Software for analyzing the user's emotional state from their facial expressions and voice.

[0909] Natural language processing technology: Used to analyze email text and URLs.

[0910] System operation explanation

[0911] The device receives an email or URL: When a user receives a new email or clicks on a new website URL, the device recognizes this.

[0912] Generating an analysis request: The device generates an analysis request to the AI ​​model based on the received email body and URL. This request includes detailed information about the email content and URL.

[0913] Performing fraud analysis: The device sends an analysis request to the AI ​​model, which uses natural language processing techniques to analyze the content of emails and URLs and calculates a score indicating the likelihood of fraud.

[0914] Emotional state detection: The device activates the camera and microphone to collect the user's facial expressions and voice in real time, and the emotion engine then analyzes this data to determine the user's emotional state.

[0915] Integrated display of warning messages: The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to adjust the content and display method of the warning message displayed to the user. For example, if the user is surprised, a more detailed warning message will be displayed, and if the user is calm, a simple message will be displayed.

[0916] Examples and prompts

[0917] For example, if a user receives an email that reads, "Congratulations! You have won a $1000 gift card. Click here to claim.", the device will generate the following prompt:

[0918] Given the following email content and header, detect if it's a phishing email and provide a confidence score. Email content: "Congratulations! You have won a $1000 gift card. Click here to claim." Email header: "From: no-reply@fakesite.com, Subject: Claim your prize"

[0919] This prompt is sent to an AI model that analyzes the email for its potential fraudulent nature. Based on the analysis results and the user's emotional state, a warning message such as "This email may be fraudulent. Please investigate further" is displayed.

[0920] In this way, the system can achieve highly accurate detection of fraudulent activity and present warning messages that adapt to the user's emotional state.

[0921] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0922] Step 1:

[0923] The device receives a new email or URL. By receiving an email or clicking a URL, the device obtains this information. The input is the new email or the accessed URL, and the output is the email body or URL to be analyzed.

[0924] Step 2:

[0925] The device generates a request to analyze the email body and URL. Specifically, it generates a prompt message containing detailed information about the email content and URL. The input is the email body and URL, and the output is an analysis request to send to the AI ​​model.

[0926] Step 3:

[0927] The analysis request generated by the device is sent to an AI model to analyze the likelihood of fraud. The AI ​​model uses natural language processing techniques to analyze the content of emails and URLs and generate a score indicating fraud. The input is the analysis request, and the output is a fraud likelihood score.

[0928] Step 4:

[0929] The device detects the user's emotional state using a camera and microphone. Specifically, the camera captures the user's facial expressions and the microphone records their voice. The emotion engine analyzes this data and determines the user's emotions. The input is camera images and voice data, and the output is the user's emotional state.

[0930] Step 5:

[0931] The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to determine the most appropriate content and display method of the warning message. The input is the fraud probability score and emotional state, and the output is an optimized warning message.

[0932] Step 6:

[0933] The terminal displays a warning message to the user. The content of the message is a warning about the risk of fraud, and appropriate wording is used depending on the user's emotional state. The input is the optimized warning message, and the output is the warning information that the user confirms.

[0934] Step 7:

[0935] The user checks the warning message and takes appropriate action to avoid the risk of fraud. For example, the user may delete the email, not click on the link, or restrict access to the site. The input is the warning message, and the output is the user's appropriate response.

[0936] 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.

[0937] 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.

[0938] 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.

[0939] [Third embodiment]

[0940] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0941] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0942] 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).

[0943] 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.

[0944] 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.

[0945] 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).

[0946] 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.

[0947] 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.

[0948] 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.

[0949] 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.

[0950] 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.

[0951] 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."

[0952] The present invention provides a system for automatically detecting fraudulent emails and websites in email applications and web browsing applications, and for issuing warnings to users. The processing of a program for implementing this system is specifically explained below in natural language.

[0953] System Overview

[0954] The system is embedded within email or web browsing applications and analyzes incoming emails and visited URLs to determine potential fraudulent activity using AI models running on the device.

[0955] Example of an email application

[0956] 1. Receiving emails and notifications

[0957] When a user starts an email application, the server receives new emails and transfers them to the terminal.

[0958] The terminal notifies the user that new mail has arrived.

[0959] 2. Generate email analysis request

[0960] When a user opens a new email, the device extracts the email body and header information, including the sender's email address, subject, body content, links, and attachment information.

[0961] The device generates an analysis request to the AI ​​model based on this information.

[0962] 3. Analysis using AI models

[0963] The device sends the analysis request it generates to the AI ​​model.

[0964] The AI ​​model analyzes the email body and links, and uses natural language processing techniques to examine the content, comparing it with known phishing and spam email characteristics to detect suspicious phrases and links.

[0965] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[0966] 4. Receiving analysis results and generating alerts

[0967] The device receives the analysis results from the AI ​​model.

[0968] Based on the analysis results, the device will issue a warning to the user. For example, if the analysis results indicate a high probability of fraud, the device will display a warning message saying, "This email may be fraudulent."

[0969] 5. User Response

[0970] Users should acknowledge the warning and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[0971] Example of a web browsing application

[0972] 1. Accessing the URL

[0973] A user clicks on a new URL in a web browsing application.

[0974] The device extracts the URL.

[0975] 2. URL parsing request generation

[0976] The device generates an analysis request to send the URL information to the AI ​​model.

[0977] 3. Analysis using AI models

[0978] The device sends the analysis request it generates to the AI ​​model.

[0979] The AI ​​model analyzes the URL destination (web page content) and compares it with known phishing sites and malicious domains.

[0980] 4. Receiving analysis results and generating alerts

[0981] If the device receives the analysis results from the AI ​​model and determines that there is a high possibility of fraud, the device will display a warning saying, "This site may be fraudulent."

[0982] 5. User Response

[0983] Users should check the warning and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[0984] In this way, the present invention is a system that automatically detects fraud risks in email and web browsing applications and warns users, preventing fraud damage before it occurs. In particular, by using an AI model, it achieves more accurate analysis than conventional simple filtering.

[0985] The processing flow will be explained below.

[0986] Step 1:

[0987] The server receives new emails and forwards them to the device, which causes the email application to recognize that new emails have arrived.

[0988] Step 2:

[0989] The device will notify the user of new emails. A notification will pop up to let you know that new email has arrived.

[0990] Step 3:

[0991] The user clicks to open a new email, which starts the process of displaying the email contents.

[0992] Step 4:

[0993] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[0994] Step 5:

[0995] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[0996] Step 6:

[0997] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[0998] Step 7:

[0999] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[1000] Step 8:

[1001] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[1002] Step 9:

[1003] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[1004] Step 10:

[1005] The device receives the analysis results and score and generates a warning to the user based on the results. For example, if there is a high possibility of fraud, it generates a warning message such as "This email may be fraudulent."

[1006] Step 11:

[1007] The device displays a warning message to the user. A message is displayed on the user interface to inform the user of the danger.

[1008] Step 12:

[1009] Users can prevent themselves from falling victim to fraud by checking the warning message and taking action such as deleting the email or not clicking on the link.

[1010] In this way, by performing specific processing at each step, a system is realized that can reliably detect fraudulent emails and warn users.

[1011] Example 1

[1012] 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."

[1013] Conventional email and web browsing applications have difficulty detecting fraud risks and providing appropriate warnings to users. This puts users at high risk of being exposed to phishing emails and fraudulent websites. The present invention aims to reduce such risks and enable users to safely communicate electronically and browse the web.

[1014] 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.

[1015] In this invention, the server includes means for a terminal to receive data related to electronic communications or websites, means for the terminal to generate a request for analyzing the data, means for transmitting the request to a machine learning model to analyze the possibility of fraud, means for notifying a user of a warning based on the analysis result, and means for guiding the user's operation based on the warning, thereby enabling the user to quickly recognize the risk of fraud and take appropriate measures.

[1016] "Device" refers to a hardware device used for electronic communication or website access, including, for example, a computer, smartphone, or tablet.

[1017] "Electronic Communications" means information such as text, images, links, etc. sent or received via email or messaging applications.

[1018] "Website" means a collection of web pages publicly available on the Internet and is a resource of information accessible to users.

[1019] "Data" refers to any information received from an electronic communication or website, including text, images, links, and metadata.

[1020] A "machine learning model" is an algorithm that analyzes large amounts of data, learns patterns and features from that data, and predicts and classifies new data.

[1021] "Natural language processing technology" is a general term for technologies that allow computers to understand, generate, and analyze human language, and includes text analysis, translation, and dialogue systems.

[1022] "Analysis Request" refers to a packet of information sent to a machine learning model to analyze data.

[1023] "Fraud" refers to malicious acts such as phishing emails and fraudulent websites that attempt to illegally obtain users' personal information or money.

[1024] "Warning" refers to a message that alerts the user based on the analysis results and provides information to help avoid risky behavior.

[1025] "Guiding user operations" means showing the user appropriate actions to take based on the warning, such as deleting the email or avoiding clicking on links.

[1026] The present invention provides a system for automatically detecting fraudulent emails and fraudulent websites in email applications and web browsing applications, and for issuing a warning to the user. A detailed description of how to implement the system of the present invention is provided below.

[1027] Hardware and software used

[1028] Hardware: Devices (computers, smartphones, tablets, etc.)

[1029] software:

[1030] Email application

[1031] Web browsing applications

[1032] Machine learning models (e.g., the BERT model, which is specialized for natural language processing)

[1033] System Overview

[1034] The system is embedded within email or web browsing applications and uses machine learning models running on the device to analyze emails received and website URLs visited in real time to identify potential fraudulent activity.

[1035] Example from an email application

[1036] Receiving emails and notifications

[1037] The server receives the user's email and forwards new emails to the terminal.

[1038] The device will notify the user of new emails by sounding an alert and displaying a pop-up message.

[1039] Generate email analysis request

[1040] When a user opens an email application and clicks on a new email, the device extracts the email body and header information.

[1041] Based on the information extracted by the device, an analysis request is generated for the machine learning model.

[1042] Analysis using machine learning models

[1043] Sends device-generated analysis requests to machine learning models.

[1044] Machine learning models analyze email text and links, and use natural language processing techniques to inspect the content.

[1045] The model matches the characteristics of known phishing and spam emails to detect suspicious phrases and links.

[1046] The analysis generates a score indicating the likelihood of fraud.

