System

Abstract

A system is provided.SOLUTION: A system comprising: means for recording a conversation; means for converting the recorded conversation into text data; means for analyzing the text data to detect specific keywords and context; means for evaluating a likelihood of fraud based on the detected keywords and context; means for generating an alert message and transmitting the alert message to a user terminal if the likelihood of fraud is high; and means for displaying the alert message on the user terminal.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In recent years, the methods of special fraud have become more sophisticated, and many people have fallen victim to them. In particular, frauds committed over the phone or in face-to-face conversations make it difficult for victims to detect the risk of fraud, which contributes to the expansion of damage. In addition, it is difficult to detect fraud early, so there is a need for a method to prevent damage before it occurs. In order to solve these problems, the present invention aims to provide a system that automatically analyzes the content of conversations and warns of possible fraud in real time. [Means for solving the problem]

[0005] The present invention provides a system that includes a means for recording conversation content, a means for converting the recorded conversation content from voice data to text data, a means for analyzing the text data and detecting specific keywords and context, a means for evaluating the possibility of fraud based on the detected keywords and context, a means for generating a warning message and transmitting it to a user terminal when a high possibility of fraud is detected, and a means for displaying the warning message on the user terminal. The system also includes a means for securely transmitting the voice data to a server and a means for allowing the user to select a corresponding action based on the notified warning message. This provides a system that can detect fraud risks in real time during a conversation and quickly issue a warning to the user, thereby preventing fraud damage before it occurs.

[0006] A "means for recording conversation content" is a device or function that digitally records a user's telephone conversations or face-to-face conversations.

[0007] "Audio data" refers to data that represents recorded conversations in digital form.

[0008] "Text data" is data expressed as a string of characters converted from voice data using voice recognition technology.

[0009] "Specific keywords" are words or phrases that suggest fraudulent activity, such as "transfer" or "account."

[0010] "Contextual detection methods" are algorithms or techniques that understand the circumstances or conditions under which certain keywords appear and assess the likelihood that those keywords are part of a fraud.

[0011] A "means for assessing the likelihood of fraud" is a technology for assessing the likelihood of fraud as a number or score based on detected keywords and context.

[0012] A "warning message" is a message containing a warning that is sent to the user when it is determined that there is a high possibility of fraud.

[0013] A "user terminal" is an electronic device used by a user, such as a mobile phone or computer, that is equipped with a display device for displaying a warning message.

[0014] The "means for displaying a warning message" refers to software or hardware for displaying a warning message on the screen of a user terminal.

[0015] The "means for transmitting to a server" refers to a communication technology for transmitting the recorded audio data to a remote server. [Brief explanation of the drawings]

[0016] [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

[0017] 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.

[0018] First, the terms used in the following description will be explained.

[0019] 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).

[0020] 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.

[0021] 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.

[0022] 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.

[0023] 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."

[0024] [First embodiment]

[0025] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0026] 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.

[0027] 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).

[0028] 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.

[0029] 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.

[0030] 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.

[0031] 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.

[0032] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] 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."

[0037] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[0038] System configuration overview

[0039] User terminal

[0040] Voice Recorder: Has the ability to record the user's conversations.

[0041] Audio file sending function: Sends recorded audio data to the server.

[0042] Warning display function: Receives warning messages from the server and displays them on the screen.

[0043] server

[0044] Speech recognition engine: Converts voice data into text data.

[0045] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[0046] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[0047] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal.

[0048] Program processing

[0049] Audio Acquisition

[0050] (Device) records conversations. When a user is having a phone or face-to-face conversation, the audio of that conversation is stored digitally.

[0051] Speech-to-text conversion

[0052] The device sends the recorded audio file to the server, where a speech recognition engine is used to convert the audio data into text data.

[0053] Text analytics

[0054] The server analyzes the converted text data using a natural language processing engine to detect specific keywords (e.g., "money," "transfer," "account information," "urgent") and context.

[0055] Determining the likelihood of fraud

[0056] The server determines the possibility of fraud based on the analysis results. If certain conditions are met, a risk score is calculated and the fraud risk level is calculated.

[0057] Warning

[0058] If the server determines that the risk score is high, it generates a warning message and sends it to the user's device, which then displays the warning on its screen to alert the user.

[0059] Specific examples

[0060] Consider the example where the user is on a call.

[0061] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[0062] Processing flow

[0063] 1. (Device) records the call audio and sends it to the server.

[0064] 2. (Server) converts the audio file into text data.

[0065] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[0066] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[0067] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[0068] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[0069] In this way, the present invention analyzes conversation content in real time and quickly detects and warns of fraud risks, helping users prevent themselves from becoming victims of fraud.

[0070] The processing flow will be explained below.

[0071] Step 1:

[0072] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[0073] Step 2:

[0074] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[0075] Step 3:

[0076] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The speech recognition engine uses advanced speech recognition technology such as Google® Speech-to-Text API or IBM Watson®.

[0077] Step 4:

[0078] The server inputs the converted text data into a natural language processing engine, which analyzes the text data and detects specific keywords (e.g., "transfer," "account," "urgent"). Natural language processing technologies used include SpaCy and NLTK.

[0079] Step 5:

[0080] The server analyzes the context based on the results of keyword detection. This contextual analysis allows us to understand the context in which the keywords are used and collects data to predict the likelihood of fraud.

[0081] Step 6:

[0082] The server evaluates the likelihood of fraud based on the collected data. The evaluation is performed using statistical models and machine learning algorithms to calculate a risk score. For example, if certain conditions are met, a risk score is calculated on a scale of 0 to 100.

[0083] Step 7:

[0084] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action.

[0085] Step 8:

[0086] The server generates a warning message and sends it to the user's device. This communication also uses an encrypted protocol to ensure security.

[0087] Step 9:

[0088] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[0089] Step 10:

[0090] The user checks the warning message and takes appropriate action. If the possibility of fraud is judged to be high, the user can choose to take action such as ending the call to prevent fraud from occurring. If necessary, the system also provides a function to automatically end the call based on the user's selection.

[0091] Example 1

[0092] 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."

[0093] In recent years, there has been a demand for technology to prevent special frauds, but current methods have difficulty detecting potential fraud in real time and issuing warnings to users. In particular, as fraud methods become more sophisticated and diverse, even a momentary error in judgment can result in significant damage. This leaves users constantly exposed to the risk of fraud and makes it difficult to have conversations with peace of mind. Therefore, a system is needed that can detect potential fraud in real time and issue immediate warnings.

[0094] 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.

[0095] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to the user terminal if the possibility of fraud is high, means for displaying the warning message on the user terminal, means for transmitting the voice data to the server via a network, and means for calculating a risk score in the analysis and determining the fraud risk level, thereby enabling rapid detection and warning of fraud risks in real time.

[0096] "Means for recording conversations" refers to devices or functions that digitally record a user's conversations.

[0097] "Means for converting the recorded conversation content from voice data to text data" refers to a function that uses voice recognition technology to analyze recorded voice data and convert it into text data as character string information.

[0098] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses natural language processing technology to extract and analyze important keywords and related contexts contained in text data.

[0099] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or rule sets for assessing the risk of specialized fraud based on analyzed keywords and contextual information.

[0100] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function for automatically creating a warning message and sending it to a user terminal when it is determined that there is a high risk of fraud.

[0101] "Means for displaying a warning message on a user terminal" refers to a function for visually displaying a received warning message on a display or the like of a user terminal.

[0102] "Means for transmitting audio data to a server via a network" refers to the function of transmitting recorded audio data to a server using the Internet or other networks.

[0103] "Means for calculating a risk score in the analysis and determining the fraud risk level" refers to an algorithm or method for deriving a risk score from the results of text analysis and determining the fraud risk level based on that score.

[0104] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[0105] System configuration overview

[0106] User terminal

[0107] Audio Recorder: Has the ability to record user conversations. The device's audio recorder records conversations in real time in digital format.

[0108] Audio file transmission function: This function transmits recorded audio data to a server via the network. When recording is finished, the audio data is automatically sent to the server.

[0109] Warning display function: Receives warning messages from the server and displays them on the screen. Received warning messages are displayed on the display to alert the user.

[0110] server

[0111] Speech recognition engine: Converts voice data into text data. Examples of use include Google Speech-to-Text and IBM Watson Speech to Text.

[0112] Natural language processing engine: Analyzes text data and detects specific keywords and contexts. Examples of use include spaCy and BERT.

[0113] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data. Calculates a risk score based on specific keywords and context to determine the risk level.

[0114] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal. A warning message is automatically generated and sent to the user's terminal.

[0115] Specific examples

[0116] Consider the example where the user is on a call.

[0117] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[0118] Processing flow

[0119] 1. (Device) records the call audio and sends it to the server.

[0120] When a user has a conversation over the phone or in person, the device's audio recorder stores the audio in digital form.

[0121] 2. (Server) converts the audio file into text data.

[0122] The server's speech recognition engine (e.g., Google Speech-to-Text) receives the audio file as input and converts it into text information.

[0123] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[0124] The server's natural language processing engine (e.g., spaCy or BERT) performs the analysis and detects specific keywords and context.

[0125] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[0126] A risk score is calculated based on the keywords and context detected, with a score of 80 or above being considered high risk.

[0127] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[0128] If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user terminal.

[0129] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[0130] As soon as the user terminal receives the warning message, it immediately displays it on the screen to alert the user.

[0131] Example prompts for generative AI models

[0132] "Analyze this conversation and assess its fraud risk: 'Please transfer $5,000 to this account immediately. Hurry or you'll be in trouble.'"

[0133] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0134] Step 1:

[0135] (Device) records user conversations.

[0136] Input: User's spoken speech

[0137] Specific operation: The device's audio recording device records the conversation in real time in digital format.

[0138] Output: Recorded audio data (digital format)

[0139] Step 2:

[0140] (Device) sends the recorded audio data to the server.

[0141] Input: Recorded audio data

[0142] Specific operation: When recording is finished, the audio data is automatically sent to the server via the network. Communication to the server is performed using a secure protocol (e.g., HTTPS).

[0143] Output: Audio data sent to the server

[0144] Step 3:

[0145] The server starts a speech recognition engine and converts the speech data into text data.

[0146] Input: Audio data sent to the server

[0147] Specific operation: The server's speech recognition engine (for example, Google Speech-to-Text) analyzes the audio file and converts it into text data as string information.

[0148] Output: Converted text data

[0149] Step 4:

[0150] (Server) analyzes the converted text data using a natural language processing engine.

[0151] Input: Converted text data

[0152] How it works: The server's natural language processing engine (such as spaCy or BERT) performs detailed analysis of the text data to detect specific keywords and context. It extracts important keywords such as "money," "transfer," "account information," and "hurry," and analyzes the context.

[0153] Output: Analysis results (detected keywords and context information)

[0154] Step 5:

[0155] (Server) evaluates the possibility of fraud based on the analysis results.

[0156] Input: Analysis results (detected keywords and context information)

[0157] Specific behavior: Calculate a risk score based on the detected keywords and context. For example, assign a weight to each keyword and evaluate the match with the context to score the overall risk. If the risk score is high, it is determined to be high risk.

[0158] Output: Risk score, risk assessment result

[0159] Step 6:

[0160] The server generates a warning message and sends it to the user terminal.

[0161] Input: Risk assessment result (if high risk)

[0162] Specific operation: If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user's device stating, "This call may be a special fraud. Please be careful."

[0163] Output: Generated warning message

[0164] Step 7:

[0165] (Device) receives the warning message and displays it on the screen.

[0166] Input: The generated warning message

[0167] Specific operation: As soon as the user device receives a warning message, it immediately displays it on the screen to alert the user. In addition to the visual display, additional notification methods such as audio alerts can also be used.

[0168] Output: Display warning message, notify user

[0169] (Application example 1)

[0170] 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."

[0171] In modern society, the methods used by special fraudsters are becoming increasingly sophisticated, and many people are falling victim to fraud. Elderly people and users unfamiliar with the Internet are particularly susceptible to fraud. As a result, not only financial damage but also psychological damage is severe. The present invention aims to prevent such special frauds from occurring. Specifically, the objective of this invention is to provide a system that prevents fraud damage by detecting possible fraud in real time and issuing a warning to the user.

[0172] 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.

[0173] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from audio data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud, means for displaying the warning message on the user terminal, means for the user to select a response action when the warning message is displayed, and means for analyzing the text data using a generative AI model. This allows for highly accurate detection of the possibility of fraud in real time, enabling the user to take immediate response action, thereby preventing fraud damage.

[0174] A "conversation recording device" is a device or software that has the ability to record a user's telephone or face-to-face conversations in real time and store them in digital form.

[0175] "Means for converting recorded conversation content from voice data to text data" refers to a system that uses voice recognition technology to convert recorded voice files into text format.

[0176] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses a natural language processing engine to analyze important words and phrases contained in text data and their contexts.

[0177] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or functions that calculate a fraud risk score and determine the likelihood of fraud based on the analyzed text data.

[0178] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function that generates a warning message and sends it to a user terminal via a network when it is determined that there is a high risk of fraud.

[0179] The "means for displaying a warning message on a user terminal" refers to a display or application for visually displaying to the user the warning message received on the user terminal.

[0180] "Means for the user to select a response action when a warning message is displayed" refers to an interface or function that allows the user to check the warning message and then select an appropriate response action themselves.

[0181] "Means for securely transmitting voice data to a server" refers to a function that uses encryption technology and communication protocols to maintain data integrity and confidentiality when transmitting voice data from a user terminal to a server.

[0182] "Means for analyzing text data using a generative AI model" refers to a function that uses a natural language processing engine that incorporates AI technology to analyze text data with high accuracy.

[0183] As an embodiment of the present invention, a system for detecting and warning against special frauds will be described in detail. This system mainly operates using a user terminal, a server, and a communication network. Specific embodiments of this technology are described below.

[0184] User terminal

[0185] The user terminal implements the following main functions:

[0186] 1. Voice recording function

[0187] The user terminal is equipped with a microphone for recording voice, which automatically records the voice during conversation and stores it in digital format.

[0188] 2. Audio file sending function

[0189] The recorded audio data is securely transmitted to the server using encryption technology, which is essential to maintain data integrity and confidentiality.

[0190] 3. Warning display function

[0191] Displays warning messages received from the server to the user. Warning messages are displayed in a popup or notification bar so that the user can see them immediately.

[0192] server

[0193] The server performs the primary processing for analyzing the voice data and assessing the likelihood of fraud.

[0194] 1. Speech Recognition Engine

[0195] The server receives the voice data sent from the user's device and converts it into text data using a voice recognition engine, which uses the Google speech recognition API.

[0196] 2. Text Analysis Engine

[0197] A natural language processing engine using generative AI models analyzes text data, detecting specific keywords (e.g., "money," "transfer," "account") and context to assess the likelihood of fraud.

[0198] 3. Fraud Detection Algorithm

[0199] A fraud risk score is calculated based on the detected keywords and context. If the risk score exceeds a certain threshold, it is determined to be highly likely to be fraud. For example, if a conversation such as "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble" is detected, it is determined to be high risk.

[0200] 4. Warning generation function

[0201] If the risk is determined to be high, a warning message is generated and sent to the user's terminal. The warning message may include a statement such as, "This call may be a special fraud. Please be careful."

[0202] Specific examples

[0203] A specific scenario is shown below.

[0204] Example Scenario

[0205] While on the phone, the user receives the message, "Please transfer 500,000 yen to this account immediately."

[0206] The user terminal records this conversation and sends the audio file to the server.

[0207] The server converts the audio file into text data and detects specific keywords such as "account," "transfer," and "urgent."

[0208] Based on this text data, a generative AI model calculates a fraud risk score and determines it as high risk.

[0209] The server generates a warning message and sends it to the user's terminal stating, "This call may be a special fraud. Please be careful."

[0210] The user terminal displays a warning message and the user takes appropriate action.

[0211] Prompt Sentence Examples

[0212] Below are some example prompts for the generative AI model:

[0213] "Please rate the likelihood of fraud for the following text:

[0214] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[0215] With such detailed embodiments, the present invention is able to detect fraud risks in real time and issue immediate warnings to users.

[0216] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0217] Step 1: Recording audio data

[0218] The user terminal records telephone or face-to-face conversations in real time, and the recorded data is stored in digital format. The input is the user's conversational voice, and the output is the recorded voice data.

[0219] Step 2: Send your recordings

[0220] The device transmits the recorded voice data to the server. The transmission is performed using encryption technology to maintain data integrity and confidentiality. The input is the recorded voice data, and the output is the transmission of encrypted voice data.

[0221] Step 3: Convert audio data to text

[0222] The server converts the received voice data into text data using a voice recognition engine (e.g., Google voice recognition API). The input is encrypted voice data, and the output is text data.

[0223] Step 4: Analyzing the text data

[0224] The server uses a generative AI model to analyze the text data and detect specific keywords and context. Specifically, a natural language processing engine scans the text data and searches for keywords such as "money," "transfer," and "account." The input is the text data, and the output is the analyzed keywords and context information.

[0225] Step 5: Fraud risk assessment

[0226] The server calculates a fraud risk score based on the analyzed keywords and context using a pre-trained fraud detection algorithm, where the input is keywords and context information and the output is a risk score.

[0227] Step 6: Generate a warning message

[0228] Once the risk score is determined, the server automatically generates a warning message if it determines that there is a high possibility of fraud. The warning message includes content such as "This call may be a special fraud. Please be careful." The input is the risk score, and the output is the warning message.

[0229] Step 7: Sending a warning message

[0230] The server sends the generated warning message to the user terminal. The input is the warning message and the output is the transmission to the user terminal.

[0231] Step 8: Displaying warning messages

[0232] The terminal displays the received warning message to the user. A pop-up notification or an alert message is used for display so that the user can immediately check it. The input is the warning message, and the output is the warning displayed on the user's screen.

[0233] Step 9: User Actions

[0234] The user checks the warning message displayed on the terminal and selects an appropriate response action. Based on the warning, the user may end the call or perform additional confirmation. The input is the warning message, and the output is the user's response action.

[0235] 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.

[0236] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[0237] System configuration overview

[0238] User terminal

[0239] Voice Recorder: Provides the ability to record a user's conversations and store them in digital form.

[0240] Audio file sending function: Sends recorded audio data to the server.

[0241] Emotion recognition function: Analyzes the user's conversational voice in real time and recognizes emotions.

[0242] Warning display function: Receives warning messages from the server and displays them on the screen.

[0243] server

[0244] Speech recognition engine: Converts voice data into text data.

[0245] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[0246] Emotion engine: Receives user emotion data and evaluates it in combination with analytical data.

[0247] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[0248] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[0249] Program processing

[0250] Audio Acquisition

[0251] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[0252] Speech-to-text conversion

[0253] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[0254] emotion recognition

[0255] The device analyzes the user's speech in real time and recognizes emotions. Emotion recognition technology uses algorithms based on characteristics such as tone, pitch, and speed of speech.

[0256] Text analytics

[0257] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The natural language processing engine then analyzes the text data to detect specific keywords and context.

[0258] Emotional Data Integration

[0259] The server integrates emotional data into the results of text analysis. Emotional data reflects the user's psychological state, and combining this information improves the accuracy of fraud likelihood assessment.

[0260] Determining the likelihood of fraud

[0261] The server evaluates the likelihood of fraud based on the integrated data. Statistical models and machine learning algorithms are used to calculate a risk score. The inclusion of emotional data also enhances the accuracy of the risk assessment.

[0262] Warning

[0263] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action. The warning message also includes additional information tailored to the user's psychological state based on emotional data.

[0264] Displaying warnings

[0265] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[0266] Specific examples

[0267] Consider the example where the user is on a call.

[0268] Call details: "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into problems."

[0269] Processing flow

[0270] 1. (Device) records the call audio and sends it to the server.

[0271] 2. (Server) converts the audio file into text data, and a natural language processing engine detects keywords such as "account," "transfer," and "urgent" and analyzes the context.

[0272] 3. (Device) analyzes the user's emotions in real time and generates emotion data. In this case, it detects that the user is feeling tense.

[0273] 4. (Server) integrates the emotion data and evaluates the likelihood of fraud. The risk score is evaluated as 90, which is determined to be high risk.

[0274] 5. (Server) generates a warning message saying, "This call may be a fraudulent transaction. Please be careful." and sends it to the user. Based on the emotional data, it also includes additional information such as, "You seem nervous, but please remain calm."

[0275] 6. (Device) displays a warning message, and the user can respond appropriately. If necessary, the call can be terminated, preventing fraud before it occurs.

[0276] The present invention simultaneously analyzes the conversation content and user emotions to achieve more accurate fraud detection and user assistance. This system allows users to quickly recognize fraud risks and respond appropriately.

[0277] The processing flow will be explained below.

[0278] Step 1:

[0279] The device records the user's conversation. The recording is initiated by the user's operation or a voice trigger, and is saved in a digital format, such as WAV or MP3.

[0280] Step 2:

[0281] The (device) sends the recorded audio file to the server using an encrypted communication protocol (e.g., HTTPS).

[0282] Step 3:

[0283] While recording, the device analyzes the user's speech in real time and uses an emotion recognition engine to detect the user's emotional state, using algorithms based on voice tone, pitch, speed, etc.

[0284] Step 4:

[0285] The server inputs the received audio file into a speech recognition engine, which converts the audio data into text data. The speech recognition engine used is Google Speech-to-Text API or IBM Watson.

[0286] Step 5:

[0287] The server inputs the text data into a natural language processing engine, which analyzes specific keywords (e.g., "transfer," "account," "urgent") and context. Natural language processing is performed using tools such as SpaCy and NLTK.

[0288] Step 6:

[0289] The server detects the context based on the results of text analysis and assesses the likelihood of fraud based on this, using statistical models and machine learning algorithms.

[0290] Step 7:

[0291] The device recognizes the emotion data in real time and sends it to the server. This data may include the user's emotions such as tension, impatience, and fear.

[0292] Step 8:

[0293] The server integrates the results of text analysis with emotion data and reassesssses the likelihood of fraud. Including emotion data improves the accuracy of the assessment.

[0294] Step 9:

[0295] The server calculates a risk score based on the integrated data, and if it is determined to be high risk, it generates a warning message informing the user that it may be a scam and urging them to act calmly.

[0296] Step 10:

[0297] The server generates a warning message and sends it to the user's device. This message is sent using a secure communication protocol.

[0298] Step 11:

[0299] (Device) displays received warning messages to the user. The messages are displayed in a pop-up format and have an attention-grabbing design.

[0300] Step 12:

[0301] The user checks the warning message and takes appropriate action. For example, if it is determined that there is a high possibility of fraud, the user can take action such as terminating the call. The message also includes additional information (e.g., "Please remain calm") tailored to the user's psychological state based on emotional data.

[0302] In this way, the present invention provides a system that analyzes conversation content and emotions in real time, assesses the possibility of fraud with high accuracy, and quickly issues a warning to the user, thereby preventing fraud damage before it occurs.

[0303] Example 2

[0304] 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."

[0305] Special fraud remains a serious social problem, with many people falling victim to it. Conventional fraud detection systems make judgments based solely on the analysis of text data, making it difficult to accurately assess the likelihood of fraud. Furthermore, warnings are issued without taking into account the user's emotions, which can lead to users being unable to respond appropriately. Therefore, there is a need for a new system that can more accurately detect potential fraud and provide warning messages that take the user's emotions into account.

[0306] 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.

[0307] In this invention, the server includes a means for analyzing the user's conversation voice in real time and recognizing emotions, a means for evaluating the possibility of fraud based on detected keywords and context and the recognized emotion data, and a means for generating a warning message and sending it to the user terminal if the possibility of fraud is high. This allows for a more accurate warning by combining text data and emotion data to evaluate the possibility of fraud.

[0308] "Conversation recording means" refers to a method or device for digitally recording and storing a user's voice.

[0309] "Means for converting voice data into text data" refers to technology or devices that analyze recorded voice data and convert it into corresponding text data.

[0310] "Means for analyzing text data and detecting specific keywords and context" refers to technology that analyzes text data converted from audio data and identifies specific words and phrases and the context behind them.

[0311] "Means for analyzing a user's conversational voice in real time and recognizing emotions" refers to a method or device that processes a user's voice in real time and identifies emotions from characteristics of the voice such as tone, pitch, and speed.

[0312] "Means for assessing the likelihood of fraud based on detected keywords and context, and recognized sentiment data" refers to algorithms and technologies that combine analysis results with sentiment data to determine the likelihood of fraud.

[0313] "Means for generating and sending a warning message to a user terminal" refers to technology or equipment that, when a possibility of fraud is determined, creates an appropriate warning message and sends that message to a user terminal.

[0314] "Means for displaying a warning message on a user terminal" refers to a method or device for visually displaying a received warning message on a user terminal.

[0315] "Means for safely transmitting voice data and emotion data to a server" refers to technology for transmitting voice data and emotion data to a server after implementing safety measures such as encryption.

[0316] "Means for the user to select a response action based on the notified warning message and additional information" refers to technology or devices that allow the user to select the most appropriate action based on the received warning message and the provided additional information.

[0317] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[0318] System configuration overview

[0319] User terminal

[0320] Audio Recording Device: A device that has the ability to record a user's conversation and store it digitally. Examples include smartphones and dedicated recording devices.

[0321] Audio file transmission function: Recorded audio data is sent to the server using an encrypted communication protocol (e.g. HTTPS).

[0322] Emotion Recognition: Analyzes the user's speech in real time to recognize emotions. Emotion recognition technology uses algorithms based on voice characteristics such as tone, pitch, and speed.

[0323] Warning display function: Receives warning messages from the server and displays them on the screen. For example, a warning is displayed in pop-up format on a smartphone screen.

[0324] server

[0325] Speech recognition engine: Converts voice data into text data. Specifically, Google Cloud Speech-to-Text API is used.

[0326] Natural language processing engines: Analyze text data to detect specific keywords and context. Examples include OpenAI® GPT-3®.

[0327] Emotion Engine: Receives user emotion data, combines it with analytical data, evaluates it, and uses emotion recognition algorithms to analyze the user's mental state.

[0328] Fraud detection algorithms: Evaluate the likelihood of fraud based on the analyzed data. Statistical models and machine learning algorithms are used to calculate a risk score.

[0329] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[0330] Specific examples

[0331] Consider an example where a user is on a call. For example, if the user receives a scam call saying, "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble," the system will perform the following process.

[0332] 1. (Device) records the call audio and sends it to the server. The recording is saved in a digital format such as WAV or MP3 and transmitted securely over HTTPS.

[0333] 2. (Server) receives the audio file and converts it to text data using the Google Cloud Speech-to-Text API.

[0334] 3. (Server) analyzes the text data using a natural language processing engine such as OpenAI GPT-3 to detect keywords and context such as "account," "transfer," and "urgent."

[0335] 4. (Device) analyzes the user's emotions in real time and generates emotional data such as "tension."

[0336] 5. (Server) integrates the emotion data and evaluates the likelihood of fraud. As a result of the evaluation, the risk score is 90, which is determined to be high risk.

[0337] 6. (Server) generates a warning message saying, "This call may be a special fraud. Please be careful." and sends it to the user's terminal, along with additional information saying, "You seem nervous, but please remain calm."

[0338] 7. (Device) displays the received warning message to the user, who can then take appropriate action. For example, the user can end the call if necessary, preventing fraud before it occurs.

[0339] Prompt Sentence Examples

[0340] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[0341] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0342] Step 1: Start recording audio

[0343] (Device) records the user's conversation. When the user starts a call, the device's audio recording device automatically starts and records the conversation in digital format. The input is the user's voice, and the output is a digital audio file (WAV or MP3).

[0344] Step 2: Send the audio file

[0345] The device sends the recorded audio file to the server. The audio file is sent securely using an encrypted communication protocol such as HTTPS. The input is the recorded audio file, and the output is the audio data sent to the server.