[1047] Receiving analysis results and generating alerts

[1048] The device receives the analysis results and issues a warning to the user. For example, if it determines that the email is likely to be fraudulent, it displays a warning message saying, "This email may be fraudulent."

[1049] User response

[1050] Users acknowledge the warning and take action to reduce the risk of fraud, such as deleting the email or not clicking on the link.

[1051] Example of a web browsing application

[1052] URL access

[1053] When a user clicks on a new URL in a web browsing application, the device extracts the URL.

[1054] URL parsing request generation

[1055] The device generates an analysis request to send the URL information to the machine learning model.

[1056] Analysis using machine learning models

[1057] Sends device-generated analysis requests to machine learning models.

[1058] Machine learning models analyze the URL destination (web page content) and compare it with known phishing sites and malicious domains.

[1059] Receiving analysis results and generating alerts

[1060] If the device receives the analysis results and determines that there is a high possibility of fraud, it will display a warning message saying, "This site may be fraudulent."

[1061] User response

[1062] Users can acknowledge the warning and take action to avoid the risk of fraud, such as not visiting the site or hiding it.

[1063] Examples of specific examples and prompts

[1064] Example of an email application

[1065] When a user opens a new email containing an "Unsubscribe" link, a machine learning model identifies the link as a phishing risk and displays a warning.

[1066] Prompt Sentence Examples

[1067] Please check the safety of this email before clicking any links, as it is likely a scam.

[1068] Example of a web browsing application

[1069] When a user attempts to visit a specific online shopping site, a machine learning model identifies the site as a known fraudulent site and displays a warning.

[1070] Prompt Sentence Examples

[1071] This site may be a scam. Please consider other safe ways to purchase.

[1072] This invention realizes a system that can instantly detect fraudulent behavior during email and web browsing, allowing users to avoid the risks of phishing emails and fraudulent sites. In particular, by using machine learning models, it is possible to perform more accurate analysis than conventional methods.

[1073] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1074] (Processing flow in email application)

[1075] Step 1:

[1076] The server receives the user's email and forwards new emails to the device.

[1077] Input: User's email account information, new emails

[1078] Output: New email notification to device

[1079] Specific operation: Records emails from various mail servers (IMAP / POP3) in the receiving log and forwards them to the mail application.

[1080] Step 2:

[1081] The device will notify the user that new email has arrived.

[1082] Input: New email

[1083] Output: New mail notification to user

[1084] Specific operation: On the smartphone, the message "You have new email" will appear in the notification bar and a notification sound will be heard on the device itself.

[1085] Step 3:

[1086] The user opens their email application and clicks on new email.

[1087] The device extracts the email body and header information.

[1088] Input: New emails opened by the user, including the email body and header information

[1089] Output: Extracted email body and header information

[1090] What happens: The sender address, subject, body, links, and attachment information are separated for analysis.

[1091] Step 4:

[1092] Based on the information extracted by the device, an analysis request is generated for the AI ​​model.

[1093] Input: Extracted email body and header information

[1094] Output: Analysis request to the AI ​​model

[1095] Specific operation: The analysis request is structured in JSON format or other format and prepared for sending to the machine learning model.

[1096] Step 5:

[1097] The device generates an analysis request and sends it to the AI ​​model.

[1098] Input: Analysis request to the AI ​​model

[1099] Output: Sending data to the AI ​​model

[1100] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[1101] Step 6:

[1102] The AI ​​model analyzes the email body and links, and uses natural language processing technology to inspect the content.

[1103] Input: Analysis request email body and link

[1104] Output: Analysis results (fraud likelihood score)

[1105] What it does: Contextual analysis, spam filtering, and matching links to known phishing phrases.

[1106] Step 7:

[1107] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[1108] Input: Parsed email body and links

[1109] Output: Analysis results with fraud probability score

[1110] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[1111] Step 8:

[1112] The device receives the analysis results from the AI ​​model.

[1113] Input: Analysis results with fraud likelihood score

[1114] Output: A warning notice to the user

[1115] Specific Actions: Based on the estimated risk, a warning message is prepared and displayed in the user interface.

[1116] Step 9:

[1117] The device will display a warning to the user.

[1118] Input: warning message

[1119] Output: A warning notice to alert the user

[1120] Specific behavior: If the email is rated as "likely fraudulent," a pop-up message will appear saying "This email may be fraudulent."

[1121] Step 10:

[1122] Users should acknowledge the warning and take action to avoid the risk of fraud.

[1123] Input: warning message

[1124] Output: User response (delete email, avoid clicking links)

[1125] Specific action: The user clicks the "Delete" button to delete the email.

[1126] (Processing flow in a web browsing application)

[1127] Step 1:

[1128] A user clicks on a new URL in their web browsing application.

[1129] Input: User clicks on a URL

[1130] Output: Extracted URLs

[1131] What it does: The browser captures the click event and records the URL being accessed.

[1132] Step 2:

[1133] The device will extract the URL.

[1134] Input: Clicked URL

[1135] Output: URL for analysis

[1136] What it does: Saves the URL as structured data.

[1137] Step 3:

[1138] The device generates an analysis request to send the URL information to the AI ​​model.

[1139] Input: A structured URL to parse

[1140] Output: Analysis request to the AI ​​model

[1141] Specific operation: Structure the analysis request in JSON format or similar.

[1142] Step 4:

[1143] The device generates an analysis request and sends it to the AI ​​model.

[1144] Input: Analysis request to the AI ​​model

[1145] Output: Sending data to the AI ​​model

[1146] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[1147] Step 5:

[1148] The AI ​​model analyzes the URL destination (webpage content) and compares it with known phishing sites and malicious domains.

[1149] Input: URL of the analysis request

[1150] Output: Analysis results

[1151] What it does: Scrape the content of web pages and compare it to a database of known phishing sites.

[1152] Step 6:

[1153] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[1154] Input: scraped web page content

[1155] Output: Analysis results with fraud probability score

[1156] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[1157] Step 7:

[1158] The device receives the analysis results and notifies the user with a warning.

[1159] Input: Analysis results with fraud likelihood score

[1160] Output: A warning notice to the user

[1161] Specific behavior: If there is a high possibility of fraud, a warning message will be displayed in a pop-up saying, "This site may be fraudulent."

[1162] Step 8:

[1163] Users should acknowledge the warning and take action to avoid the risk of fraud.

[1164] Input: warning message

[1165] Output: User response (stops visiting the site)

[1166] Specific Action: The user closes the browser tab or stops visiting the site.

[1167] (Application example 1)

[1168] 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."

[1169] In today's world, as internet-based communications become increasingly common, the number of victims of fraudulent emails and phishing sites is rapidly increasing. These fraudulent activities often result in the leakage of personal information and financial losses, posing a significant risk to many users. Current technology makes it difficult to detect these fraudulent activities in advance and implement countermeasures, creating a demand for systems with advanced analytical capabilities. The purpose of this invention is to solve these problems and enable users to use email and websites safely.

[1170] 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.

[1171] In this invention, the server includes: a terminal receiving an email body or URL; a terminal generating a request for analyzing the email body or URL; a terminal sending the request to an AI model to analyze the possibility of fraud; a terminal issuing a warning to the user based on the analysis results; a smartphone installed on the smartphone for analyzing emails and websites in real time; and a terminal displaying a warning if the possibility of fraud is high based on the analysis results. This enables sophisticated analysis of fraudulent emails and phishing sites and displaying a warning to the user in real time.

[1172] A "terminal" is an electronic device such as a smartphone, tablet, or PC that is operated by a user.

[1173] "Email body" refers to the body of the message written by the sender of the email, which may contain text and links.

[1174] "URL" stands for Uniform Resource Locator and indicates the address of a web page on the Internet.

[1175] An "analysis request" is request data generated to allow an AI model to determine the possibility of fraud based on the content of an email body or URL.

[1176] An "AI model" is an analytical engine that utilizes artificial intelligence technology and is designed to detect fraudulent emails and phishing sites.

[1177] "Natural language processing" is a technique in which AI models understand and analyze human language, and is used to scrutinize the content of emails and web pages.

[1178] A "warning message" is a notification message that is displayed to warn the user of possible fraud based on the analysis results.

[1179] "Real-time" refers to the instantaneous reception and analysis of data, allowing users to obtain results immediately.

[1180] This invention is a system that automatically detects fraudulent emails and websites in email and web browsing applications and issues a warning to the user. This system is mainly composed of a terminal and a server, and the terminal receives emails and URLs, analyzes their contents, and notifies the user.

[1181] Program processing explanation

[1182] The server first receives the email body or URL sent from the device. The email body includes the sender address, subject, body text, links, etc. The server generates a request to analyze this content and sends it to an AI model. This AI model uses natural language processing (NLP) technology and is designed to detect potential fraud and phishing. Specifically, it analyzes the content by comparing it with keywords and phrases commonly found in fraudulent emails, as well as information about the domains to which the links lead.

[1183] As a result of the analysis, the AI ​​model assigns a score indicating the likelihood of fraud and sends it back to the server. The server generates and notifies the user of a warning message based on this score. For example, if the score is high, a warning such as "This email may be fraudulent" or "This site may be fraudulent" will be displayed.

[1184] The system of the present invention mainly uses the following hardware and software. The terminal is a general computing device such as a smartphone or tablet. The server can be a cloud server or a dedicated server in a data center. In addition, the AI ​​model uses a deep learning framework such as TensorFlow or PyTorch.

[1185] Specific examples

[1186] For example, when a user receives an email disguised as a message from a bank, the body of the email often contains keywords such as "transfer" and "password change." When the user opens the email, the device sends the email body to a server, where the AI ​​model analyzes these keywords. If the analysis returns a score of "very likely to be fraudulent," the user is shown a warning message saying, "This email may be fraudulent."

[1187] Prompt Sentence Examples

[1188] 1. Enter the sender, subject, and body of the email you received and assess its likelihood of fraud.

[1189] 2. "Analyze the URL destination to determine whether it is phishing."

[1190] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1191] Step 1:

[1192] The device receives the email or URL.

[1193] What it does: The device receives new emails or website URLs via applications the user has opened. For emails, information such as the sender, subject, body, and links is extracted.

[1194] Input: Received email body or URL

[1195] Output: Extracted email and URL information

[1196] Step 2:

[1197] The device generates an analysis request.

[1198] What it does: Based on the information in the received email or URL, it creates an analysis request to send to the AI ​​model. This request includes the text of the email body and the URL information.

[1199] Input: Extracted email and URL information

[1200] Output: Analysis request (email body text, URL information, etc.)

[1201] Step 3:

[1202] The device sends an analysis request to the server.

[1203] Specific operation: The generated analysis request is sent to the server, which receives it and prepares to pass the request to the AI ​​model.

[1204] Input: Parse request

[1205] Output: The parsing request sent to the server

[1206] Step 4:

[1207] The server sends an analysis request to the AI ​​model and begins the analysis.

[1208] How it works: When the server receives an analysis request, it passes it to the AI ​​model, which uses natural language processing techniques to analyze the content of the email body and URLs to detect potential fraud.

[1209] Input: Parse request

[1210] Output: Analysis results (e.g., fraud probability score)

[1211] Step 5:

[1212] The server receives the analysis results and sends them to the terminal.

[1213] How it works: The server receives the analysis results returned by the AI ​​model and sends them to the device, which includes a score indicating the likelihood of fraud.

[1214] Input: Analysis results (e.g., fraud likelihood score)

[1215] Output: Analysis results sent to the device

[1216] Step 6:

[1217] The device notifies the user of a warning based on the analysis results.

[1218] Specific operation: When the device receives the analysis results, if there is a high possibility of fraud, it will display a warning message to the user. The warning message will include content such as "This email may be fraudulent" or "This site may be fraudulent."

[1219] Input: Analysis result (fraud likelihood score)

[1220] Output: The warning message displayed to the user.

[1221] 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.

[1222] This invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. It also incorporates an emotion engine to understand the user's emotional state and take appropriate action. Below, we will explain in detail the processing of the program for implementing this system.

[1223] System Overview

[1224] This system is embedded in email or web browsing applications, analyzes incoming emails and accessed URLs, and determines whether fraud is likely. It also uses an emotion engine to recognize the user's emotional state and adjust the display method and content of warning messages accordingly.

[1225] Example of an email application

[1226] 1. Receiving emails and notifications

[1227] The server receives new emails and forwards them to the terminal.

[1228] The terminal notifies the user that new mail has arrived.

[1229] 2. Generate email analysis request

[1230] The user clicks to open a new email.

[1231] The terminal extracts the email body and email header.

[1232] The device generates an analysis request to the AI ​​model based on the information extracted.

[1233] 3. Analysis using AI models

[1234] The device sends the analysis request generated to the AI ​​model.

[1235] The AI ​​model analyzes the email text and links, and uses natural language processing techniques to detect suspicious phrases and links.

[1236] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[1237] 4. Emotion Recognition by Emotion Engine

[1238] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[1239] An emotion engine analyzes this data to determine the user's emotional state.

[1240] 5. Integration of analysis results and emotion recognition results

[1241] The device receives the analysis results from the AI ​​model and the emotion recognition results from the emotion engine.

[1242] Based on these results, the terminal determines the content and display method of a warning message to the user.

[1243] 6. Displaying a warning message

[1244] The device will display a warning message to the user. For example, if the user is nervous, it will display a friendly message such as "This email may be fraudulent. Please proceed with caution."

[1245] If the user is calm, a simple message such as "This email may be fraudulent" will be displayed.

[1246] 7. User Response

[1247] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[1248] Example of a web browsing application

[1249] 1. Accessing the URL

[1250] A user clicks on a new URL in a web browsing application.

[1251] The device extracts the URL.

[1252] 2. URL parsing request generation

[1253] The device generates an analysis request to the AI ​​model based on the URL information.

[1254] 3. Analysis using AI models

[1255] The device sends the analysis request generated to the AI ​​model.

[1256] An AI model analyzes URL destinations and compares them with known phishing sites and malicious domains.

[1257] 4. Emotion Recognition by Emotion Engine

[1258] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[1259] An emotion engine analyzes this data to determine the user's emotional state.

[1260] 5. Integration of analysis results and emotion recognition results

[1261] The device receives the analysis results and emotion recognition results from the AI ​​model and determines the content and display method of the warning message to the user based on these.

[1262] 6. Displaying a warning message

[1263] The device displays a warning message to the user, for example, if the user is surprised, it displays a descriptive message such as "This site may be fraudulent. Please investigate further."

[1264] If the user is calm, a simple message such as "This site may be a scam" will be displayed.

[1265] 7. User Response

[1266] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[1267] Thus, the present invention is a system that recognizes a user's emotional state in email and web browsing applications and adjusts the content and display of warning messages accordingly, allowing users to more effectively avoid the risk of fraud.

[1268] The processing flow will be explained below.

[1269] Step 1:

[1270] The server receives new emails and forwards them to the device, which then recognizes that new emails have been received.

[1271] Step 2:

[1272] The device will notify the user of new mail. A notification will pop up to let you know that new mail has arrived.

[1273] Step 3:

[1274] The user clicks to open a new email, which starts the process of displaying the email contents.

[1275] Step 4:

[1276] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[1277] Step 5:

[1278] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[1279] Step 6:

[1280] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[1281] Step 7:

[1282] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[1283] Step 8:

[1284] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[1285] Step 9:

[1286] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[1287] Step 10:

[1288] The device receives the analysis results from the AI ​​model and generates a warning for the user based on the results.

[1289] Step 11:

[1290] The device activates an emotion engine to recognize the user's emotions, and collects the user's facial expressions and voice using the device's camera or microphone.

[1291] Step 12:

[1292] The emotion engine analyzes the collected data to determine the user's current emotional state, categorizing it into emotion categories such as nervous, angry, surprised, or calm.

[1293] Step 13:

[1294] The device combines the analysis results of the AI ​​model with the emotion recognition results of the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it will select a friendly and reassuring message.

[1295] Step 14:

[1296] The device will then display appropriate warning messages to the user, adjusting the tone and details of the messages to match the user's emotional state.

[1297] Step 15:

[1298] Users should review the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[1299] In this way, by performing specific processing at each step, a system is realized that not only reliably detects fraudulent emails and warns users, but also uses an emotion engine to take appropriate action taking into account the user's emotional state.

[1300] Example 2

[1301] 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."

[1302] In today's digital communications, damage caused by fraudulent emails and phishing sites is increasing. In particular, these fraudulent activities use sophisticated techniques, making them difficult to detect with conventional security systems. Furthermore, since warning messages are provided uniformly without taking into account the user's emotional state, many users ignore the warnings or fail to take appropriate action, which is a problem. Therefore, there is a need for a system that provides more appropriate warnings by taking into account the user's emotional state.

[1303] 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.

[1304] In this invention, the terminal includes means for receiving an email body or a URL, means for generating a request for analyzing the email body or the URL, means for sending a request to a generative AI model and analyzing the possibility of fraud, means for detecting the user's facial expressions and voice using a camera or microphone and activating an emotion engine that determines the user's emotional state, means for notifying the user of a warning based on the analysis result and the emotional state, means for determining the content and display method of the warning message based on the analysis result and the emotion recognition result, and means for displaying a warning message customized according to the user's emotional state. This makes it possible to effectively notify the user of the risk of fraud and encourage appropriate action while taking the user's emotional state into consideration.

[1305] "Terminal" refers to a computing device used by a user, on which an email application or a web browsing application runs.

[1306] The "mail body" refers to the main content of an email sent or received by a user, and is the part that contains the written message or content.

[1307] "URL" refers to the address that specifies the destination of a link to a web page or resource.

[1308] "Means for generating a request" refers to the function for constructing and generating the request data required to analyze the email body or URL.

[1309] A "generative AI model" refers to an artificial intelligence model that uses machine learning technology to analyze email text and URLs to assess the likelihood of fraud.

[1310] "Means for analyzing" refers to the ability of the generative AI model to assess the likelihood of fraud based on the generated request.

[1311] An "emotion engine" refers to a system that uses a camera or microphone to capture a user's facial expressions and voice, and analyzes that data to determine the user's emotional state.

[1312] "Means for notifying warnings" refers to the ability to display warnings to users based on the likelihood of fraud or their emotional state.

[1313] "Means for determining the content and display method of the warning message" refers to a function that takes into consideration the analysis results and emotion recognition results and determines what kind of warning message to display to the user and how.

[1314] A "customized alert message" refers to an alert message whose content and presentation are tailored to the user's current emotional state.

[1315] The present invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. Furthermore, by combining it with an emotion engine, the system grasps the user's emotional state and takes appropriate action. Detailed embodiments for implementing this system are described below.

[1316] The system is integrated into email or web browsing applications running on a device. The device requires a camera, microphone, display, and internet connection for its hardware. The software includes email client software, a web browser, a generative AI model for natural language processing (NLP), and an emotion engine for analyzing the user's emotional state.

[1317] First, the server receives a new email and forwards it to the device. The device displays a notification of the new email to the user. When the user opens the email, the device extracts the email body and headers. Based on this, the device generates an analysis request for the AI ​​model and sends it to the generative AI model.

[1318] The generative AI model analyzes the email body and links, using natural language processing to detect suspicious phrases and links, and assigns a score indicating the likelihood of fraud. The analysis results are then sent back to the device.

[1319] In parallel, the device activates an emotion engine that uses the camera and microphone to detect the user's facial expressions and voice, and determines the user's emotional state based on the collected data.

[1320] The device integrates the analysis results from the AI ​​model and the emotion recognition results from the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it may display "This email may be fraudulent. Please proceed with caution," but if the user is calm, it may simply display "This email may be fraudulent." The device displays an appropriate warning message to the user, who can then confirm the message and take appropriate action to avoid the risk of fraud.

[1321] It works similarly in web browsing applications. When a user clicks on a new URL, the device extracts the URL and generates an analysis request from the AI ​​model. The generative AI model analyzes the URL's destination and compares it with known phishing sites and malicious domains. The emotion engine determines the user's emotional state and, based on the analysis and emotion recognition results, determines the content and display method of the warning message. The device then displays an appropriate warning to the user and prevents them from accessing the dangerous site.

[1322] A specific example of this is the prompt:

[1323] Email applications:

[1324] "This email may be fraudulent. Please review it carefully and verify with other sources before proceeding."

[1325] Web browsing applications:

[1326] "This site may be a scam. Please avoid visiting."

[1327] This system provides warning messages that take into account the user's emotional state, allowing for more effective prevention of fraud risks.