[0346] Step 3: Convert speech to text

[0347] The server inputs the received audio file into a speech recognition engine such as the Google Cloud Speech-to-Text API. The speech recognition engine analyzes the audio data and converts it into text data. The input is the audio file, and the output is the converted text data.

[0348] Step 4: Natural Language Processing Analysis

[0349] The server inputs the text data into a natural language processing engine such as OpenAI GPT-3. The engine analyzes the text data and detects specific keywords and contexts such as "account," "transfer," and "urgent." The input is the text data, and the output is the analyzed keywords and context information.

[0350] Step 5: Start Emotion Recognition

[0351] The device analyzes the user's speech in real time and recognizes emotions. It uses emotion recognition algorithms to identify the user's emotions from features such as voice tone, pitch, and speed. The input is real-time voice data, and the output is recognized emotion data.

[0352] Step 6: Sending Emotion Data

[0353] (Device) sends the recognized emotion data to the server. The input is the emotion data, and the output is the emotion data sent to the server.

[0354] Step 7: Data synthesis and evaluation

[0355] The (server) integrates the analyzed keyword, context, and sentiment data. Based on the integrated data, it assesses the likelihood of fraud. It calculates a risk score using statistical models and machine learning algorithms. The input is keyword and context information and sentiment data, and the output is a risk score.

[0356] Step 8: Generate a warning message

[0357] If the server determines the risk score is high, it generates a warning message that includes information indicating the likelihood of fraud and additional information based on the user's sentiment. The input is the risk score and the output is the generated warning message.

[0358] Step 9: Sending a warning message

[0359] (Server) generates and sends the generated warning message to the user terminal. The input is the warning message, and the output is the warning message sent to the user terminal.

[0360] Step 10: Displaying warning messages

[0361] Displays to the user the warning message received by (the terminal). The warning message appears on the screen in a pop-up format and is designed to grab the user's attention. The input is the warning message and the output is the warning message displayed to the user.

[0362] (Application example 2)

[0363] 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."

[0364] In recent years, the methods of special fraud have become more sophisticated, and the damage caused by them has been increasing. Conventional fraud detection systems assess the possibility of fraud solely by analyzing the content of conversations, which limits their accuracy and leaves users at a high risk of falling victim to fraud. Therefore, there is a need for a more accurate fraud detection system that takes into account not only the content of conversations but also the user's emotional state.

[0365] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for recording the content of the conversation, means for converting the recorded content of the conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for analyzing the user's conversation voice in real time and recognizing emotions, means for integrating the emotion data into the text analysis results and evaluating the possibility of fraud, means for generating a warning message and sending it to the user terminal if there is a high possibility of fraud, and means for displaying the warning message on the user terminal. This makes it possible to simultaneously consider the content of the conversation and the user's emotional state and more accurately evaluate the possibility of fraud.

[0366] A "conversation recording device" is a device or system that has the ability to digitally record and store a user's calls and voice messages.

[0367] The "means for converting voice data into text data" is a voice recognition engine that has the function of analyzing recorded voice data and converting it into corresponding text data.

[0368] The "means for analyzing text data and detecting specific keywords and context" is a natural language processing engine for identifying important words and phrases within the text data and analyzing their context.

[0369] The "means for analyzing the user's conversation voice in real time and recognizing emotions" is an emotion recognition engine for analyzing emotions in real time based on the user's voice data.

[0370] The "means for integrating emotion data with text analysis results to assess the likelihood of fraud" is an algorithm for combining emotion recognition results and text analysis results to assess the likelihood of fraud.

[0371] "Means for generating a warning message and sending it to the user terminal when there is a high possibility of fraud" refers to a system that has the function of creating a warning message and notifying the user terminal when it is determined that there is a high risk of fraud.

[0372] The "means for displaying a warning message on a user terminal" refers to a device or software that has a function for visually displaying a warning message on the screen of a user terminal.

[0373] 1. Overall system configuration

[0374] This system consists of a user terminal, a server, and a communication network. The main components are a user terminal that includes a means for recording conversations and sending the audio data to the server, and the server that includes a means for analyzing the received audio data and evaluating the possibility of fraud.

[0375] 2. User Device

[0376] The user terminal has the following functions:

[0377] Voice Recorder: A device that has the ability to record user calls and voice messages. The recorded voice data is saved in a digital format (e.g., WAV or MP3).

[0378] Audio data transmission function: Recorded audio data is encrypted (e.g., AES encrypted) and then securely transmitted to the server. HTTPS is used as the communication protocol.

[0379] Emotion recognition function: Analyzes the user's speech in real time and recognizes emotions. IBM Watson Tone Analyzer and other engines are used as emotion recognition engines.

[0380] Warning display function: Displays the warning message sent from the server on the screen of the user's terminal.

[0381] 3. Server

[0382] The server has the following functions:

[0383] Speech recognition engine: An engine for converting voice data into text data, such as the Google Cloud Speech-to-Text API.

[0384] Natural language processing engines: SpaCy and NLTK are used to analyze text data and detect specific keywords and context.

[0385] Emotion recognition engine: Receives emotion data sent from the user's device and integrates it into the text analysis results. IBM Watson Tone Analyzer and other technologies are used.

[0386] Fraud detection algorithm: An algorithm that assesses the likelihood of fraud based on text analysis and sentiment data. It uses Scikit-learn and TENSORFLOW® models.

[0387] Warning generation function: If there is a high risk of fraud, a warning message is generated and sent to the user's terminal.

[0388] 4. Data Analysis and Alert Generation

[0389] The server inputs the received voice data into a voice recognition engine to generate text data. The generated text data is then analyzed by a natural language processing engine to detect specific keywords and context. At the same time, an emotion recognition engine analyzes the user's emotional data and integrates it into the text analysis results. Finally, a fraud detection algorithm evaluates the data and calculates a risk score. If the risk is assessed as high, a warning message is generated and sent to the user's device.

[0390] 5. Specific usage examples

[0391] Suppose a user is on a call and says something like, "Please transfer 500,000 yen to this account right away. If we don't hurry, we'll be in big trouble." In this case, the following process will occur:

[0392] 1. The user device records the call audio and sends it to the server.

[0393] 2. The server converts the audio file into text data using a speech recognition engine and detects important keywords such as "transfer" and "urgent."

[0394] 3. The emotion recognition engine detects high levels of tension in the user's voice.

[0395] 4. The fraud detection algorithm integrates the emotional data and evaluates the risk score as 90, determining it to be high risk.

[0396] 5. The server generates a warning message stating, "This call may be a special fraud. Please remain calm." and sends it to the user's terminal.

[0397] 6. The user terminal displays a warning message on the screen, allowing the user to take appropriate action.

[0398] Example prompt sentence:

[0399] "If a transfer request is made during a call, analyze the data in real time and assess the risk of special fraud."

[0400] "Please also integrate the results of user sentiment analysis and generate warning messages based on the evaluation results."

[0401] In this way, the present invention simultaneously analyzes the conversation content and user emotions, achieving more accurate fraud detection and user assistance.

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

[0403] Step 1:

[0404] When a user initiates a call, the device records the conversation in real time. The recording is saved in a digital format (e.g., WAV or MP3). The input is the user's voice, and the output is the recorded audio file.

[0405] Step 2:

[0406] The device encrypts the recorded audio file (e.g., AES encryption) and sends it to the server. The input is the recorded audio file, and the output is the encrypted audio file.

[0407] Step 3:

[0408] The server inputs the received audio file into a speech recognition engine (e.g., Google Cloud Speech-to-Text API) and converts the audio data into text data. The input is an encrypted audio file, and the output is text data.

[0409] Step 4:

[0410] The server inputs the generated text data into a natural language processing engine (e.g., spaCy or NLTK) to detect specific keywords and contexts. The input is the text data, and the output is the detected keywords and context information.

[0411] Step 5:

[0412] The device analyzes the user's speech in real time and generates emotion data using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The input is the user's speech data, and the output is emotion data.

[0413] Step 6:

[0414] The server integrates the text analysis results and sentiment data and evaluates the likelihood of fraud using a fraud detection algorithm (e.g., Scikit-learn or TensorFlow model). The inputs are the text analysis results and sentiment data, and the output is a risk score.

[0415] Step 7:

[0416] If the risk score is high, the server uses the warning generation function to generate a warning message and send it to the user terminal. The input is the risk score, and the output is the warning message.

[0417] Step 8:

[0418] The user terminal displays the received warning message on the screen. The input is the warning message, and the output is the warning message displayed on the screen.

[0419] The above is a detailed description of the processing flow of the system program that realizes the application example.

[0420] 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.

[0421] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (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.

[0422] 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.

[0423] [Second embodiment]

[0424] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0425] 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.

[0426] 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).

[0427] 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.

[0428] 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.

[0429] 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).

[0430] 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.

[0431] 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.

[0432] 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.

[0433] 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.

[0434] 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.

[0435] 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."

[0436] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[0437] System configuration overview

[0438] User terminal

[0439] Voice Recorder: Has the ability to record the user's conversations.

[0440] Audio file sending function: Sends recorded audio data to the server.

[0441] Warning display function: Receives warning messages from the server and displays them on the screen.

[0442] server

[0443] Speech recognition engine: Converts voice data into text data.

[0444] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[0445] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[0446] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal.

[0447] Program processing

[0448] Audio Acquisition

[0449] (Device) records conversations. When a user is having a phone or face-to-face conversation, the audio of that conversation is stored digitally.

[0450] Speech-to-text conversion

[0451] The device sends the recorded audio file to the server, where a speech recognition engine is used to convert the audio data into text data.

[0452] Text analytics

[0453] The server analyzes the converted text data using a natural language processing engine to detect specific keywords (e.g., "money," "transfer," "account information," "urgent") and context.

[0454] Determining the likelihood of fraud

[0455] The server determines the possibility of fraud based on the analysis results. If certain conditions are met, a risk score is calculated and the fraud risk level is calculated.

[0456] Warning

[0457] If the server determines that the risk score is high, it generates a warning message and sends it to the user's device, which then displays the warning on its screen to alert the user.

[0458] Specific examples

[0459] Consider the example where the user is on a call.

[0460] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[0461] Processing flow

[0462] 1. (Device) records the call audio and sends it to the server.

[0463] 2. (Server) converts the audio file into text data.

[0464] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[0465] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[0466] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[0467] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[0468] In this way, the present invention analyzes conversation content in real time and quickly detects and warns of fraud risks, helping users prevent themselves from becoming victims of fraud.

[0469] The processing flow will be explained below.

[0470] Step 1:

[0471] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[0472] Step 2:

[0473] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[0474] Step 3:

[0475] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The speech recognition engine uses advanced speech recognition technologies such as Google Speech-to-Text API and IBM Watson.

[0476] Step 4:

[0477] The server inputs the converted text data into a natural language processing engine, which analyzes the text data and detects specific keywords (e.g., "transfer," "account," "urgent"). Natural language processing technologies used include SpaCy and NLTK.

[0478] Step 5:

[0479] The server analyzes the context based on the results of keyword detection. This contextual analysis allows us to understand the context in which the keywords are used and collects data to predict the likelihood of fraud.

[0480] Step 6:

[0481] The server evaluates the likelihood of fraud based on the collected data. The evaluation is performed using statistical models and machine learning algorithms to calculate a risk score. For example, if certain conditions are met, a risk score is calculated on a scale of 0 to 100.

[0482] Step 7:

[0483] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action.

[0484] Step 8:

[0485] The server generates a warning message and sends it to the user's device. This communication also uses an encrypted protocol to ensure security.

[0486] Step 9:

[0487] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[0488] Step 10:

[0489] The user checks the warning message and takes appropriate action. If the possibility of fraud is judged to be high, the user can choose to take action such as ending the call to prevent fraud from occurring. If necessary, the system also provides a function to automatically end the call based on the user's selection.

[0490] Example 1

[0491] 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."

[0492] In recent years, there has been a demand for technology to prevent special frauds, but current methods have difficulty detecting potential fraud in real time and issuing warnings to users. In particular, as fraud methods become more sophisticated and diverse, even a momentary error in judgment can result in significant damage. This leaves users constantly exposed to the risk of fraud and makes it difficult to have conversations with peace of mind. Therefore, a system is needed that can detect potential fraud in real time and issue immediate warnings.

[0493] 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.

[0494] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to the user terminal if the possibility of fraud is high, means for displaying the warning message on the user terminal, means for transmitting the voice data to the server via a network, and means for calculating a risk score in the analysis and determining the fraud risk level, thereby enabling rapid detection and warning of fraud risks in real time.

[0495] "Means for recording conversations" refers to devices or functions that digitally record a user's conversations.

[0496] "Means for converting the recorded conversation content from voice data to text data" refers to a function that uses voice recognition technology to analyze recorded voice data and convert it into text data as character string information.

[0497] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses natural language processing technology to extract and analyze important keywords and related contexts contained in text data.

[0498] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or rule sets for assessing the risk of specialized fraud based on analyzed keywords and contextual information.

[0499] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function for automatically creating a warning message and sending it to a user terminal when it is determined that there is a high risk of fraud.

[0500] "Means for displaying a warning message on a user terminal" refers to a function for visually displaying a received warning message on a display or the like of a user terminal.

[0501] "Means for transmitting audio data to a server via a network" refers to the function of transmitting recorded audio data to a server using the Internet or other networks.

[0502] "Means for calculating a risk score in the analysis and determining the fraud risk level" refers to an algorithm or method for deriving a risk score from the results of text analysis and determining the fraud risk level based on that score.

[0503] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[0504] System configuration overview

[0505] User terminal

[0506] Audio Recorder: Has the ability to record user conversations. The device's audio recorder records conversations in real time in digital format.

[0507] Audio file transmission function: This function transmits recorded audio data to a server via the network. When recording is finished, the audio data is automatically sent to the server.

[0508] Warning display function: Receives warning messages from the server and displays them on the screen. Received warning messages are displayed on the display to alert the user.

[0509] server

[0510] Speech recognition engine: Converts voice data into text data. Examples of use include Google Speech-to-Text and IBM Watson Speech to Text.