[1328] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1329] Step 1: Receiving emails and notifications

[1330] The server receives new emails. The input is the new email and the output is the received email data. Specifically, the server receives new emails using the SMTP protocol and saves them in the user's mailbox. The server then transfers this email data to the terminal. The terminal displays a pop-up notification of new emails to the user. The input is the email data received from the server and the output is the notification to the user.

[1331] Step 2: Generate an email analysis request

[1332] The user clicks to open a new email. The input is the user's click action, and the output is the email body and email header data. The device extracts the email body and email header. The device obtains this data using the email client's internal API and generates an analysis request for the generative AI model. The input is the extracted email data, and the output is the analysis request.

[1333] Step 3: Analysis by AI model

[1334] The device sends the analysis request it generated to the generative AI model. The input is the analysis request, and the output is the analysis result. Specifically, data is sent using an HTTP POST request. The generative AI model analyzes the email body and links, and uses natural language processing technology to detect suspicious phrases and links. The output is an analysis result with a score for the likelihood of fraud. The analysis result is returned to the device in JSON format.

[1335] Step 4: Emotion Recognition with the Emotion Engine

[1336] The device starts the emotion engine and detects the user's facial expressions and voice using a camera or microphone. The input is real-time data of the user's facial expressions and voice, and the output is the analyzed emotional state. The emotion engine analyzes this data and determines the user's emotional state. The output is the user's emotional state (e.g., nervous, surprised, calm, etc.).

[1337] Step 5: Integrating analysis results with emotion recognition results

[1338] The device receives the analysis results from the generative AI model and the emotion recognition results from the emotion engine. The inputs are the analysis results and emotion recognition results, and the output is the integrated information used for notifications. Specifically, the device runs an algorithm based on the score of the analysis results and the emotion recognition results to determine the content and display method of the warning message to the user.

[1339] Step 6: Displaying warning messages

[1340] The terminal displays a customized warning message to the user. The input is the integrated information, and the output is a warning message to the user. The terminal uses a widget in an email client or web browser to display a pop-up message on the screen. For example, if the user is nervous, it displays "This email may be fraudulent. Please proceed with caution," and if the user is calm, it displays a simple message saying "This email may be fraudulent."

[1341] Step 7: User Action

[1342] The user checks the warning message and takes appropriate action to avoid the risk of fraud. The input is the warning message, and the output is the user's action. Specifically, the user may take action such as deleting the email or not clicking on the link. For example, if the user sees the warning message and selects "Delete fraudulent email," the email will be deleted.

[1343] (Application example 2)

[1344] 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."

[1345] Conventional email and web browsing applications are ineffective at detecting fraudulent emails and websites, and warnings to users are provided in a generic format that is not adapted to the user's emotional state. As a result, users are at increased risk of falling victim to fraud because they are unable to take appropriate action. Therefore, the present invention aims to improve the accuracy of fraud detection and provide warnings that are adapted to the user's emotional state, encouraging the user to take more effective action.

[1346] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1347] In this invention, the server includes, in an email or web browsing application, a means for a terminal to receive an email body or URL, a means for the terminal to generate a request for analyzing the email body or URL, a means for the terminal to send the request to an AI model to analyze the possibility of fraud, a means for the terminal to detect a user's emotional state using a camera and microphone, a means for an emotion engine to analyze the user's emotional state, and a means for the terminal to notify a user of a warning based on the analysis results of the AI ​​model and the emotion analysis results of the emotion engine. This enables highly accurate detection of fraudulent activities and the provision of a warning message adapted to the user's emotional state.

[1348] A "terminal" is an electronic device operated by a user, and is a device for receiving emails and browsing the web.

[1349] The "mail body" is the content of the email, and is information that includes text, images, links, and the like.

[1350] "URL" is a Uniform Resource Locator that indicates the address of a web page, and is an identifier for specifying the destination of browsing access.

[1351] An "AI model" is an analytical system that uses artificial intelligence algorithms and is a model trained to perform a specific task.

[1352] An "analysis request" is a command sent from a device to an AI model that contains the information necessary to analyze specific data.

[1353] "Fraudulence" is a probability or score that indicates whether an email or website is attempting to commit fraud.

[1354] The "emotional state" indicates the psychological state of the user, and is an emotion judged from facial expressions, voice, etc.

[1355] An "emotion engine" is software that analyzes a user's emotional state from their facial expressions and voice.

[1356] A "warning message" is a notification message that alerts users to risks and caution, and is in the form of text or audio to notify them of possible fraudulent activity.

[1357] A "camera" is a device that captures images and videos and is used to detect a user's facial expression.

[1358] A "microphone" is a device that records sound and is used to detect the user's voice.

[1359] This invention realizes a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and then analyzes the user's emotional state and responds appropriately. The specific hardware and software that make up the system, as well as the actual processing procedures, are described below.

[1360] System Configuration

[1361] The system consists of the following major hardware and software components:

[1362] Device: An electronic device operated by a user, such as a smartphone, tablet, or computer.

[1363] Camera: A built-in camera used to detect the user's facial expressions.

[1364] Microphone: A microphone built into the device, used to detect the user's voice.

[1365] AI model: An artificial intelligence model that analyzes the content of emails and URLs to determine the likelihood of fraud.

[1366] Emotion engine: Software for analyzing the user's emotional state from their facial expressions and voice.

[1367] Natural language processing technology: Used to analyze email text and URLs.

[1368] System operation explanation

[1369] The device receives an email or URL: When a user receives a new email or clicks on a new website URL, the device recognizes this.

[1370] Generating an analysis request: The device generates an analysis request to the AI ​​model based on the received email body and URL. This request includes detailed information about the email content and URL.

[1371] Performing fraud analysis: The device sends an analysis request to the AI ​​model, which uses natural language processing techniques to analyze the content of emails and URLs and calculates a score indicating the likelihood of fraud.

[1372] Emotional state detection: The device activates the camera and microphone to collect the user's facial expressions and voice in real time, and the emotion engine then analyzes this data to determine the user's emotional state.

[1373] Integrated display of warning messages: The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to adjust the content and display method of the warning message displayed to the user. For example, if the user is surprised, a more detailed warning message will be displayed, and if the user is calm, a simple message will be displayed.

[1374] Examples and prompts

[1375] For example, if a user receives an email that reads, "Congratulations! You have won a $1000 gift card. Click here to claim.", the device will generate the following prompt:

[1376] Given the following email content and header, detect if it's a phishing email and provide a confidence score. Email content: "Congratulations! You have won a $1000 gift card. Click here to claim." Email header: "From: no-reply@fakesite.com, Subject: Claim your prize"

[1377] This prompt is sent to an AI model that analyzes the email for its potential fraudulent nature. Based on the analysis results and the user's emotional state, a warning message such as "This email may be fraudulent. Please investigate further" is displayed.

[1378] In this way, the system can achieve highly accurate detection of fraudulent activity and present warning messages that adapt to the user's emotional state.

[1379] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1380] Step 1:

[1381] The device receives a new email or URL. By receiving an email or clicking a URL, the device obtains this information. The input is the new email or the accessed URL, and the output is the email body or URL to be analyzed.

[1382] Step 2:

[1383] The device generates a request to analyze the email body and URL. Specifically, it generates a prompt message containing detailed information about the email content and URL. The input is the email body and URL, and the output is an analysis request to send to the AI ​​model.

[1384] Step 3:

[1385] The analysis request generated by the device is sent to an AI model to analyze the likelihood of fraud. The AI ​​model uses natural language processing techniques to analyze the content of emails and URLs and generate a score indicating fraud. The input is the analysis request, and the output is a fraud likelihood score.

[1386] Step 4:

[1387] The device detects the user's emotional state using a camera and microphone. Specifically, the camera captures the user's facial expressions and the microphone records their voice. The emotion engine analyzes this data and determines the user's emotions. The input is camera images and voice data, and the output is the user's emotional state.

[1388] Step 5:

[1389] The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to determine the most appropriate content and display method of the warning message. The input is the fraud probability score and emotional state, and the output is an optimized warning message.

[1390] Step 6:

[1391] The terminal displays a warning message to the user. The content of the message is a warning about the risk of fraud, and appropriate wording is used depending on the user's emotional state. The input is the optimized warning message, and the output is the warning information that the user confirms.

[1392] Step 7:

[1393] The user checks the warning message and takes appropriate action to avoid the risk of fraud. For example, the user may delete the email, not click on the link, or restrict access to the site. The input is the warning message, and the output is the user's appropriate response.

[1394] 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.

[1395] 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.

[1396] 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.

[1397] [Fourth embodiment]

[1398] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1399] 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.

[1400] 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).

[1401] 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.

[1402] 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.

[1403] 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).

[1404] 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.

[1405] 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.

[1406] 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.

[1407] 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.

[1408] 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.

[1409] 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.

[1410] 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."

[1411] The present invention provides a system for automatically detecting fraudulent emails and websites in email applications and web browsing applications, and for issuing warnings to users. The processing of a program for implementing this system is specifically explained below in natural language.

[1412] System Overview

[1413] The system is embedded within email or web browsing applications and analyzes incoming emails and visited URLs to determine potential fraudulent activity using AI models running on the device.

[1414] Example of an email application

[1415] 1. Receiving emails and notifications

[1416] When a user starts an email application, the server receives new emails and transfers them to the terminal.

[1417] The terminal notifies the user that new mail has arrived.

[1418] 2. Generate email analysis request

[1419] When a user opens a new email, the device extracts the email body and header information, including the sender's email address, subject, body content, links, and attachment information.

[1420] The device generates an analysis request to the AI ​​model based on this information.

[1421] 3. Analysis using AI models

[1422] The device sends the analysis request it generates to the AI ​​model.

[1423] The AI ​​model analyzes the email body and links, and uses natural language processing techniques to examine the content, comparing it with known phishing and spam email characteristics to detect suspicious phrases and links.

[1424] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[1425] 4. Receiving analysis results and generating alerts

[1426] The device receives the analysis results from the AI ​​model.

[1427] Based on the analysis results, the device will issue a warning to the user. For example, if the analysis results indicate a high probability of fraud, the device will display a warning message saying, "This email may be fraudulent."

[1428] 5. User Response

[1429] Users should acknowledge the warning and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[1430] Example of a web browsing application

[1431] 1. Accessing the URL

[1432] A user clicks on a new URL in a web browsing application.