[0511] Natural language processing engine: Analyzes text data and detects specific keywords and contexts. Examples of use include spaCy and BERT.

[0512] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data. Calculates a risk score based on specific keywords and context to determine the risk level.

[0513] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal. A warning message is automatically generated and sent to the user's terminal.

[0514] Specific examples

[0515] Consider the example where the user is on a call.

[0516] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[0517] Processing flow

[0518] 1. (Device) records the call audio and sends it to the server.

[0519] When a user has a conversation over the phone or in person, the device's audio recorder stores the audio in digital form.

[0520] 2. (Server) converts the audio file into text data.

[0521] The server's speech recognition engine (e.g., Google Speech-to-Text) receives the audio file as input and converts it into text information.

[0522] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[0523] The server's natural language processing engine (e.g., spaCy or BERT) performs the analysis and detects specific keywords and context.

[0524] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[0525] A risk score is calculated based on the keywords and context detected, with a score of 80 or above being considered high risk.

[0526] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[0527] If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user terminal.

[0528] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[0529] As soon as the user terminal receives the warning message, it immediately displays it on the screen to alert the user.

[0530] Example prompts for generative AI models

[0531] "Analyze this conversation and assess its fraud risk: 'Please transfer $5,000 to this account immediately. Hurry or you'll be in trouble.'"

[0532] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0533] Step 1:

[0534] (Device) records user conversations.

[0535] Input: User's spoken speech

[0536] Specific operation: The device's audio recording device records the conversation in real time in digital format.

[0537] Output: Recorded audio data (digital format)

[0538] Step 2:

[0539] (Device) sends the recorded audio data to the server.

[0540] Input: Recorded audio data

[0541] Specific operation: When recording is finished, the audio data is automatically sent to the server via the network. Communication to the server is performed using a secure protocol (e.g., HTTPS).

[0542] Output: Audio data sent to the server

[0543] Step 3:

[0544] The server starts a speech recognition engine and converts the speech data into text data.

[0545] Input: Audio data sent to the server

[0546] Specific operation: The server's speech recognition engine (for example, Google Speech-to-Text) analyzes the audio file and converts it into text data as string information.

[0547] Output: Converted text data

[0548] Step 4:

[0549] (Server) analyzes the converted text data using a natural language processing engine.

[0550] Input: Converted text data

[0551] How it works: The server's natural language processing engine (such as spaCy or BERT) performs detailed analysis of the text data to detect specific keywords and context. It extracts important keywords such as "money," "transfer," "account information," and "hurry," and analyzes the context.

[0552] Output: Analysis results (detected keywords and context information)

[0553] Step 5:

[0554] (Server) evaluates the possibility of fraud based on the analysis results.

[0555] Input: Analysis results (detected keywords and context information)

[0556] Specific behavior: Calculate a risk score based on the detected keywords and context. For example, assign a weight to each keyword and evaluate the match with the context to score the overall risk. If the risk score is high, it is determined to be high risk.

[0557] Output: Risk score, risk assessment result

[0558] Step 6:

[0559] The server generates a warning message and sends it to the user terminal.

[0560] Input: Risk assessment result (if high risk)

[0561] Specific operation: If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user's device stating, "This call may be a special fraud. Please be careful."

[0562] Output: Generated warning message

[0563] Step 7:

[0564] (Device) receives the warning message and displays it on the screen.

[0565] Input: The generated warning message

[0566] Specific operation: As soon as the user device receives a warning message, it immediately displays it on the screen to alert the user. In addition to the visual display, additional notification methods such as audio alerts can also be used.

[0567] Output: Display warning message, notify user

[0568] (Application example 1)

[0569] 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."

[0570] In modern society, the methods used by special fraudsters are becoming increasingly sophisticated, and many people are falling victim to fraud. Elderly people and users unfamiliar with the Internet are particularly susceptible to fraud. As a result, not only financial damage but also psychological damage is severe. The present invention aims to prevent such special frauds from occurring. Specifically, the objective of this invention is to provide a system that prevents fraud damage by detecting possible fraud in real time and issuing a warning to the user.

[0571] 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.

[0572] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from audio data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud, means for displaying the warning message on the user terminal, means for the user to select a response action when the warning message is displayed, and means for analyzing the text data using a generative AI model. This allows for highly accurate detection of the possibility of fraud in real time, enabling the user to take immediate response action, thereby preventing fraud damage.

[0573] A "conversation recording device" is a device or software that has the ability to record a user's telephone or face-to-face conversations in real time and store them in digital form.

[0574] "Means for converting recorded conversation content from voice data to text data" refers to a system that uses voice recognition technology to convert recorded voice files into text format.

[0575] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses a natural language processing engine to analyze important words and phrases contained in text data and their contexts.

[0576] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or functions that calculate a fraud risk score and determine the likelihood of fraud based on the analyzed text data.

[0577] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function that generates a warning message and sends it to a user terminal via a network when it is determined that there is a high risk of fraud.

[0578] The "means for displaying a warning message on a user terminal" refers to a display or application for visually displaying to the user the warning message received on the user terminal.

[0579] "Means for the user to select a response action when a warning message is displayed" refers to an interface or function that allows the user to check the warning message and then select an appropriate response action themselves.

[0580] "Means for securely transmitting voice data to a server" refers to a function that uses encryption technology and communication protocols to maintain data integrity and confidentiality when transmitting voice data from a user terminal to a server.

[0581] "Means for analyzing text data using a generative AI model" refers to a function that uses a natural language processing engine that incorporates AI technology to analyze text data with high accuracy.

[0582] As an embodiment of the present invention, a system for detecting and warning against special frauds will be described in detail. This system mainly operates using a user terminal, a server, and a communication network. Specific embodiments of this technology are described below.

[0583] User terminal

[0584] The user terminal implements the following main functions:

[0585] 1. Voice recording function

[0586] The user terminal is equipped with a microphone for recording voice, which automatically records the voice during conversation and stores it in digital format.

[0587] 2. Audio file sending function

[0588] The recorded audio data is securely transmitted to the server using encryption technology, which is essential to maintain data integrity and confidentiality.

[0589] 3. Warning display function

[0590] Displays warning messages received from the server to the user. Warning messages are displayed in a popup or notification bar so that the user can see them immediately.

[0591] server

[0592] The server performs the primary processing for analyzing the voice data and assessing the likelihood of fraud.

[0593] 1. Speech Recognition Engine

[0594] The server receives the voice data sent from the user's device and converts it into text data using a voice recognition engine, which uses the Google speech recognition API.

[0595] 2. Text Analysis Engine

[0596] A natural language processing engine using generative AI models analyzes text data, detecting specific keywords (e.g., "money," "transfer," "account") and context to assess the likelihood of fraud.

[0597] 3. Fraud Detection Algorithm

[0598] A fraud risk score is calculated based on the detected keywords and context. If the risk score exceeds a certain threshold, it is determined to be highly likely to be fraud. For example, if a conversation such as "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble" is detected, it is determined to be high risk.

[0599] 4. Warning generation function

[0600] If the risk is determined to be high, a warning message is generated and sent to the user's terminal. The warning message may include a statement such as, "This call may be a special fraud. Please be careful."

[0601] Specific examples

[0602] A specific scenario is shown below.

[0603] Example Scenario

[0604] While on the phone, the user receives the message, "Please transfer 500,000 yen to this account immediately."

[0605] The user terminal records this conversation and sends the audio file to the server.

[0606] The server converts the audio file into text data and detects specific keywords such as "account," "transfer," and "urgent."

[0607] Based on this text data, a generative AI model calculates a fraud risk score and determines it as high risk.

[0608] The server generates a warning message and sends it to the user's terminal stating, "This call may be a special fraud. Please be careful."

[0609] The user terminal displays a warning message and the user takes appropriate action.

[0610] Prompt Sentence Examples

[0611] Below are some example prompts for the generative AI model:

[0612] "Please rate the likelihood of fraud for the following text:

[0613] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[0614] With such detailed embodiments, the present invention is able to detect fraud risks in real time and issue immediate warnings to users.

[0615] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0616] Step 1: Recording audio data

[0617] The user terminal records telephone or face-to-face conversations in real time, and the recorded data is stored in digital format. The input is the user's conversational voice, and the output is the recorded voice data.

[0618] Step 2: Send your recordings

[0619] The device transmits the recorded voice data to the server. The transmission is performed using encryption technology to maintain data integrity and confidentiality. The input is the recorded voice data, and the output is the transmission of encrypted voice data.

[0620] Step 3: Convert audio data to text

[0621] The server converts the received voice data into text data using a voice recognition engine (e.g., Google voice recognition API). The input is encrypted voice data, and the output is text data.

[0622] Step 4: Analyzing the text data

[0623] The server uses a generative AI model to analyze the text data and detect specific keywords and context. Specifically, a natural language processing engine scans the text data and searches for keywords such as "money," "transfer," and "account." The input is the text data, and the output is the analyzed keywords and context information.

[0624] Step 5: Fraud risk assessment

[0625] The server calculates a fraud risk score based on the analyzed keywords and context using a pre-trained fraud detection algorithm, where the input is keywords and context information and the output is a risk score.

[0626] Step 6: Generate a warning message

[0627] Once the risk score is determined, the server automatically generates a warning message if it determines that there is a high possibility of fraud. The warning message includes content such as "This call may be a special fraud. Please be careful." The input is the risk score, and the output is the warning message.

[0628] Step 7: Sending a warning message

[0629] The server sends the generated warning message to the user terminal. The input is the warning message and the output is the transmission to the user terminal.

[0630] Step 8: Displaying warning messages

[0631] The terminal displays the received warning message to the user. A pop-up notification or an alert message is used for display so that the user can immediately check it. The input is the warning message, and the output is the warning displayed on the user's screen.

[0632] Step 9: User Actions

[0633] The user checks the warning message displayed on the terminal and selects an appropriate response action. Based on the warning, the user may end the call or perform additional confirmation. The input is the warning message, and the output is the user's response action.

[0634] 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.

[0635] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[0636] System configuration overview

[0637] User terminal

[0638] Voice Recorder: Provides the ability to record a user's conversations and store them in digital form.

[0639] Audio file sending function: Sends recorded audio data to the server.

[0640] Emotion recognition function: Analyzes the user's conversational voice in real time and recognizes emotions.

[0641] Warning display function: Receives warning messages from the server and displays them on the screen.

[0642] server

[0643] Speech recognition engine: Converts voice data into text data.

[0644] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[0645] Emotion engine: Receives user emotion data and evaluates it in combination with analytical data.

[0646] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[0647] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[0648] Program processing

[0649] Audio Acquisition

[0650] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[0651] Speech-to-text conversion

[0652] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[0653] emotion recognition

[0654] The device analyzes the user's speech in real time and recognizes emotions. Emotion recognition technology uses algorithms based on characteristics such as tone, pitch, and speed of speech.

[0655] Text analytics

[0656] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The natural language processing engine then analyzes the text data to detect specific keywords and context.

[0657] Emotional Data Integration

[0658] The server integrates emotional data into the results of text analysis. Emotional data reflects the user's psychological state, and combining this information improves the accuracy of fraud likelihood assessment.

[0659] Determining the likelihood of fraud

[0660] The server evaluates the likelihood of fraud based on the integrated data. Statistical models and machine learning algorithms are used to calculate a risk score. The inclusion of emotional data also enhances the accuracy of the risk assessment.

[0661] Warning

[0662] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action. The warning message also includes additional information tailored to the user's psychological state based on emotional data.

[0663] Displaying warnings

[0664] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[0665] Specific examples

[0666] Consider the example where the user is on a call.

[0667] Call details: "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into problems."

[0668] Processing flow

[0669] 1. (Device) records the call audio and sends it to the server.

[0670] 2. (Server) converts the audio file into text data, and a natural language processing engine detects keywords such as "account," "transfer," and "urgent" and analyzes the context.

[0671] 3. (Device) analyzes the user's emotions in real time and generates emotion data. In this case, it detects that the user is feeling tense.

[0672] 4. (Server) integrates the emotion data and evaluates the likelihood of fraud. The risk score is evaluated as 90, which is determined to be high risk.

[0673] 5. (Server) generates a warning message saying, "This call may be a fraudulent transaction. Please be careful." and sends it to the user. Based on the emotional data, it also includes additional information such as, "You seem nervous, but please remain calm."

[0674] 6. (Device) displays a warning message, and the user can respond appropriately. If necessary, the call can be terminated, preventing fraud before it occurs.

[0675] The present invention simultaneously analyzes the conversation content and user emotions to achieve more accurate fraud detection and user assistance. This system allows users to quickly recognize fraud risks and respond appropriately.

[0676] The processing flow will be explained below.

[0677] Step 1:

[0678] The device records the user's conversation. The recording is initiated by the user's operation or a voice trigger, and is saved in a digital format, such as WAV or MP3.

[0679] Step 2:

[0680] The (device) sends the recorded audio file to the server using an encrypted communication protocol (e.g., HTTPS).

[0681] Step 3:

[0682] While recording, the device analyzes the user's speech in real time and uses an emotion recognition engine to detect the user's emotional state, using algorithms based on voice tone, pitch, speed, etc.

[0683] Step 4:

[0684] The server inputs the received audio file into a speech recognition engine, which converts the audio data into text data. The speech recognition engine used is Google Speech-to-Text API or IBM Watson.

[0685] Step 5:

[0686] The server inputs the text data into a natural language processing engine, which analyzes specific keywords (e.g., "transfer," "account," "urgent") and context. Natural language processing is performed using tools such as SpaCy and NLTK.

[0687] Step 6:

[0688] The server detects the context based on the results of text analysis and assesses the likelihood of fraud based on this, using statistical models and machine learning algorithms.

[0689] Step 7:

[0690] The device recognizes the emotion data in real time and sends it to the server. This data may include the user's emotions such as tension, impatience, and fear.

[0691] Step 8:

[0692] The server integrates the results of text analysis with emotion data and reassesssses the likelihood of fraud. Including emotion data improves the accuracy of the assessment.

[0693] Step 9:

[0694] The server calculates a risk score based on the integrated data, and if it is determined to be high risk, it generates a warning message informing the user that it may be a scam and urging them to act calmly.

[0695] Step 10:

[0696] The server generates a warning message and sends it to the user's device. This message is sent using a secure communication protocol.

[0697] Step 11:

[0698] (Device) displays received warning messages to the user. The messages are displayed in a pop-up format and have an attention-grabbing design.

[0699] Step 12:

[0700] The user checks the warning message and takes appropriate action. For example, if it is determined that there is a high possibility of fraud, the user can take action such as terminating the call. The message also includes additional information (e.g., "Please remain calm") tailored to the user's psychological state based on emotional data.

[0701] In this way, the present invention provides a system that analyzes conversation content and emotions in real time, assesses the possibility of fraud with high accuracy, and quickly issues a warning to the user, thereby preventing fraud damage before it occurs.

[0702] Example 2

[0703] 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."

[0704] Special fraud remains a serious social problem, with many people falling victim to it. Conventional fraud detection systems make judgments based solely on the analysis of text data, making it difficult to accurately assess the likelihood of fraud. Furthermore, warnings are issued without taking into account the user's emotions, which can lead to users being unable to respond appropriately. Therefore, there is a need for a new system that can more accurately detect potential fraud and provide warning messages that take the user's emotions into account.

[0705] 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.

[0706] In this invention, the server includes a means for analyzing the user's conversation voice in real time and recognizing emotions, a means for evaluating the possibility of fraud based on detected keywords and context and the recognized emotion data, and a means for generating a warning message and sending it to the user terminal if the possibility of fraud is high. This allows for a more accurate warning by combining text data and emotion data to evaluate the possibility of fraud.

[0707] "Conversation recording means" refers to a method or device for digitally recording and storing a user's voice.

[0708] "Means for converting voice data into text data" refers to technology or devices that analyze recorded voice data and convert it into corresponding text data.

[0709] "Means for analyzing text data and detecting specific keywords and context" refers to technology that analyzes text data converted from audio data and identifies specific words and phrases and the context behind them.

[0710] "Means for analyzing a user's conversational voice in real time and recognizing emotions" refers to a method or device that processes a user's voice in real time and identifies emotions from characteristics of the voice such as tone, pitch, and speed.

[0711] "Means for assessing the likelihood of fraud based on detected keywords and context, and recognized sentiment data" refers to algorithms and technologies that combine analysis results with sentiment data to determine the likelihood of fraud.

[0712] "Means for generating and sending a warning message to a user terminal" refers to technology or equipment that, when a possibility of fraud is determined, creates an appropriate warning message and sends that message to a user terminal.

[0713] "Means for displaying a warning message on a user terminal" refers to a method or device for visually displaying a received warning message on a user terminal.

[0714] "Means for safely transmitting voice data and emotion data to a server" refers to technology for transmitting voice data and emotion data to a server after implementing safety measures such as encryption.

[0715] "Means for the user to select a response action based on the notified warning message and additional information" refers to technology or devices that allow the user to select the most appropriate action based on the received warning message and the provided additional information.

[0716] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[0717] System configuration overview

[0718] User terminal

[0719] Audio Recording Device: A device that has the ability to record a user's conversation and store it digitally. Examples include smartphones and dedicated recording devices.

[0720] Audio file transmission function: Recorded audio data is sent to the server using an encrypted communication protocol (e.g. HTTPS).

[0721] Emotion Recognition: Analyzes the user's speech in real time to recognize emotions. Emotion recognition technology uses algorithms based on voice characteristics such as tone, pitch, and speed.

[0722] Warning display function: Receives warning messages from the server and displays them on the screen. For example, a warning is displayed in pop-up format on a smartphone screen.

[0723] server

[0724] Speech recognition engine: Converts voice data into text data. Specifically, Google Cloud Speech-to-Text API is used.

[0725] Natural language processing engines: Analyze text data to detect specific keywords and context. Examples include OpenAI GPT-3.

[0726] Emotion Engine: Receives user emotion data, combines it with analytical data, evaluates it, and uses emotion recognition algorithms to analyze the user's mental state.

[0727] Fraud detection algorithms: Evaluate the likelihood of fraud based on the analyzed data. Statistical models and machine learning algorithms are used to calculate a risk score.

[0728] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[0729] Specific examples

[0730] Consider an example where a user is on a call. For example, if the user receives a scam call saying, "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble," the system will perform the following process.

[0731] 1. (Device) records the call audio and sends it to the server. The recording is saved in a digital format such as WAV or MP3 and transmitted securely over HTTPS.

[0732] 2. (Server) receives the audio file and converts it to text data using the Google Cloud Speech-to-Text API.

[0733] 3. (Server) analyzes the text data using a natural language processing engine such as OpenAI GPT-3 to detect keywords and context such as "account," "transfer," and "urgent."

[0734] 4. (Device) analyzes the user's emotions in real time and generates emotional data such as "tension."

[0735] 5. (Server) integrates the emotion data and evaluates the likelihood of fraud. As a result of the evaluation, the risk score is 90, which is determined to be high risk.

[0736] 6. (Server) generates a warning message saying, "This call may be a special fraud. Please be careful." and sends it to the user's terminal, along with additional information saying, "You seem nervous, but please remain calm."

[0737] 7. (Device) displays the received warning message to the user, who can then take appropriate action. For example, the user can end the call if necessary, preventing fraud before it occurs.

[0738] Prompt Sentence Examples

[0739] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[0740] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0741] Step 1: Start recording audio

[0742] (Device) records the user's conversation. When the user starts a call, the device's audio recording device automatically starts and records the conversation in digital format. The input is the user's voice, and the output is a digital audio file (WAV or MP3).

[0743] Step 2: Send the audio file

[0744] The device sends the recorded audio file to the server. The audio file is sent securely using an encrypted communication protocol such as HTTPS. The input is the recorded audio file, and the output is the audio data sent to the server.

[0745] Step 3: Convert speech to text

[0746] The server inputs the received audio file into a speech recognition engine such as the Google Cloud Speech-to-Text API. The speech recognition engine analyzes the audio data and converts it into text data. The input is the audio file, and the output is the converted text data.

[0747] Step 4: Natural Language Processing Analysis

[0748] The server inputs the text data into a natural language processing engine such as OpenAI GPT-3. The engine analyzes the text data and detects specific keywords and contexts such as "account," "transfer," and "urgent." The input is the text data, and the output is the analyzed keywords and context information.

[0749] Step 5: Start Emotion Recognition

[0750] The device analyzes the user's speech in real time and recognizes emotions. It uses emotion recognition algorithms to identify the user's emotions from features such as voice tone, pitch, and speed. The input is real-time voice data, and the output is recognized emotion data.

[0751] Step 6: Sending Emotion Data

[0752] (Device) sends the recognized emotion data to the server. The input is the emotion data, and the output is the emotion data sent to the server.

[0753] Step 7: Data synthesis and evaluation

[0754] The (server) integrates the analyzed keyword, context, and sentiment data. Based on the integrated data, it assesses the likelihood of fraud. It calculates a risk score using statistical models and machine learning algorithms. The input is keyword and context information and sentiment data, and the output is a risk score.

[0755] Step 8: Generate a warning message

[0756] If the server determines the risk score is high, it generates a warning message that includes information indicating the likelihood of fraud and additional information based on the user's sentiment. The input is the risk score and the output is the generated warning message.

[0757] Step 9: Sending a warning message

[0758] (Server) generates and sends the generated warning message to the user terminal. The input is the warning message, and the output is the warning message sent to the user terminal.

[0759] Step 10: Displaying warning messages

[0760] Displays to the user the warning message received by (the terminal). The warning message appears on the screen in a pop-up format and is designed to grab the user's attention. The input is the warning message and the output is the warning message displayed to the user.

[0761] (Application example 2)

[0762] 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."

[0763] In recent years, the methods of special fraud have become more sophisticated, and the damage caused by them has been increasing. Conventional fraud detection systems assess the possibility of fraud solely by analyzing the content of conversations, which limits their accuracy and leaves users at a high risk of falling victim to fraud. Therefore, there is a need for a more accurate fraud detection system that takes into account not only the content of conversations but also the user's emotional state.

[0764] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for recording the content of the conversation, means for converting the recorded content of the conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for analyzing the user's conversation voice in real time and recognizing emotions, means for integrating the emotion data into the text analysis results and evaluating the possibility of fraud, means for generating a warning message and sending it to the user terminal if there is a high possibility of fraud, and means for displaying the warning message on the user terminal. This makes it possible to simultaneously consider the content of the conversation and the user's emotional state and more accurately evaluate the possibility of fraud.

[0765] A "conversation recording device" is a device or system that has the ability to digitally record and store a user's calls and voice messages.

[0766] The "means for converting voice data into text data" is a voice recognition engine that has the function of analyzing recorded voice data and converting it into corresponding text data.

[0767] The "means for analyzing text data and detecting specific keywords and context" is a natural language processing engine for identifying important words and phrases within the text data and analyzing their context.

[0768] The "means for analyzing the user's conversation voice in real time and recognizing emotions" is an emotion recognition engine for analyzing emotions in real time based on the user's voice data.

[0769] The "means for integrating emotion data with text analysis results to assess the likelihood of fraud" is an algorithm for combining emotion recognition results and text analysis results to assess the likelihood of fraud.

[0770] "Means for generating a warning message and sending it to the user terminal when there is a high possibility of fraud" refers to a system that has the function of creating a warning message and notifying the user terminal when it is determined that there is a high risk of fraud.

[0771] The "means for displaying a warning message on a user terminal" refers to a device or software that has a function for visually displaying a warning message on the screen of a user terminal.

[0772] 1. Overall system configuration

[0773] This system consists of a user terminal, a server, and a communication network. The main components are a user terminal that includes a means for recording conversations and sending the audio data to the server, and the server that includes a means for analyzing the received audio data and evaluating the possibility of fraud.

[0774] 2. User Device

[0775] The user terminal has the following functions:

[0776] Voice Recorder: A device that has the ability to record user calls and voice messages. The recorded voice data is saved in a digital format (e.g., WAV or MP3).

[0777] Audio data transmission function: Recorded audio data is encrypted (e.g., AES encrypted) and then securely transmitted to the server. HTTPS is used as the communication protocol.

[0778] Emotion recognition function: Analyzes the user's speech in real time and recognizes emotions. IBM Watson Tone Analyzer and other engines are used as emotion recognition engines.

[0779] Warning display function: Displays the warning message sent from the server on the screen of the user's terminal.

[0780] 3. Server

[0781] The server has the following functions:

[0782] Speech recognition engine: An engine for converting voice data into text data, such as the Google Cloud Speech-to-Text API.

[0783] Natural language processing engines: SpaCy and NLTK are used to analyze text data and detect specific keywords and context.

[0784] Emotion recognition engine: Receives emotion data sent from the user's device and integrates it into the text analysis results. IBM Watson Tone Analyzer and other technologies are used.

[0785] Fraud detection algorithm: An algorithm that assesses the likelihood of fraud based on text analysis and sentiment data. It uses Scikit-learn and TensorFlow models.

[0786] Warning generation function: If there is a high risk of fraud, a warning message is generated and sent to the user's terminal.

[0787] 4. Data Analysis and Alert Generation

[0788] The server inputs the received voice data into a voice recognition engine to generate text data. The generated text data is then analyzed by a natural language processing engine to detect specific keywords and context. At the same time, an emotion recognition engine analyzes the user's emotional data and integrates it into the text analysis results. Finally, a fraud detection algorithm evaluates the data and calculates a risk score. If the risk is assessed as high, a warning message is generated and sent to the user's device.

[0789] 5. Specific usage examples

[0790] Suppose a user is on a call and says something like, "Please transfer 500,000 yen to this account right away. If we don't hurry, we'll be in big trouble." In this case, the following process will occur:

[0791] 1. The user device records the call audio and sends it to the server.

[0792] 2. The server converts the audio file into text data using a speech recognition engine and detects important keywords such as "transfer" and "urgent."

[0793] 3. The emotion recognition engine detects high levels of tension in the user's voice.

[0794] 4. The fraud detection algorithm integrates the emotional data and evaluates the risk score as 90, determining it to be high risk.

[0795] 5. The server generates a warning message stating, "This call may be a special fraud. Please remain calm." and sends it to the user's terminal.

[0796] 6. The user terminal displays a warning message on the screen, allowing the user to take appropriate action.

[0797] Example prompt sentence:

[0798] "If a transfer request is made during a call, analyze the data in real time and assess the risk of special fraud."

[0799] "Please also integrate the results of user sentiment analysis and generate warning messages based on the evaluation results."

[0800] In this way, the present invention simultaneously analyzes the conversation content and user emotions, achieving more accurate fraud detection and user assistance.

[0801] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0802] Step 1:

[0803] When a user initiates a call, the device records the conversation in real time. The recording is saved in a digital format (e.g., WAV or MP3). The input is the user's voice, and the output is the recorded audio file.

[0804] Step 2:

[0805] The device encrypts the recorded audio file (e.g., AES encryption) and sends it to the server. The input is the recorded audio file, and the output is the encrypted audio file.

[0806] Step 3:

[0807] The server inputs the received audio file into a speech recognition engine (e.g., Google Cloud Speech-to-Text API) and converts the audio data into text data. The input is an encrypted audio file, and the output is text data.

[0808] Step 4:

[0809] The server inputs the generated text data into a natural language processing engine (e.g., spaCy or NLTK) to detect specific keywords and contexts. The input is the text data, and the output is the detected keywords and context information.

[0810] Step 5:

[0811] The device analyzes the user's speech in real time and generates emotion data using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The input is the user's speech data, and the output is emotion data.

[0812] Step 6:

[0813] The server integrates the text analysis results and sentiment data and evaluates the likelihood of fraud using a fraud detection algorithm (e.g., Scikit-learn or TensorFlow model). The inputs are the text analysis results and sentiment data, and the output is a risk score.

[0814] Step 7:

[0815] If the risk score is high, the server uses the warning generation function to generate a warning message and send it to the user terminal. The input is the risk score, and the output is the warning message.