[1433] The device extracts the URL.

[1434] 2. URL parsing request generation

[1435] The device generates an analysis request to send the URL information to the AI ​​model.

[1436] 3. Analysis using AI models

[1437] The device sends the analysis request it generates to the AI ​​model.

[1438] The AI ​​model analyzes the URL destination (web page content) and compares it with known phishing sites and malicious domains.

[1439] 4. Receiving analysis results and generating alerts

[1440] If the device receives the analysis results from the AI ​​model and determines that there is a high possibility of fraud, the device will display a warning saying, "This site may be fraudulent."

[1441] 5. User Response

[1442] Users should check the warning and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[1443] In this way, the present invention is a system that automatically detects fraud risks in email and web browsing applications and warns users, preventing fraud damage before it occurs. In particular, by using an AI model, it achieves more accurate analysis than conventional simple filtering.

[1444] The processing flow will be explained below.

[1445] Step 1:

[1446] The server receives new emails and forwards them to the device, which causes the email application to recognize that new emails have arrived.

[1447] Step 2:

[1448] The device will notify the user of new emails. A notification will pop up to let you know that new email has arrived.

[1449] Step 3:

[1450] The user clicks to open a new email, which starts the process of displaying the email contents.

[1451] Step 4:

[1452] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[1453] Step 5:

[1454] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[1455] Step 6:

[1456] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[1457] Step 7:

[1458] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[1459] Step 8:

[1460] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[1461] Step 9:

[1462] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[1463] Step 10:

[1464] The device receives the analysis results and score and generates a warning to the user based on the results. For example, if there is a high possibility of fraud, it generates a warning message such as "This email may be fraudulent."

[1465] Step 11:

[1466] The device displays a warning message to the user. A message is displayed on the user interface to inform the user of the danger.

[1467] Step 12:

[1468] Users can prevent themselves from falling victim to fraud by checking the warning message and taking action such as deleting the email or not clicking on the link.

[1469] In this way, by performing specific processing at each step, a system is realized that can reliably detect fraudulent emails and warn users.

[1470] Example 1

[1471] 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."

[1472] Conventional email and web browsing applications have difficulty detecting fraud risks and providing appropriate warnings to users. This puts users at high risk of being exposed to phishing emails and fraudulent websites. The present invention aims to reduce such risks and enable users to safely communicate electronically and browse the web.

[1473] 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.

[1474] In this invention, the server includes means for a terminal to receive data related to electronic communications or websites, means for the terminal to generate a request for analyzing the data, means for transmitting the request to a machine learning model to analyze the possibility of fraud, means for notifying a user of a warning based on the analysis result, and means for guiding the user's operation based on the warning, thereby enabling the user to quickly recognize the risk of fraud and take appropriate measures.

[1475] "Device" refers to a hardware device used for electronic communication or website access, including, for example, a computer, smartphone, or tablet.

[1476] "Electronic Communications" means information such as text, images, links, etc. sent or received via email or messaging applications.

[1477] "Website" means a collection of web pages publicly available on the Internet and is a resource of information accessible to users.

[1478] "Data" refers to any information received from an electronic communication or website, including text, images, links, and metadata.

[1479] A "machine learning model" is an algorithm that analyzes large amounts of data, learns patterns and features from that data, and predicts and classifies new data.

[1480] "Natural language processing technology" is a general term for technologies that allow computers to understand, generate, and analyze human language, and includes text analysis, translation, and dialogue systems.

[1481] "Analysis Request" refers to a packet of information sent to a machine learning model to analyze data.

[1482] "Fraud" refers to malicious acts such as phishing emails and fraudulent websites that attempt to illegally obtain users' personal information or money.

[1483] "Warning" refers to a message that alerts the user based on the analysis results and provides information to help avoid risky behavior.

[1484] "Guiding user operations" means showing the user appropriate actions to take based on the warning, such as deleting the email or avoiding clicking on links.

[1485] The present invention provides a system for automatically detecting fraudulent emails and fraudulent websites in email applications and web browsing applications, and for issuing a warning to the user. A detailed description of how to implement the system of the present invention is provided below.

[1486] Hardware and software used

[1487] Hardware: Devices (computers, smartphones, tablets, etc.)

[1488] software:

[1489] Email application

[1490] Web browsing applications

[1491] Machine learning models (e.g., the BERT model, which is specialized for natural language processing)

[1492] System Overview

[1493] The system is embedded within email or web browsing applications and uses machine learning models running on the device to analyze emails received and website URLs visited in real time to identify potential fraudulent activity.

[1494] Example from an email application

[1495] Receiving emails and notifications

[1496] The server receives the user's email and forwards new emails to the terminal.

[1497] The device will notify the user of new emails by sounding an alert and displaying a pop-up message.

[1498] Generate email analysis request

[1499] When a user opens an email application and clicks on a new email, the device extracts the email body and header information.

[1500] Based on the information extracted by the device, an analysis request is generated for the machine learning model.

[1501] Analysis using machine learning models

[1502] Sends device-generated analysis requests to machine learning models.

[1503] Machine learning models analyze email text and links, and use natural language processing techniques to inspect the content.

[1504] The model matches the characteristics of known phishing and spam emails to detect suspicious phrases and links.

[1505] The analysis generates a score indicating the likelihood of fraud.

[1506] Receiving analysis results and generating alerts

[1507] The device receives the analysis results and issues a warning to the user. For example, if it determines that the email is likely to be fraudulent, it displays a warning message saying, "This email may be fraudulent."

[1508] User response

[1509] Users acknowledge the warning and take action to reduce the risk of fraud, such as deleting the email or not clicking on the link.

[1510] Example of a web browsing application

[1511] URL access

[1512] When a user clicks on a new URL in a web browsing application, the device extracts the URL.

[1513] URL parsing request generation

[1514] The device generates an analysis request to send the URL information to the machine learning model.

[1515] Analysis using machine learning models

[1516] Sends device-generated analysis requests to machine learning models.

[1517] Machine learning models analyze the URL destination (web page content) and compare it with known phishing sites and malicious domains.

[1518] Receiving analysis results and generating alerts

[1519] If the device receives the analysis results and determines that there is a high possibility of fraud, it will display a warning message saying, "This site may be fraudulent."

[1520] User response

[1521] Users can acknowledge the warning and take action to avoid the risk of fraud, such as not visiting the site or hiding it.

[1522] Examples of specific examples and prompts

[1523] Example of an email application

[1524] When a user opens a new email containing an "Unsubscribe" link, a machine learning model identifies the link as a phishing risk and displays a warning.

[1525] Prompt Sentence Examples

[1526] Please check the safety of this email before clicking any links, as it is likely a scam.

[1527] Example of a web browsing application

[1528] When a user attempts to visit a specific online shopping site, a machine learning model identifies the site as a known fraudulent site and displays a warning.

[1529] Prompt Sentence Examples

[1530] This site may be a scam. Please consider other safe ways to purchase.

[1531] This invention realizes a system that can instantly detect fraudulent behavior during email and web browsing, allowing users to avoid the risks of phishing emails and fraudulent sites. In particular, by using machine learning models, it is possible to perform more accurate analysis than conventional methods.

[1532] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1533] (Processing flow in email application)

[1534] Step 1:

[1535] The server receives the user's email and forwards new emails to the device.

[1536] Input: User's email account information, new emails

[1537] Output: New email notification to device

[1538] Specific operation: Records emails from various mail servers (IMAP / POP3) in the receiving log and forwards them to the mail application.

[1539] Step 2:

[1540] The device will notify the user that new email has arrived.

[1541] Input: New email

[1542] Output: New mail notification to user

[1543] Specific operation: On the smartphone, the message "You have new email" will appear in the notification bar and a notification sound will be heard on the device itself.

[1544] Step 3:

[1545] The user opens their email application and clicks on new email.

[1546] The device extracts the email body and header information.

[1547] Input: New emails opened by the user, including the email body and header information

[1548] Output: Extracted email body and header information

[1549] What happens: The sender address, subject, body, links, and attachment information are separated for analysis.

[1550] Step 4:

[1551] Based on the information extracted by the device, an analysis request is generated for the AI ​​model.

[1552] Input: Extracted email body and header information

[1553] Output: Analysis request to the AI ​​model

[1554] Specific operation: The analysis request is structured in JSON format or other format and prepared for sending to the machine learning model.

[1555] Step 5:

[1556] The device generates an analysis request and sends it to the AI ​​model.

[1557] Input: Analysis request to the AI ​​model

[1558] Output: Sending data to the AI ​​model

[1559] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[1560] Step 6:

[1561] The AI ​​model analyzes the email body and links, and uses natural language processing technology to inspect the content.

[1562] Input: Analysis request email body and link

[1563] Output: Analysis results (fraud likelihood score)

[1564] What it does: Contextual analysis, spam filtering, and matching links to known phishing phrases.

[1565] Step 7:

[1566] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[1567] Input: Parsed email body and links

[1568] Output: Analysis results with fraud probability score

[1569] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[1570] Step 8:

[1571] The device receives the analysis results from the AI ​​model.

[1572] Input: Analysis results with fraud likelihood score

[1573] Output: A warning notice to the user

[1574] Specific Actions: Based on the estimated risk, a warning message is prepared and displayed in the user interface.

[1575] Step 9:

[1576] The device will display a warning to the user.

[1577] Input: warning message

[1578] Output: A warning notice to alert the user

[1579] Specific behavior: If the email is rated as "likely fraudulent," a pop-up message will appear saying "This email may be fraudulent."

[1580] Step 10:

[1581] Users should acknowledge the warning and take action to avoid the risk of fraud.

[1582] Input: warning message

[1583] Output: User response (delete email, avoid clicking links)

[1584] Specific action: The user clicks the "Delete" button to delete the email.

[1585] (Processing flow in a web browsing application)

[1586] Step 1:

[1587] A user clicks on a new URL in their web browsing application.

[1588] Input: User clicks on a URL

[1589] Output: Extracted URLs

[1590] What it does: The browser captures the click event and records the URL being accessed.

[1591] Step 2:

[1592] The device will extract the URL.

[1593] Input: Clicked URL

[1594] Output: URL for analysis

[1595] What it does: Saves the URL as structured data.

[1596] Step 3:

[1597] The device generates an analysis request to send the URL information to the AI ​​model.