[0816] Step 8:

[0817] The user terminal displays the received warning message on the screen. The input is the warning message, and the output is the warning message displayed on the screen.

[0818] The above is a detailed description of the processing flow of the system program that realizes the application example.

[0819] 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.

[0820] 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.

[0821] 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.

[0822] [Third embodiment]

[0823] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0824] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0825] 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).

[0826] 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.

[0827] 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.

[0828] 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).

[0829] 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.

[0830] 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.

[0831] 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.

[0832] 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.

[0833] 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.

[0834] 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."

[0835] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[0836] System configuration overview

[0837] User terminal

[0838] Voice Recorder: Has the ability to record the user's conversations.

[0839] Audio file sending function: Sends recorded audio data to the server.

[0840] Warning display function: Receives warning messages from the server and displays them on the screen.

[0841] server

[0842] Speech recognition engine: Converts voice data into text data.

[0843] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[0844] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[0845] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal.

[0846] Program processing

[0847] Audio Acquisition

[0848] (Device) records conversations. When a user is having a phone or face-to-face conversation, the audio of that conversation is stored digitally.

[0849] Speech-to-text conversion

[0850] The device sends the recorded audio file to the server, where a speech recognition engine is used to convert the audio data into text data.

[0851] Text analytics

[0852] The server analyzes the converted text data using a natural language processing engine to detect specific keywords (e.g., "money," "transfer," "account information," "urgent") and context.

[0853] Determining the likelihood of fraud

[0854] The server determines the possibility of fraud based on the analysis results. If certain conditions are met, a risk score is calculated and the fraud risk level is calculated.

[0855] Warning

[0856] If the server determines that the risk score is high, it generates a warning message and sends it to the user's device, which then displays the warning on its screen to alert the user.

[0857] Specific examples

[0858] Consider the example where the user is on a call.

[0859] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[0860] Processing flow

[0861] 1. (Device) records the call audio and sends it to the server.

[0862] 2. (Server) converts the audio file into text data.

[0863] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[0864] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[0865] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[0866] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[0867] In this way, the present invention analyzes conversation content in real time and quickly detects and warns of fraud risks, helping users prevent themselves from becoming victims of fraud.

[0868] The processing flow will be explained below.

[0869] Step 1:

[0870] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[0871] Step 2:

[0872] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[0873] Step 3:

[0874] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The speech recognition engine uses advanced speech recognition technologies such as Google Speech-to-Text API and IBM Watson.

[0875] Step 4:

[0876] The server inputs the converted text data into a natural language processing engine, which analyzes the text data and detects specific keywords (e.g., "transfer," "account," "urgent"). Natural language processing technologies used include SpaCy and NLTK.

[0877] Step 5:

[0878] The server analyzes the context based on the results of keyword detection. This contextual analysis allows us to understand the context in which the keywords are used and collects data to predict the likelihood of fraud.

[0879] Step 6:

[0880] The server evaluates the likelihood of fraud based on the collected data. The evaluation is performed using statistical models and machine learning algorithms to calculate a risk score. For example, if certain conditions are met, a risk score is calculated on a scale of 0 to 100.

[0881] Step 7:

[0882] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action.

[0883] Step 8:

[0884] The server generates a warning message and sends it to the user's device. This communication also uses an encrypted protocol to ensure security.

[0885] Step 9:

[0886] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[0887] Step 10:

[0888] The user checks the warning message and takes appropriate action. If the possibility of fraud is judged to be high, the user can choose to take action such as ending the call to prevent fraud from occurring. If necessary, the system also provides a function to automatically end the call based on the user's selection.

[0889] Example 1

[0890] 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."

[0891] In recent years, there has been a demand for technology to prevent special frauds, but current methods have difficulty detecting potential fraud in real time and issuing warnings to users. In particular, as fraud methods become more sophisticated and diverse, even a momentary error in judgment can result in significant damage. This leaves users constantly exposed to the risk of fraud and makes it difficult to have conversations with peace of mind. Therefore, a system is needed that can detect potential fraud in real time and issue immediate warnings.

[0892] 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.

[0893] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to the user terminal if the possibility of fraud is high, means for displaying the warning message on the user terminal, means for transmitting the voice data to the server via a network, and means for calculating a risk score in the analysis and determining the fraud risk level, thereby enabling rapid detection and warning of fraud risks in real time.

[0894] "Means for recording conversations" refers to devices or functions that digitally record a user's conversations.

[0895] "Means for converting the recorded conversation content from voice data to text data" refers to a function that uses voice recognition technology to analyze recorded voice data and convert it into text data as character string information.

[0896] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses natural language processing technology to extract and analyze important keywords and related contexts contained in text data.

[0897] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or rule sets for assessing the risk of specialized fraud based on analyzed keywords and contextual information.

[0898] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function for automatically creating a warning message and sending it to a user terminal when it is determined that there is a high risk of fraud.

[0899] "Means for displaying a warning message on a user terminal" refers to a function for visually displaying a received warning message on a display or the like of a user terminal.

[0900] "Means for transmitting audio data to a server via a network" refers to the function of transmitting recorded audio data to a server using the Internet or other networks.

[0901] "Means for calculating a risk score in the analysis and determining the fraud risk level" refers to an algorithm or method for deriving a risk score from the results of text analysis and determining the fraud risk level based on that score.

[0902] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[0903] System configuration overview

[0904] User terminal

[0905] Audio Recorder: Has the ability to record user conversations. The device's audio recorder records conversations in real time in digital format.

[0906] Audio file transmission function: This function transmits recorded audio data to a server via the network. When recording is finished, the audio data is automatically sent to the server.

[0907] Warning display function: Receives warning messages from the server and displays them on the screen. Received warning messages are displayed on the display to alert the user.

[0908] server

[0909] Speech recognition engine: Converts voice data into text data. Examples of use include Google Speech-to-Text and IBM Watson Speech to Text.

[0910] Natural language processing engine: Analyzes text data and detects specific keywords and contexts. Examples of use include spaCy and BERT.

[0911] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data. Calculates a risk score based on specific keywords and context to determine the risk level.

[0912] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal. A warning message is automatically generated and sent to the user's terminal.

[0913] Specific examples

[0914] Consider the example where the user is on a call.

[0915] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[0916] Processing flow

[0917] 1. (Device) records the call audio and sends it to the server.

[0918] When a user has a conversation over the phone or in person, the device's audio recorder stores the audio in digital form.

[0919] 2. (Server) converts the audio file into text data.

[0920] The server's speech recognition engine (e.g., Google Speech-to-Text) receives the audio file as input and converts it into text information.

[0921] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[0922] The server's natural language processing engine (e.g., spaCy or BERT) performs the analysis and detects specific keywords and context.

[0923] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[0924] A risk score is calculated based on the keywords and context detected, with a score of 80 or above being considered high risk.

[0925] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[0926] If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user terminal.

[0927] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[0928] As soon as the user terminal receives the warning message, it immediately displays it on the screen to alert the user.

[0929] Example prompts for generative AI models

[0930] "Analyze this conversation and assess its fraud risk: 'Please transfer $5,000 to this account immediately. Hurry or you'll be in trouble.'"

[0931] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0932] Step 1:

[0933] (Device) records user conversations.

[0934] Input: User's spoken speech

[0935] Specific operation: The device's audio recording device records the conversation in real time in digital format.

[0936] Output: Recorded audio data (digital format)

[0937] Step 2:

[0938] (Device) sends the recorded audio data to the server.

[0939] Input: Recorded audio data

[0940] Specific operation: When recording is finished, the audio data is automatically sent to the server via the network. Communication to the server is performed using a secure protocol (e.g., HTTPS).

[0941] Output: Audio data sent to the server

[0942] Step 3:

[0943] The server starts a speech recognition engine and converts the speech data into text data.

[0944] Input: Audio data sent to the server

[0945] Specific operation: The server's speech recognition engine (for example, Google Speech-to-Text) analyzes the audio file and converts it into text data as string information.

[0946] Output: Converted text data

[0947] Step 4:

[0948] (Server) analyzes the converted text data using a natural language processing engine.

[0949] Input: Converted text data

[0950] How it works: The server's natural language processing engine (such as spaCy or BERT) performs detailed analysis of the text data to detect specific keywords and context. It extracts important keywords such as "money," "transfer," "account information," and "hurry," and analyzes the context.

[0951] Output: Analysis results (detected keywords and context information)

[0952] Step 5:

[0953] (Server) evaluates the possibility of fraud based on the analysis results.

[0954] Input: Analysis results (detected keywords and context information)

[0955] Specific behavior: Calculate a risk score based on the detected keywords and context. For example, assign a weight to each keyword and evaluate the match with the context to score the overall risk. If the risk score is high, it is determined to be high risk.

[0956] Output: Risk score, risk assessment result

[0957] Step 6:

[0958] The server generates a warning message and sends it to the user terminal.

[0959] Input: Risk assessment result (if high risk)

[0960] Specific operation: If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user's device stating, "This call may be a special fraud. Please be careful."

[0961] Output: Generated warning message

[0962] Step 7:

[0963] (Device) receives the warning message and displays it on the screen.

[0964] Input: The generated warning message

[0965] Specific operation: As soon as the user device receives a warning message, it immediately displays it on the screen to alert the user. In addition to the visual display, additional notification methods such as audio alerts can also be used.

[0966] Output: Display warning message, notify user

[0967] (Application example 1)

[0968] 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."

[0969] In modern society, the methods used by special fraudsters are becoming increasingly sophisticated, and many people are falling victim to fraud. Elderly people and users unfamiliar with the Internet are particularly susceptible to fraud. As a result, not only financial damage but also psychological damage is severe. The present invention aims to prevent such special frauds from occurring. Specifically, the objective of this invention is to provide a system that prevents fraud damage by detecting possible fraud in real time and issuing a warning to the user.

[0970] 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.

[0971] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from audio data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud, means for displaying the warning message on the user terminal, means for the user to select a response action when the warning message is displayed, and means for analyzing the text data using a generative AI model. This allows for highly accurate detection of the possibility of fraud in real time, enabling the user to take immediate response action, thereby preventing fraud damage.

[0972] A "conversation recording device" is a device or software that has the ability to record a user's telephone or face-to-face conversations in real time and store them in digital form.

[0973] "Means for converting recorded conversation content from voice data to text data" refers to a system that uses voice recognition technology to convert recorded voice files into text format.

[0974] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses a natural language processing engine to analyze important words and phrases contained in text data and their contexts.

[0975] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or functions that calculate a fraud risk score and determine the likelihood of fraud based on the analyzed text data.

[0976] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function that generates a warning message and sends it to a user terminal via a network when it is determined that there is a high risk of fraud.

[0977] The "means for displaying a warning message on a user terminal" refers to a display or application for visually displaying to the user the warning message received on the user terminal.

[0978] "Means for the user to select a response action when a warning message is displayed" refers to an interface or function that allows the user to check the warning message and then select an appropriate response action themselves.

[0979] "Means for securely transmitting voice data to a server" refers to a function that uses encryption technology and communication protocols to maintain data integrity and confidentiality when transmitting voice data from a user terminal to a server.

[0980] "Means for analyzing text data using a generative AI model" refers to a function that uses a natural language processing engine that incorporates AI technology to analyze text data with high accuracy.

[0981] As an embodiment of the present invention, a system for detecting and warning against special frauds will be described in detail. This system mainly operates using a user terminal, a server, and a communication network. Specific embodiments of this technology are described below.

[0982] User terminal

[0983] The user terminal implements the following main functions:

[0984] 1. Voice recording function

[0985] The user terminal is equipped with a microphone for recording voice, which automatically records the voice during conversation and stores it in digital format.

[0986] 2. Audio file sending function

[0987] The recorded audio data is securely transmitted to the server using encryption technology, which is essential to maintain data integrity and confidentiality.

[0988] 3. Warning display function

[0989] Displays warning messages received from the server to the user. Warning messages are displayed in a popup or notification bar so that the user can see them immediately.

[0990] server

[0991] The server performs the primary processing for analyzing the voice data and assessing the likelihood of fraud.

[0992] 1. Speech Recognition Engine

[0993] The server receives the voice data sent from the user's device and converts it into text data using a voice recognition engine, which uses the Google speech recognition API.

[0994] 2. Text Analysis Engine

[0995] A natural language processing engine using generative AI models analyzes text data, detecting specific keywords (e.g., "money," "transfer," "account") and context to assess the likelihood of fraud.

[0996] 3. Fraud Detection Algorithm

[0997] A fraud risk score is calculated based on the detected keywords and context. If the risk score exceeds a certain threshold, it is determined to be highly likely to be fraud. For example, if a conversation such as "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble" is detected, it is determined to be high risk.

[0998] 4. Warning generation function

[0999] If the risk is determined to be high, a warning message is generated and sent to the user's terminal. The warning message may include a statement such as, "This call may be a special fraud. Please be careful."

[1000] Specific examples

[1001] A specific scenario is shown below.

[1002] Example Scenario

[1003] While on the phone, the user receives the message, "Please transfer 500,000 yen to this account immediately."

[1004] The user terminal records this conversation and sends the audio file to the server.

[1005] The server converts the audio file into text data and detects specific keywords such as "account," "transfer," and "urgent."

[1006] Based on this text data, a generative AI model calculates a fraud risk score and determines it as high risk.

[1007] The server generates a warning message and sends it to the user's terminal stating, "This call may be a special fraud. Please be careful."

[1008] The user terminal displays a warning message and the user takes appropriate action.

[1009] Prompt Sentence Examples

[1010] Below are some example prompts for the generative AI model:

[1011] "Please rate the likelihood of fraud for the following text:

[1012] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[1013] With such detailed embodiments, the present invention is able to detect fraud risks in real time and issue immediate warnings to users.

[1014] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1015] Step 1: Recording audio data

[1016] The user terminal records telephone or face-to-face conversations in real time, and the recorded data is stored in digital format. The input is the user's conversational voice, and the output is the recorded voice data.