[1598] Input: A structured URL to parse

[1599] Output: Analysis request to the AI ​​model

[1600] Specific operation: Structure the analysis request in JSON format or similar.

[1601] Step 4:

[1602] The device generates an analysis request and sends it to the AI ​​model.

[1603] Input: Analysis request to the AI ​​model

[1604] Output: Sending data to the AI ​​model

[1605] Specific operations: Send analysis requests to AI models using HTTP requests or API calls.

[1606] Step 5:

[1607] The AI ​​model analyzes the URL destination (webpage content) and compares it with known phishing sites and malicious domains.

[1608] Input: URL of the analysis request

[1609] Output: Analysis results

[1610] What it does: Scrape the content of web pages and compare it to a database of known phishing sites.

[1611] Step 6:

[1612] The AI ​​model generates an analysis and assigns a score indicating the likelihood of fraud.

[1613] Input: scraped web page content

[1614] Output: Analysis results with fraud probability score

[1615] What it does: Uses a scoring algorithm to quantify the likelihood of fraud.

[1616] Step 7:

[1617] The device receives the analysis results and notifies the user with a warning.

[1618] Input: Analysis results with fraud likelihood score

[1619] Output: A warning notice to the user

[1620] Specific behavior: If there is a high possibility of fraud, a warning message will be displayed in a pop-up saying, "This site may be fraudulent."

[1621] Step 8:

[1622] Users should acknowledge the warning and take action to avoid the risk of fraud.

[1623] Input: warning message

[1624] Output: User response (stops visiting the site)

[1625] Specific Action: The user closes the browser tab or stops visiting the site.

[1626] (Application example 1)

[1627] 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."

[1628] In today's world, as internet-based communications become increasingly common, the number of victims of fraudulent emails and phishing sites is rapidly increasing. These fraudulent activities often result in the leakage of personal information and financial losses, posing a significant risk to many users. Current technology makes it difficult to detect these fraudulent activities in advance and implement countermeasures, creating a demand for systems with advanced analytical capabilities. The purpose of this invention is to solve these problems and enable users to use email and websites safely.

[1629] 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.

[1630] In this invention, the server includes: a terminal receiving an email body or URL; a terminal generating a request for analyzing the email body or URL; a terminal sending the request to an AI model to analyze the possibility of fraud; a terminal issuing a warning to the user based on the analysis results; a smartphone installed on the smartphone for analyzing emails and websites in real time; and a terminal displaying a warning if the possibility of fraud is high based on the analysis results. This enables sophisticated analysis of fraudulent emails and phishing sites and displaying a warning to the user in real time.

[1631] A "terminal" is an electronic device such as a smartphone, tablet, or PC that is operated by a user.

[1632] "Email body" refers to the body of the message written by the sender of the email, which may contain text and links.

[1633] "URL" stands for Uniform Resource Locator and indicates the address of a web page on the Internet.

[1634] An "analysis request" is request data generated to allow an AI model to determine the possibility of fraud based on the content of an email body or URL.

[1635] An "AI model" is an analytical engine that utilizes artificial intelligence technology and is designed to detect fraudulent emails and phishing sites.

[1636] "Natural language processing" is a technique in which AI models understand and analyze human language, and is used to scrutinize the content of emails and web pages.

[1637] A "warning message" is a notification message that is displayed to warn the user of possible fraud based on the analysis results.

[1638] "Real-time" refers to the instantaneous reception and analysis of data, allowing users to obtain results immediately.

[1639] This invention is a system that automatically detects fraudulent emails and websites in email and web browsing applications and issues a warning to the user. This system is mainly composed of a terminal and a server, and the terminal receives emails and URLs, analyzes their contents, and notifies the user.

[1640] Program processing explanation

[1641] The server first receives the email body or URL sent from the device. The email body includes the sender address, subject, body text, links, etc. The server generates a request to analyze this content and sends it to an AI model. This AI model uses natural language processing (NLP) technology and is designed to detect potential fraud and phishing. Specifically, it analyzes the content by comparing it with keywords and phrases commonly found in fraudulent emails, as well as information about the domains to which the links lead.

[1642] As a result of the analysis, the AI ​​model assigns a score indicating the likelihood of fraud and sends it back to the server. The server generates and notifies the user of a warning message based on this score. For example, if the score is high, a warning such as "This email may be fraudulent" or "This site may be fraudulent" will be displayed.

[1643] The system of the present invention mainly uses the following hardware and software. The terminal is a general computing device such as a smartphone or tablet. The server can be a cloud server or a dedicated server in a data center. In addition, the AI ​​model uses a deep learning framework such as TensorFlow or PyTorch.

[1644] Specific examples

[1645] For example, when a user receives an email disguised as a message from a bank, the body of the email often contains keywords such as "transfer" and "password change." When the user opens the email, the device sends the email body to a server, where the AI ​​model analyzes these keywords. If the analysis returns a score of "very likely to be fraudulent," the user is shown a warning message saying, "This email may be fraudulent."

[1646] Prompt Sentence Examples

[1647] 1. Enter the sender, subject, and body of the email you received and assess its likelihood of fraud.

[1648] 2. "Analyze the URL destination to determine whether it is phishing."

[1649] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1650] Step 1:

[1651] The device receives the email or URL.

[1652] What it does: The device receives new emails or website URLs via applications the user has opened. For emails, information such as the sender, subject, body, and links is extracted.

[1653] Input: Received email body or URL

[1654] Output: Extracted email and URL information

[1655] Step 2:

[1656] The device generates an analysis request.

[1657] What it does: Based on the information in the received email or URL, it creates an analysis request to send to the AI ​​model. This request includes the text of the email body and the URL information.

[1658] Input: Extracted email and URL information

[1659] Output: Analysis request (email body text, URL information, etc.)

[1660] Step 3:

[1661] The device sends an analysis request to the server.

[1662] Specific operation: The generated analysis request is sent to the server, which receives it and prepares to pass the request to the AI ​​model.

[1663] Input: Parse request

[1664] Output: The parsing request sent to the server

[1665] Step 4:

[1666] The server sends an analysis request to the AI ​​model and begins the analysis.

[1667] How it works: When the server receives an analysis request, it passes it to the AI ​​model, which uses natural language processing techniques to analyze the content of the email body and URLs to detect potential fraud.

[1668] Input: Parse request

[1669] Output: Analysis results (e.g., fraud probability score)

[1670] Step 5:

[1671] The server receives the analysis results and sends them to the terminal.

[1672] How it works: The server receives the analysis results returned by the AI ​​model and sends them to the device, which includes a score indicating the likelihood of fraud.

[1673] Input: Analysis results (e.g., fraud likelihood score)

[1674] Output: Analysis results sent to the device

[1675] Step 6:

[1676] The device notifies the user of a warning based on the analysis results.

[1677] Specific operation: When the device receives the analysis results, if there is a high possibility of fraud, it will display a warning message to the user. The warning message will include content such as "This email may be fraudulent" or "This site may be fraudulent."

[1678] Input: Analysis result (fraud likelihood score)

[1679] Output: The warning message displayed to the user.

[1680] 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.

[1681] This invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. It also incorporates an emotion engine to understand the user's emotional state and take appropriate action. Below, we will explain in detail the processing of the program for implementing this system.

[1682] System Overview

[1683] This system is embedded in email or web browsing applications, analyzes incoming emails and accessed URLs, and determines whether fraud is likely. It also uses an emotion engine to recognize the user's emotional state and adjust the display method and content of warning messages accordingly.

[1684] Example of an email application

[1685] 1. Receiving emails and notifications

[1686] The server receives new emails and forwards them to the terminal.

[1687] The terminal notifies the user that new mail has arrived.

[1688] 2. Generate email analysis request

[1689] The user clicks to open a new email.

[1690] The terminal extracts the email body and email header.

[1691] The device generates an analysis request to the AI ​​model based on the information extracted.

[1692] 3. Analysis using AI models

[1693] The device sends the analysis request it generates to the AI ​​model.

[1694] The AI ​​model analyzes the email text and links, and uses natural language processing techniques to detect suspicious phrases and links.

[1695] The AI ​​model generates an analysis result and assigns a score indicating the likelihood of fraud.

[1696] 4. Emotion Recognition by Emotion Engine

[1697] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[1698] An emotion engine analyzes this data to determine the user's emotional state.

[1699] 5. Integration of analysis results and emotion recognition results

[1700] The device receives the analysis results from the AI ​​model and the emotion recognition results from the emotion engine.

[1701] Based on these results, the terminal determines the content and display method of a warning message to the user.

[1702] 6. Displaying a warning message

[1703] The device will display a warning message to the user. For example, if the user is nervous, it will display a friendly message such as "This email may be fraudulent. Please proceed with caution."

[1704] If the user is calm, a simple message such as "This email may be fraudulent" will be displayed.

[1705] 7. User Response

[1706] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[1707] Example of a web browsing application

[1708] 1. Accessing the URL

[1709] A user clicks on a new URL in a web browsing application.

[1710] The device extracts the URL.

[1711] 2. URL parsing request generation

[1712] The device generates an analysis request to the AI ​​model based on the URL information.

[1713] 3. Analysis using AI models

[1714] The device sends the analysis request it generates to the AI ​​model.

[1715] An AI model analyzes URL destinations and compares them with known phishing sites and malicious domains.

[1716] 4. Emotion Recognition by Emotion Engine

[1717] The device activates an emotion engine and detects the user's facial expressions and voice using a camera or microphone.

[1718] An emotion engine analyzes this data to determine the user's emotional state.

[1719] 5. Integration of analysis results and emotion recognition results

[1720] The device receives the analysis results and emotion recognition results from the AI ​​model and determines the content and display method of the warning message to the user based on these.

[1721] 6. Displaying a warning message

[1722] The device displays a warning message to the user, for example, if the user is surprised, it displays a descriptive message such as "This site may be fraudulent. Please investigate further."

[1723] If the user is calm, a simple message such as "This site may be a scam" will be displayed.

[1724] 7. User Response

[1725] Users should check the warning message and take appropriate action to avoid the risk of fraud, such as not visiting the site or hiding it.

[1726] Thus, the present invention is a system that recognizes a user's emotional state in email and web browsing applications and adjusts the content and display of warning messages accordingly, allowing users to more effectively avoid the risk of fraud.

[1727] The processing flow will be explained below.

[1728] Step 1:

[1729] The server receives new emails and forwards them to the device, which then recognizes that new emails have been received.

[1730] Step 2:

[1731] The device will notify the user of new mail. A notification will pop up to let you know that new mail has arrived.

[1732] Step 3:

[1733] The user clicks to open a new email, which starts the process of displaying the email contents.

[1734] Step 4:

[1735] The device extracts the email body and email headers (sender, subject, received time, etc.) and extracts the necessary data for analysis.

[1736] Step 5:

[1737] The device generates an analysis request to the AI ​​model based on the extracted email body, links, and sender information. The request contains all the necessary information for analysis.

[1738] Step 6:

[1739] The device sends the generated analysis request to the AI ​​model, which starts the analysis process.

[1740] Step 7:

[1741] The AI ​​model receives the analysis request and analyzes the email body and links, using natural language processing techniques to thoroughly inspect the email content and detect suspicious phrases and links.

[1742] Step 8:

[1743] The AI ​​model compares email content and links with known phishing and spam patterns to assess their likelihood of fraud, and generates a score based on the results.

[1744] Step 9:

[1745] The AI ​​model sends back an analysis and score to the device, which indicates the likelihood of fraud.

[1746] Step 10:

[1747] The device receives the analysis results from the AI ​​model and generates a warning for the user based on the results.

[1748] Step 11:

[1749] The device activates an emotion engine to recognize the user's emotions, and collects the user's facial expressions and voice using the device's camera or microphone.

[1750] Step 12:

[1751] The emotion engine analyzes the collected data to determine the user's current emotional state, categorizing it into emotion categories such as nervous, angry, surprised, or calm.

[1752] Step 13:

[1753] The device combines the analysis results of the AI ​​model with the emotion recognition results of the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it will select a friendly and reassuring message.

[1754] Step 14:

[1755] The device will then display appropriate warning messages to the user, adjusting the tone and details of the messages to match the user's emotional state.

[1756] Step 15:

[1757] Users should review the warning message and take appropriate action to avoid the risk of fraud, such as deleting the email or not clicking on any links.

[1758] In this way, by performing specific processing at each step, a system is realized that not only reliably detects fraudulent emails and warns users, but also uses an emotion engine to take appropriate action taking into account the user's emotional state.

[1759] Example 2

[1760] 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."

[1761] In today's digital communications, damage caused by fraudulent emails and phishing sites is increasing. In particular, these fraudulent activities use sophisticated techniques, making them difficult to detect with conventional security systems. Furthermore, since warning messages are provided uniformly without taking into account the user's emotional state, many users ignore the warnings or fail to take appropriate action, which is a problem. Therefore, there is a need for a system that provides more appropriate warnings by taking into account the user's emotional state.

[1762] 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.

[1763] In this invention, the terminal includes means for receiving an email body or a URL, means for generating a request for analyzing the email body or the URL, means for sending a request to a generative AI model and analyzing the possibility of fraud, means for detecting the user's facial expressions and voice using a camera or microphone and activating an emotion engine that determines the user's emotional state, means for notifying the user of a warning based on the analysis result and the emotional state, means for determining the content and display method of the warning message based on the analysis result and the emotion recognition result, and means for displaying a warning message customized according to the user's emotional state. This makes it possible to effectively notify the user of the risk of fraud and encourage appropriate action while taking the user's emotional state into consideration.

[1764] "Terminal" refers to a computing device used by a user, on which an email application or a web browsing application runs.

[1765] The "mail body" refers to the main content of an email sent or received by a user, and is the part that contains the written message or content.

[1766] "URL" refers to the address that specifies the destination of a link to a web page or resource.

[1767] "Means for generating a request" refers to the function for constructing and generating the request data required to analyze the email body or URL.

[1768] A "generative AI model" refers to an artificial intelligence model that uses machine learning technology to analyze email text and URLs to assess the likelihood of fraud.

[1769] "Means for analyzing" refers to the ability of the generative AI model to assess the likelihood of fraud based on the generated request.

[1770] An "emotion engine" refers to a system that uses a camera or microphone to capture a user's facial expressions and voice, and analyzes that data to determine the user's emotional state.

[1771] "Means for notifying warnings" refers to the ability to display warnings to users based on the likelihood of fraud or their emotional state.

[1772] "Means for determining the content and display method of the warning message" refers to a function that takes into consideration the analysis results and emotion recognition results and determines what kind of warning message to display to the user and how.

[1773] A "customized alert message" refers to an alert message whose content and presentation are tailored to the user's current emotional state.

[1774] The present invention is a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and issues a warning to the user. Furthermore, by combining it with an emotion engine, the system grasps the user's emotional state and takes appropriate action. Detailed embodiments for implementing this system are described below.

[1775] The system is integrated into email or web browsing applications running on a device. The device requires a camera, microphone, display, and internet connection for its hardware. The software includes email client software, a web browser, a generative AI model for natural language processing (NLP), and an emotion engine for analyzing the user's emotional state.

[1776] First, the server receives a new email and forwards it to the device. The device displays a notification of the new email to the user. When the user opens the email, the device extracts the email body and headers. Based on this, the device generates an analysis request for the AI ​​model and sends it to the generative AI model.

[1777] The generative AI model analyzes the email body and links, using natural language processing to detect suspicious phrases and links, and assigns a score indicating the likelihood of fraud. The analysis results are then sent back to the device.

[1778] In parallel, the device activates an emotion engine that uses the camera and microphone to detect the user's facial expressions and voice, and determines the user's emotional state based on the collected data.

[1779] The device integrates the analysis results from the AI ​​model and the emotion recognition results from the emotion engine to determine the content and display method of the warning message to the user. For example, if the user is nervous, it may display "This email may be fraudulent. Please proceed with caution," but if the user is calm, it may simply display "This email may be fraudulent." The device displays an appropriate warning message to the user, who can then confirm the message and take appropriate action to avoid the risk of fraud.

[1780] It works similarly in web browsing applications. When a user clicks on a new URL, the device extracts the URL and generates an analysis request from the AI ​​model. The generative AI model analyzes the URL's destination and compares it with known phishing sites and malicious domains. The emotion engine determines the user's emotional state and, based on the analysis and emotion recognition results, determines the content and display method of the warning message. The device then displays an appropriate warning to the user and prevents them from accessing the dangerous site.

[1781] A specific example of this is the prompt:

[1782] Email applications:

[1783] "This email may be fraudulent. Please review it carefully and verify with other sources before proceeding."

[1784] Web browsing applications:

[1785] "This site may be a scam. Please avoid visiting."

[1786] This system provides warning messages that take into account the user's emotional state, allowing for more effective prevention of fraud risks.

[1787] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1788] Step 1: Receiving emails and notifications

[1789] The server receives new emails. The input is the new email and the output is the received email data. Specifically, the server receives new emails using the SMTP protocol and saves them in the user's mailbox. The server then transfers this email data to the terminal. The terminal displays a pop-up notification of new emails to the user. The input is the email data received from the server and the output is the notification to the user.

[1790] Step 2: Generate an email analysis request

[1791] The user clicks to open a new email. The input is the user's click action, and the output is the email body and email header data. The device extracts the email body and email header. The device obtains this data using the email client's internal API and generates an analysis request for the generative AI model. The input is the extracted email data, and the output is the analysis request.

[1792] Step 3: Analysis by AI model

[1793] The device sends the analysis request it generated to the generative AI model. The input is the analysis request, and the output is the analysis result. Specifically, data is sent using an HTTP POST request. The generative AI model analyzes the email body and links, and uses natural language processing technology to detect suspicious phrases and links. The output is an analysis result with a score for the likelihood of fraud. The analysis result is returned to the device in JSON format.

[1794] Step 4: Emotion Recognition with the Emotion Engine

[1795] The device starts the emotion engine and detects the user's facial expressions and voice using a camera or microphone. The input is real-time data of the user's facial expressions and voice, and the output is the analyzed emotional state. The emotion engine analyzes this data and determines the user's emotional state. The output is the user's emotional state (e.g., nervous, surprised, calm, etc.).

[1796] Step 5: Integrating analysis results with emotion recognition results

[1797] The device receives the analysis results from the generative AI model and the emotion recognition results from the emotion engine. The inputs are the analysis results and emotion recognition results, and the output is the integrated information used for notifications. Specifically, the device runs an algorithm based on the score of the analysis results and the emotion recognition results to determine the content and display method of the warning message to the user.

[1798] Step 6: Displaying warning messages

[1799] The terminal displays a customized warning message to the user. The input is the integrated information, and the output is a warning message to the user. The terminal uses a widget in an email client or web browser to display a pop-up message on the screen. For example, if the user is nervous, it displays "This email may be fraudulent. Please proceed with caution," and if the user is calm, it displays a simple message saying "This email may be fraudulent."

[1800] Step 7: User Action

[1801] The user checks the warning message and takes appropriate action to avoid the risk of fraud. The input is the warning message, and the output is the user's action. Specifically, the user may take action such as deleting the email or not clicking on the link. For example, if the user sees the warning message and selects "Delete fraudulent email," the email will be deleted.

[1802] (Application example 2)

[1803] 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."

[1804] Conventional email and web browsing applications are ineffective at detecting fraudulent emails and websites, and warnings to users are provided in a generic format that is not adapted to the user's emotional state. As a result, users are at increased risk of falling victim to fraud because they are unable to take appropriate action. Therefore, the present invention aims to improve the accuracy of fraud detection and provide warnings that are adapted to the user's emotional state, encouraging the user to take more effective action.

[1805] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1806] In this invention, the server includes, in an email or web browsing application, a means for a terminal to receive an email body or URL, a means for the terminal to generate a request for analyzing the email body or URL, a means for the terminal to send the request to an AI model to analyze the possibility of fraud, a means for the terminal to detect a user's emotional state using a camera and microphone, a means for an emotion engine to analyze the user's emotional state, and a means for the terminal to notify a user of a warning based on the analysis results of the AI ​​model and the emotion analysis results of the emotion engine. This enables highly accurate detection of fraudulent activities and the provision of a warning message adapted to the user's emotional state.

[1807] A "terminal" is an electronic device operated by a user, and is a device for receiving emails and browsing the web.

[1808] The "mail body" is the content of the email, and is information that includes text, images, links, and the like.