[1017] Step 2: Send your recordings

[1018] The device transmits the recorded voice data to the server. The transmission is performed using encryption technology to maintain data integrity and confidentiality. The input is the recorded voice data, and the output is the transmission of encrypted voice data.

[1019] Step 3: Convert audio data to text

[1020] The server converts the received voice data into text data using a voice recognition engine (e.g., Google voice recognition API). The input is encrypted voice data, and the output is text data.

[1021] Step 4: Analyzing the text data

[1022] The server uses a generative AI model to analyze the text data and detect specific keywords and context. Specifically, a natural language processing engine scans the text data and searches for keywords such as "money," "transfer," and "account." The input is the text data, and the output is the analyzed keywords and context information.

[1023] Step 5: Fraud risk assessment

[1024] The server calculates a fraud risk score based on the analyzed keywords and context using a pre-trained fraud detection algorithm, where the input is keywords and context information and the output is a risk score.

[1025] Step 6: Generate a warning message

[1026] Once the risk score is determined, the server automatically generates a warning message if it determines that there is a high possibility of fraud. The warning message includes content such as "This call may be a special fraud. Please be careful." The input is the risk score, and the output is the warning message.

[1027] Step 7: Sending a warning message

[1028] The server sends the generated warning message to the user terminal. The input is the warning message and the output is the transmission to the user terminal.

[1029] Step 8: Displaying warning messages

[1030] The terminal displays the received warning message to the user. A pop-up notification or an alert message is used for display so that the user can immediately check it. The input is the warning message, and the output is the warning displayed on the user's screen.

[1031] Step 9: User Actions

[1032] The user checks the warning message displayed on the terminal and selects an appropriate response action. Based on the warning, the user may end the call or perform additional confirmation. The input is the warning message, and the output is the user's response action.

[1033] 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.

[1034] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[1035] System configuration overview

[1036] User terminal

[1037] Voice Recorder: Provides the ability to record a user's conversations and store them in digital form.

[1038] Audio file sending function: Sends recorded audio data to the server.

[1039] Emotion recognition function: Analyzes the user's conversational voice in real time and recognizes emotions.

[1040] Warning display function: Receives warning messages from the server and displays them on the screen.

[1041] server

[1042] Speech recognition engine: Converts voice data into text data.

[1043] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[1044] Emotion engine: Receives user emotion data and evaluates it in combination with analytical data.

[1045] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[1046] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[1047] Program processing

[1048] Audio Acquisition

[1049] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[1050] Speech-to-text conversion

[1051] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[1052] emotion recognition

[1053] The device analyzes the user's speech in real time and recognizes emotions. Emotion recognition technology uses algorithms based on characteristics such as tone, pitch, and speed of speech.

[1054] Text analytics

[1055] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The natural language processing engine then analyzes the text data to detect specific keywords and context.

[1056] Emotional Data Integration

[1057] The server integrates emotional data into the results of text analysis. Emotional data reflects the user's psychological state, and combining this information improves the accuracy of fraud likelihood assessment.

[1058] Determining the likelihood of fraud

[1059] The server evaluates the likelihood of fraud based on the integrated data. Statistical models and machine learning algorithms are used to calculate a risk score. The inclusion of emotional data also enhances the accuracy of the risk assessment.

[1060] Warning

[1061] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action. The warning message also includes additional information tailored to the user's psychological state based on emotional data.

[1062] Displaying warnings

[1063] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[1064] Specific examples

[1065] Consider the example where the user is on a call.

[1066] Call details: "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into problems."

[1067] Processing flow

[1068] 1. (Device) records the call audio and sends it to the server.

[1069] 2. (Server) converts the audio file into text data, and a natural language processing engine detects keywords such as "account," "transfer," and "urgent" and analyzes the context.

[1070] 3. (Device) analyzes the user's emotions in real time and generates emotion data. In this case, it detects that the user is feeling tense.

[1071] 4. (Server) integrates the emotion data and evaluates the likelihood of fraud. The risk score is evaluated as 90, which is determined to be high risk.

[1072] 5. (Server) generates a warning message saying, "This call may be a fraudulent transaction. Please be careful." and sends it to the user. Based on the emotional data, it also includes additional information such as, "You seem nervous, but please remain calm."

[1073] 6. (Device) displays a warning message, and the user can respond appropriately. If necessary, the call can be terminated, preventing fraud before it occurs.

[1074] The present invention simultaneously analyzes the conversation content and user emotions to achieve more accurate fraud detection and user assistance. This system allows users to quickly recognize fraud risks and respond appropriately.

[1075] The processing flow will be explained below.

[1076] Step 1:

[1077] The device records the user's conversation. The recording is initiated by the user's operation or a voice trigger, and is saved in a digital format, such as WAV or MP3.

[1078] Step 2:

[1079] The (device) sends the recorded audio file to the server using an encrypted communication protocol (e.g., HTTPS).

[1080] Step 3:

[1081] While recording, the device analyzes the user's speech in real time and uses an emotion recognition engine to detect the user's emotional state, using algorithms based on voice tone, pitch, speed, etc.

[1082] Step 4:

[1083] The server inputs the received audio file into a speech recognition engine, which converts the audio data into text data. The speech recognition engine used is Google Speech-to-Text API or IBM Watson.

[1084] Step 5:

[1085] The server inputs the text data into a natural language processing engine, which analyzes specific keywords (e.g., "transfer," "account," "urgent") and context. Natural language processing is performed using tools such as SpaCy and NLTK.

[1086] Step 6:

[1087] The server detects the context based on the results of text analysis and assesses the likelihood of fraud based on this, using statistical models and machine learning algorithms.

[1088] Step 7:

[1089] The device recognizes the emotion data in real time and sends it to the server. This data may include the user's emotions such as tension, impatience, and fear.

[1090] Step 8:

[1091] The server integrates the results of text analysis with emotion data and reassesssses the likelihood of fraud. Including emotion data improves the accuracy of the assessment.

[1092] Step 9:

[1093] The server calculates a risk score based on the integrated data, and if it is determined to be high risk, it generates a warning message informing the user that it may be a scam and urging them to act calmly.

[1094] Step 10:

[1095] The server generates a warning message and sends it to the user's device. This message is sent using a secure communication protocol.

[1096] Step 11:

[1097] (Device) displays received warning messages to the user. The messages are displayed in a pop-up format and have an attention-grabbing design.

[1098] Step 12:

[1099] The user checks the warning message and takes appropriate action. For example, if it is determined that there is a high possibility of fraud, the user can take action such as terminating the call. The message also includes additional information (e.g., "Please remain calm") tailored to the user's psychological state based on emotional data.

[1100] In this way, the present invention provides a system that analyzes conversation content and emotions in real time, assesses the possibility of fraud with high accuracy, and quickly issues a warning to the user, thereby preventing fraud damage before it occurs.

[1101] Example 2

[1102] 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."

[1103] Special fraud remains a serious social problem, with many people falling victim to it. Conventional fraud detection systems make judgments based solely on the analysis of text data, making it difficult to accurately assess the likelihood of fraud. Furthermore, warnings are issued without taking into account the user's emotions, which can lead to users being unable to respond appropriately. Therefore, there is a need for a new system that can more accurately detect potential fraud and provide warning messages that take the user's emotions into account.

[1104] 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.

[1105] In this invention, the server includes a means for analyzing the user's conversation voice in real time and recognizing emotions, a means for evaluating the possibility of fraud based on detected keywords and context and the recognized emotion data, and a means for generating a warning message and sending it to the user terminal if the possibility of fraud is high. This allows for a more accurate warning by combining text data and emotion data to evaluate the possibility of fraud.

[1106] "Conversation recording means" refers to a method or device for digitally recording and storing a user's voice.

[1107] "Means for converting voice data into text data" refers to technology or devices that analyze recorded voice data and convert it into corresponding text data.

[1108] "Means for analyzing text data and detecting specific keywords and context" refers to technology that analyzes text data converted from audio data and identifies specific words and phrases and the context behind them.

[1109] "Means for analyzing a user's conversational voice in real time and recognizing emotions" refers to a method or device that processes a user's voice in real time and identifies emotions from characteristics of the voice such as tone, pitch, and speed.

[1110] "Means for assessing the likelihood of fraud based on detected keywords and context, and recognized sentiment data" refers to algorithms and technologies that combine analysis results with sentiment data to determine the likelihood of fraud.

[1111] "Means for generating and sending a warning message to a user terminal" refers to technology or equipment that, when a possibility of fraud is determined, creates an appropriate warning message and sends that message to a user terminal.

[1112] "Means for displaying a warning message on a user terminal" refers to a method or device for visually displaying a received warning message on a user terminal.

[1113] "Means for safely transmitting voice data and emotion data to a server" refers to technology for transmitting voice data and emotion data to a server after implementing safety measures such as encryption.

[1114] "Means for the user to select a response action based on the notified warning message and additional information" refers to technology or devices that allow the user to select the most appropriate action based on the received warning message and the provided additional information.

[1115] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[1116] System configuration overview

[1117] User terminal

[1118] Audio Recording Device: A device that has the ability to record a user's conversation and store it digitally. Examples include smartphones and dedicated recording devices.

[1119] Audio file transmission function: Recorded audio data is sent to the server using an encrypted communication protocol (e.g. HTTPS).

[1120] Emotion Recognition: Analyzes the user's speech in real time to recognize emotions. Emotion recognition technology uses algorithms based on voice characteristics such as tone, pitch, and speed.

[1121] Warning display function: Receives warning messages from the server and displays them on the screen. For example, a warning is displayed in pop-up format on a smartphone screen.

[1122] server

[1123] Speech recognition engine: Converts voice data into text data. Specifically, Google Cloud Speech-to-Text API is used.

[1124] Natural language processing engines: Analyze text data to detect specific keywords and context. Examples include OpenAI GPT-3.

[1125] Emotion Engine: Receives user emotion data, combines it with analytical data, evaluates it, and uses emotion recognition algorithms to analyze the user's mental state.

[1126] Fraud detection algorithms: Evaluate the likelihood of fraud based on the analyzed data. Statistical models and machine learning algorithms are used to calculate a risk score.

[1127] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[1128] Specific examples

[1129] Consider an example where a user is on a call. For example, if the user receives a scam call saying, "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble," the system will perform the following process.

[1130] 1. (Device) records the call audio and sends it to the server. The recording is saved in a digital format such as WAV or MP3 and transmitted securely over HTTPS.

[1131] 2. (Server) receives the audio file and converts it to text data using the Google Cloud Speech-to-Text API.

[1132] 3. (Server) analyzes the text data using a natural language processing engine such as OpenAI GPT-3 to detect keywords and context such as "account," "transfer," and "urgent."

[1133] 4. (Device) analyzes the user's emotions in real time and generates emotional data such as "tension."

[1134] 5. (Server) integrates the emotion data and evaluates the likelihood of fraud. As a result of the evaluation, the risk score is 90, which is determined to be high risk.

[1135] 6. (Server) generates a warning message saying, "This call may be a special fraud. Please be careful." and sends it to the user's terminal, along with additional information saying, "You seem nervous, but please remain calm."

[1136] 7. (Device) displays the received warning message to the user, who can then take appropriate action. For example, the user can end the call if necessary, preventing fraud before it occurs.

[1137] Prompt Sentence Examples

[1138] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[1139] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1140] Step 1: Start recording audio

[1141] (Device) records the user's conversation. When the user starts a call, the device's audio recording device automatically starts and records the conversation in digital format. The input is the user's voice, and the output is a digital audio file (WAV or MP3).

[1142] Step 2: Send the audio file

[1143] The device sends the recorded audio file to the server. The audio file is sent securely using an encrypted communication protocol such as HTTPS. The input is the recorded audio file, and the output is the audio data sent to the server.

[1144] Step 3: Convert speech to text

[1145] The server inputs the received audio file into a speech recognition engine such as the Google Cloud Speech-to-Text API. The speech recognition engine analyzes the audio data and converts it into text data. The input is the audio file, and the output is the converted text data.

[1146] Step 4: Natural Language Processing Analysis

[1147] The server inputs the text data into a natural language processing engine such as OpenAI GPT-3. The engine analyzes the text data and detects specific keywords and contexts such as "account," "transfer," and "urgent." The input is the text data, and the output is the analyzed keywords and context information.

[1148] Step 5: Start Emotion Recognition

[1149] The device analyzes the user's speech in real time and recognizes emotions. It uses emotion recognition algorithms to identify the user's emotions from features such as voice tone, pitch, and speed. The input is real-time voice data, and the output is recognized emotion data.

[1150] Step 6: Sending Emotion Data

[1151] (Device) sends the recognized emotion data to the server. The input is the emotion data, and the output is the emotion data sent to the server.

[1152] Step 7: Data synthesis and evaluation

[1153] The (server) integrates the analyzed keyword, context, and sentiment data. Based on the integrated data, it assesses the likelihood of fraud. It calculates a risk score using statistical models and machine learning algorithms. The input is keyword and context information and sentiment data, and the output is a risk score.

[1154] Step 8: Generate a warning message

[1155] If the server determines the risk score is high, it generates a warning message that includes information indicating the likelihood of fraud and additional information based on the user's sentiment. The input is the risk score and the output is the generated warning message.

[1156] Step 9: Sending a warning message

[1157] (Server) generates and sends the generated warning message to the user terminal. The input is the warning message, and the output is the warning message sent to the user terminal.

[1158] Step 10: Displaying warning messages

[1159] Displays to the user the warning message received by (the terminal). The warning message appears on the screen in a pop-up format and is designed to grab the user's attention. The input is the warning message and the output is the warning message displayed to the user.

[1160] (Application example 2)

[1161] 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."

[1162] In recent years, the methods of special fraud have become more sophisticated, and the damage caused by them has been increasing. Conventional fraud detection systems assess the possibility of fraud solely by analyzing the content of conversations, which limits their accuracy and leaves users at a high risk of falling victim to fraud. Therefore, there is a need for a more accurate fraud detection system that takes into account not only the content of conversations but also the user's emotional state.