[1809] "URL" is a Uniform Resource Locator that indicates the address of a web page, and is an identifier for specifying the destination of browsing access.

[1810] An "AI model" is an analytical system that uses artificial intelligence algorithms and is a model trained to perform a specific task.

[1811] An "analysis request" is a command sent from a device to an AI model that contains the information necessary to analyze specific data.

[1812] "Fraudulence" is a probability or score that indicates whether an email or website is attempting to commit fraud.

[1813] The "emotional state" indicates the psychological state of the user, and is an emotion judged from facial expressions, voice, etc.

[1814] An "emotion engine" is software that analyzes a user's emotional state from their facial expressions and voice.

[1815] A "warning message" is a notification message that alerts users to risks and caution, and is in the form of text or audio to notify them of possible fraudulent activity.

[1816] A "camera" is a device that captures images and videos and is used to detect a user's facial expression.

[1817] A "microphone" is a device that records sound and is used to detect the user's voice.

[1818] This invention realizes a system that automatically detects fraudulent emails and websites in email applications and web browsing applications, and then analyzes the user's emotional state and responds appropriately. The specific hardware and software that make up the system, as well as the actual processing procedures, are described below.

[1819] System Configuration

[1820] The system consists of the following major hardware and software components:

[1821] Device: An electronic device operated by a user, such as a smartphone, tablet, or computer.

[1822] Camera: A built-in camera used to detect the user's facial expressions.

[1823] Microphone: A microphone built into the device, used to detect the user's voice.

[1824] AI model: An artificial intelligence model that analyzes the content of emails and URLs to determine the likelihood of fraud.

[1825] Emotion engine: Software for analyzing the user's emotional state from their facial expressions and voice.

[1826] Natural language processing technology: Used to analyze email text and URLs.

[1827] System operation explanation

[1828] The device receives an email or URL: When a user receives a new email or clicks on a new website URL, the device recognizes this.

[1829] Generating an analysis request: The device generates an analysis request to the AI ​​model based on the received email body and URL. This request includes detailed information about the email content and URL.

[1830] Performing fraud analysis: The device sends an analysis request to the AI ​​model, which uses natural language processing techniques to analyze the content of emails and URLs and calculates a score indicating the likelihood of fraud.

[1831] Emotional state detection: The device activates the camera and microphone to collect the user's facial expressions and voice in real time, and the emotion engine then analyzes this data to determine the user's emotional state.

[1832] Integrated display of warning messages: The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to adjust the content and display method of the warning message displayed to the user. For example, if the user is surprised, a more detailed warning message will be displayed, and if the user is calm, a simple message will be displayed.

[1833] Examples and prompts

[1834] For example, if a user receives an email that reads, "Congratulations! You have won a $1000 gift card. Click here to claim.", the device will generate the following prompt:

[1835] Given the following email content and header, detect if it's a phishing email and provide a confidence score. Email content: "Congratulations! You have won a $1000 gift card. Click here to claim." Email header: "From: no-reply@fakesite.com, Subject: Claim your prize"

[1836] This prompt is sent to an AI model that analyzes the email for its potential fraudulent nature. Based on the analysis results and the user's emotional state, a warning message such as "This email may be fraudulent. Please investigate further" is displayed.

[1837] In this way, the system can achieve highly accurate detection of fraudulent activity and present warning messages that adapt to the user's emotional state.

[1838] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1839] Step 1:

[1840] The device receives a new email or URL. By receiving an email or clicking a URL, the device obtains this information. The input is the new email or the accessed URL, and the output is the email body or URL to be analyzed.

[1841] Step 2:

[1842] The device generates a request to analyze the email body and URL. Specifically, it generates a prompt message containing detailed information about the email content and URL. The input is the email body and URL, and the output is an analysis request to send to the AI ​​model.

[1843] Step 3:

[1844] The analysis request generated by the device is sent to an AI model to analyze the likelihood of fraud. The AI ​​model uses natural language processing techniques to analyze the content of emails and URLs and generate a score indicating fraud. The input is the analysis request, and the output is a fraud likelihood score.

[1845] Step 4:

[1846] The device detects the user's emotional state using a camera and microphone. Specifically, the camera captures the user's facial expressions and the microphone records their voice. The emotion engine analyzes this data and determines the user's emotions. The input is camera images and voice data, and the output is the user's emotional state.

[1847] Step 5:

[1848] The device combines the analysis results of the AI ​​model with the emotion analysis results of the emotion engine to determine the most appropriate content and display method of the warning message. The input is the fraud probability score and emotional state, and the output is an optimized warning message.

[1849] Step 6:

[1850] The terminal displays a warning message to the user. The content of the message is a warning about the risk of fraud, and appropriate wording is used depending on the user's emotional state. The input is the optimized warning message, and the output is the warning information that the user confirms.

[1851] Step 7:

[1852] The user checks the warning message and takes appropriate action to avoid the risk of fraud. For example, the user may delete the email, not click on the link, or restrict access to the site. The input is the warning message, and the output is the user's appropriate response.

[1853] 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.

[1854] 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.

[1855] 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.

[1856] 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.

[1857] 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.

[1858] 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.

[1859] 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).

[1860] 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.

[1861] 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."

[1862] 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.

[1863] 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).

[1864] 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.

[1865] 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.

[1866] 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.

[1867] 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.

[1868] 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.

[1869] 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.

[1870] 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.

[1871] 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.

[1872] 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.

[1873] 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.

[1874] The following is further disclosed regarding the above embodiment.

[1875] (Claim 1)

[1876] In an email or web browsing application, a terminal receives an email body or a URL;

[1877] A terminal generates a request for analyzing the email body or URL;

[1878] means for transmitting the request to an AI model for analyzing the likelihood of fraud;

[1879] The system further includes a means for notifying a user of a warning based on the analysis result.

[1880] (Claim 2)

[1881] 2. The system of claim 1, wherein the AI ​​model uses natural language processing techniques to analyze the content of the email body and detect suspicious phrases, links, and sender information.

[1882] (Claim 3)

[1883] 10. The system of claim 1, wherein the AI ​​model generates a fraud likelihood score and determines the content of a warning message to be displayed to the user based on the score.

[1884] "Example 1"

[1885] (Claim 1)

[1886] means for the terminal to receive electronic communications or data relating to the website;

[1887] means for generating a request for the terminal to analyze said data;

[1888] means for transmitting the request to a machine learning model to analyze the likelihood of fraud;

[1889] means for notifying a user of a warning based on the analysis result;

[1890] The system includes a means for guiding a user's operation based on the warning.

[1891] (Claim 2)

[1892] 10. The system of claim 1, wherein the machine learning model uses natural language processing techniques to analyze the content of electronic communications to detect suspicious phrases, links, and sender information.

[1893] (Claim 3)

[1894] 10. The system of claim 1, wherein the machine learning model generates a fraud likelihood score and determines the content of a warning message to be displayed to the user based on the score.

[1895] "Application Example 1"

[1896] (Claim 1)

[1897] In an email or web browsing application, a terminal receives an email body or a URL;

[1898] A terminal generates a request for analyzing the email body or URL;

[1899] means for transmitting the request to an AI model for analyzing the likelihood of fraud;

[1900] means for notifying a user of a warning based on the analysis result;

[1901] A means of analyzing emails and websites in real time that is installed on smartphones,

[1902] A method to display warnings when there is a high possibility of fraud based on the analysis results

[1903] A system including:

[1904] (Claim 2)

[1905] 2. The system of claim 1, wherein the AI ​​model uses natural language processing techniques to analyze the content of the email body and detect suspicious phrases, links, and sender information.

[1906] (Claim 3)

[1907] 10. The system of claim 1, wherein the AI ​​model generates a fraud likelihood score and determines the content of a warning message to be displayed to the user based on the score.

[1908] "Example 2: Combining Emotion Engines"

[1909] (Claim 1)

[1910] A means for the device to receive the email body or URL;

[1911] A terminal generates a request for analyzing the email body or URL;

[1912] means for transmitting the request to a generative AI model for analyzing the likelihood of fraud;

[1913] A device detects the user's facial expression and voice using a camera or microphone, and activates an emotion engine that determines the user's emotional state;

[1914] a means for the terminal to notify a user of a warning based on the analysis result and the emotional state;

[1915] a means for determining the content and display method of a warning message based on the analysis result and the emotion recognition result;

[1916] means for displaying a customized warning message to the user according to the user's emotional state;

[1917] A system including:

[1918] (Claim 2)

[1919] 2. The system of claim 1, wherein the generative AI model uses natural language processing techniques to analyze the content of email text or URLs to detect suspicious phrases, links, and sender information.

[1920] (Claim 3)

[1921] 10. The system of claim 1, wherein the generative AI model generates a fraud likelihood score and determines the content of a warning message to be displayed to the user based on the score and the user's emotional state.

[1922] "Application example 2 when combining emotion engines"

[1923] (Claim 1)

[1924] In an email or web browsing application, a terminal receives an email body or a URL;

[1925] A terminal generates a request for analyzing the email body or URL;

[1926] means for transmitting the request to an AI model for analyzing the likelihood of fraud;

[1927] means for detecting the emotional state of a user in the terminal using a camera and a microphone;

[1928] a means for the emotion engine to analyze the user's emotional state;

[1929] The system includes a means for the terminal to notify the user of a warning based on the analysis results of the AI ​​model and the emotion analysis results of the emotion engine.

[1930] (Claim 2)

[1931] 2. The system of claim 1, wherein the AI ​​model uses natural language processing techniques to analyze the content of the email body and detect suspicious phrases, links, and sender information.

[1932] (Claim 3)

[1933] 2. The system of claim 1, wherein the AI ​​model generates a fraud likelihood score and determines the content and display method of a warning message to be displayed to the user based on the score and the emotion analysis results of the emotion engine. [Explanation of symbols]

[1934] 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. In an email or web browsing application, a terminal receives an email body or a URL; A terminal generates a request for analyzing the email body or URL; means for transmitting the request to an AI model for analyzing the likelihood of fraud; The system further includes a means for notifying a user of a warning based on the analysis result.

2. The system of claim 1 , wherein the AI ​​model uses natural language processing techniques to analyze the content of the email body and detect suspicious phrases, links, and sender information.

3. 10. The system of claim 1, wherein the AI ​​model generates a fraud likelihood score and determines the content of a warning message to be displayed to the user based on the score.

Images