[1163] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for recording the content of the conversation, means for converting the recorded content of the conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for analyzing the user's conversation voice in real time and recognizing emotions, means for integrating the emotion data into the text analysis results and evaluating the possibility of fraud, means for generating a warning message and sending it to the user terminal if there is a high possibility of fraud, and means for displaying the warning message on the user terminal. This makes it possible to simultaneously consider the content of the conversation and the user's emotional state and more accurately evaluate the possibility of fraud.

[1164] A "conversation recording device" is a device or system that has the ability to digitally record and store a user's calls and voice messages.

[1165] The "means for converting voice data into text data" is a voice recognition engine that has the function of analyzing recorded voice data and converting it into corresponding text data.

[1166] The "means for analyzing text data and detecting specific keywords and context" is a natural language processing engine for identifying important words and phrases within the text data and analyzing their context.

[1167] The "means for analyzing the user's conversation voice in real time and recognizing emotions" is an emotion recognition engine for analyzing emotions in real time based on the user's voice data.

[1168] The "means for integrating emotion data with text analysis results to assess the likelihood of fraud" is an algorithm for combining emotion recognition results and text analysis results to assess the likelihood of fraud.

[1169] "Means for generating a warning message and sending it to the user terminal when there is a high possibility of fraud" refers to a system that has the function of creating a warning message and notifying the user terminal when it is determined that there is a high risk of fraud.

[1170] The "means for displaying a warning message on a user terminal" refers to a device or software that has a function for visually displaying a warning message on the screen of a user terminal.

[1171] 1. Overall system configuration

[1172] This system consists of a user terminal, a server, and a communication network. The main components are a user terminal that includes a means for recording conversations and sending the audio data to the server, and the server that includes a means for analyzing the received audio data and evaluating the possibility of fraud.

[1173] 2. User Device

[1174] The user terminal has the following functions:

[1175] Voice Recorder: A device that has the ability to record user calls and voice messages. The recorded voice data is saved in a digital format (e.g., WAV or MP3).

[1176] Audio data transmission function: Recorded audio data is encrypted (e.g., AES encrypted) and then securely transmitted to the server. HTTPS is used as the communication protocol.

[1177] Emotion recognition function: Analyzes the user's speech in real time and recognizes emotions. IBM Watson Tone Analyzer and other engines are used as emotion recognition engines.

[1178] Warning display function: Displays the warning message sent from the server on the screen of the user's terminal.

[1179] 3. Server

[1180] The server has the following functions:

[1181] Speech recognition engine: An engine for converting voice data into text data, such as the Google Cloud Speech-to-Text API.

[1182] Natural language processing engines: SpaCy and NLTK are used to analyze text data and detect specific keywords and context.

[1183] Emotion recognition engine: Receives emotion data sent from the user's device and integrates it into the text analysis results. IBM Watson Tone Analyzer and other technologies are used.

[1184] Fraud detection algorithm: An algorithm that assesses the likelihood of fraud based on text analysis and sentiment data. It uses Scikit-learn and TensorFlow models.

[1185] Warning generation function: If there is a high risk of fraud, a warning message is generated and sent to the user's terminal.

[1186] 4. Data Analysis and Alert Generation

[1187] The server inputs the received voice data into a voice recognition engine to generate text data. The generated text data is then analyzed by a natural language processing engine to detect specific keywords and context. At the same time, an emotion recognition engine analyzes the user's emotional data and integrates it into the text analysis results. Finally, a fraud detection algorithm evaluates the data and calculates a risk score. If the risk is assessed as high, a warning message is generated and sent to the user's device.

[1188] 5. Specific usage examples

[1189] Suppose a user is on a call and says something like, "Please transfer 500,000 yen to this account right away. If we don't hurry, we'll be in big trouble." In this case, the following process will occur:

[1190] 1. The user device records the call audio and sends it to the server.

[1191] 2. The server converts the audio file into text data using a speech recognition engine and detects important keywords such as "transfer" and "urgent."

[1192] 3. The emotion recognition engine detects high levels of tension in the user's voice.

[1193] 4. The fraud detection algorithm integrates the emotional data and evaluates the risk score as 90, determining it to be high risk.

[1194] 5. The server generates a warning message stating, "This call may be a special fraud. Please remain calm." and sends it to the user's terminal.

[1195] 6. The user terminal displays a warning message on the screen, allowing the user to take appropriate action.

[1196] Example prompt sentence:

[1197] "If a transfer request is made during a call, analyze the data in real time and assess the risk of special fraud."

[1198] "Please also integrate the results of user sentiment analysis and generate warning messages based on the evaluation results."

[1199] In this way, the present invention simultaneously analyzes the conversation content and user emotions, achieving more accurate fraud detection and user assistance.

[1200] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1201] Step 1:

[1202] When a user initiates a call, the device records the conversation in real time. The recording is saved in a digital format (e.g., WAV or MP3). The input is the user's voice, and the output is the recorded audio file.

[1203] Step 2:

[1204] The device encrypts the recorded audio file (e.g., AES encryption) and sends it to the server. The input is the recorded audio file, and the output is the encrypted audio file.

[1205] Step 3:

[1206] The server inputs the received audio file into a speech recognition engine (e.g., Google Cloud Speech-to-Text API) and converts the audio data into text data. The input is an encrypted audio file, and the output is text data.

[1207] Step 4:

[1208] The server inputs the generated text data into a natural language processing engine (e.g., spaCy or NLTK) to detect specific keywords and contexts. The input is the text data, and the output is the detected keywords and context information.

[1209] Step 5:

[1210] The device analyzes the user's speech in real time and generates emotion data using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The input is the user's speech data, and the output is emotion data.

[1211] Step 6:

[1212] The server integrates the text analysis results and sentiment data and evaluates the likelihood of fraud using a fraud detection algorithm (e.g., Scikit-learn or TensorFlow model). The inputs are the text analysis results and sentiment data, and the output is a risk score.

[1213] Step 7:

[1214] If the risk score is high, the server uses the warning generation function to generate a warning message and send it to the user terminal. The input is the risk score, and the output is the warning message.

[1215] Step 8:

[1216] The user terminal displays the received warning message on the screen. The input is the warning message, and the output is the warning message displayed on the screen.

[1217] The above is a detailed description of the processing flow of the system program that realizes the application example.

[1218] 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.

[1219] 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.

[1220] 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.

[1221] [Fourth embodiment]

[1222] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1223] 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.

[1224] 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).

[1225] 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.

[1226] 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.

[1227] 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).

[1228] 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.

[1229] 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.

[1230] 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.

[1231] 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.

[1232] 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.

[1233] 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.

[1234] 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."

[1235] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[1236] System configuration overview

[1237] User terminal

[1238] Voice Recorder: Has the ability to record the user's conversations.

[1239] Audio file sending function: Sends recorded audio data to the server.

[1240] Warning display function: Receives warning messages from the server and displays them on the screen.

[1241] server

[1242] Speech recognition engine: Converts voice data into text data.

[1243] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[1244] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[1245] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal.

[1246] Program processing

[1247] Audio Acquisition

[1248] (Device) records conversations. When a user is having a phone or face-to-face conversation, the audio of that conversation is stored digitally.

[1249] Speech-to-text conversion

[1250] The device sends the recorded audio file to the server, where a speech recognition engine is used to convert the audio data into text data.

[1251] Text analytics

[1252] The server analyzes the converted text data using a natural language processing engine to detect specific keywords (e.g., "money," "transfer," "account information," "urgent") and context.

[1253] Determining the likelihood of fraud

[1254] The server determines the possibility of fraud based on the analysis results. If certain conditions are met, a risk score is calculated and the fraud risk level is calculated.

[1255] Warning

[1256] If the server determines that the risk score is high, it generates a warning message and sends it to the user's device, which then displays the warning on its screen to alert the user.

[1257] Specific examples

[1258] Consider the example where the user is on a call.

[1259] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[1260] Processing flow

[1261] 1. (Device) records the call audio and sends it to the server.

[1262] 2. (Server) converts the audio file into text data.

[1263] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[1264] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[1265] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[1266] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[1267] In this way, the present invention analyzes conversation content in real time and quickly detects and warns of fraud risks, helping users prevent themselves from becoming victims of fraud.

[1268] The processing flow will be explained below.

[1269] Step 1:

[1270] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[1271] Step 2:

[1272] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[1273] Step 3:

[1274] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The speech recognition engine uses advanced speech recognition technologies such as Google Speech-to-Text API and IBM Watson.

[1275] Step 4:

[1276] The server inputs the converted text data into a natural language processing engine, which analyzes the text data and detects specific keywords (e.g., "transfer," "account," "urgent"). Natural language processing technologies used include SpaCy and NLTK.

[1277] Step 5:

[1278] The server analyzes the context based on the results of keyword detection. This contextual analysis allows us to understand the context in which the keywords are used and collects data to predict the likelihood of fraud.

[1279] Step 6:

[1280] The server evaluates the likelihood of fraud based on the collected data. The evaluation is performed using statistical models and machine learning algorithms to calculate a risk score. For example, if certain conditions are met, a risk score is calculated on a scale of 0 to 100.

[1281] Step 7:

[1282] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action.

[1283] Step 8:

[1284] The server generates a warning message and sends it to the user's device. This communication also uses an encrypted protocol to ensure security.

[1285] Step 9:

[1286] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[1287] Step 10:

[1288] The user checks the warning message and takes appropriate action. If the possibility of fraud is judged to be high, the user can choose to take action such as ending the call to prevent fraud from occurring. If necessary, the system also provides a function to automatically end the call based on the user's selection.

[1289] Example 1

[1290] 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."

[1291] In recent years, there has been a demand for technology to prevent special frauds, but current methods have difficulty detecting potential fraud in real time and issuing warnings to users. In particular, as fraud methods become more sophisticated and diverse, even a momentary error in judgment can result in significant damage. This leaves users constantly exposed to the risk of fraud and makes it difficult to have conversations with peace of mind. Therefore, a system is needed that can detect potential fraud in real time and issue immediate warnings.

[1292] 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.

[1293] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to the user terminal if the possibility of fraud is high, means for displaying the warning message on the user terminal, means for transmitting the voice data to the server via a network, and means for calculating a risk score in the analysis and determining the fraud risk level, thereby enabling rapid detection and warning of fraud risks in real time.

[1294] "Means for recording conversations" refers to devices or functions that digitally record a user's conversations.

[1295] "Means for converting the recorded conversation content from voice data to text data" refers to a function that uses voice recognition technology to analyze recorded voice data and convert it into text data as character string information.

[1296] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses natural language processing technology to extract and analyze important keywords and related contexts contained in text data.

[1297] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or rule sets for assessing the risk of specialized fraud based on analyzed keywords and contextual information.

[1298] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function for automatically creating a warning message and sending it to a user terminal when it is determined that there is a high risk of fraud.

[1299] "Means for displaying a warning message on a user terminal" refers to a function for visually displaying a received warning message on a display or the like of a user terminal.

[1300] "Means for transmitting audio data to a server via a network" refers to the function of transmitting recorded audio data to a server using the Internet or other networks.

[1301] "Means for calculating a risk score in the analysis and determining the fraud risk level" refers to an algorithm or method for deriving a risk score from the results of text analysis and determining the fraud risk level based on that score.

[1302] The system of the present invention is for detecting the possibility of special fraud in real time and issuing a warning to users. This system operates by utilizing a user terminal, a server, and a communication network.

[1303] System configuration overview

[1304] User terminal

[1305] Audio Recorder: Has the ability to record user conversations. The device's audio recorder records conversations in real time in digital format.

[1306] Audio file transmission function: This function transmits recorded audio data to a server via the network. When recording is finished, the audio data is automatically sent to the server.

[1307] Warning display function: Receives warning messages from the server and displays them on the screen. Received warning messages are displayed on the display to alert the user.

[1308] server

[1309] Speech recognition engine: Converts voice data into text data. Examples of use include Google Speech-to-Text and IBM Watson Speech to Text.

[1310] Natural language processing engine: Analyzes text data and detects specific keywords and contexts. Examples of use include spaCy and BERT.

[1311] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data. Calculates a risk score based on specific keywords and context to determine the risk level.

[1312] Warning generation function: If there is a high possibility of fraud, a warning message is generated and sent to the user's terminal. A warning message is automatically generated and sent to the user's terminal.

[1313] Specific examples

[1314] Consider the example where the user is on a call.

[1315] What the call said: "Please transfer 500,000 yen to this account immediately. If we don't hurry, we'll have problems."

[1316] Processing flow

[1317] 1. (Device) records the call audio and sends it to the server.

[1318] When a user has a conversation over the phone or in person, the device's audio recorder stores the audio in digital form.

[1319] 2. (Server) converts the audio file into text data.

[1320] The server's speech recognition engine (e.g., Google Speech-to-Text) receives the audio file as input and converts it into text information.

[1321] 3. (Server) performs text analysis to detect keywords such as "account," "transfer," and "urgent," and analyzes the context.

[1322] The server's natural language processing engine (e.g., spaCy or BERT) performs the analysis and detects specific keywords and context.

[1323] 4. (Server) evaluates the likelihood of fraud based on these keywords and context and derives a risk score. A risk score of 80 or higher is considered high risk.

[1324] A risk score is calculated based on the keywords and context detected, with a score of 80 or above being considered high risk.

[1325] 5. (Server) generates a warning message and sends it to the user's device stating, "This call may be a special fraud. Please be careful."

[1326] If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user terminal.

[1327] 6. (Device) receives the warning message and displays it on the screen. The user can take appropriate action to prevent fraud.

[1328] As soon as the user terminal receives the warning message, it immediately displays it on the screen to alert the user.

[1329] Example prompts for generative AI models

[1330] "Analyze this conversation and assess its fraud risk: 'Please transfer $5,000 to this account immediately. Hurry or you'll be in trouble.'"

[1331] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1332] Step 1:

[1333] (Device) records user conversations.

[1334] Input: User's spoken speech

[1335] Specific operation: The device's audio recording device records the conversation in real time in digital format.

[1336] Output: Recorded audio data (digital format)

[1337] Step 2:

[1338] (Device) sends the recorded audio data to the server.

[1339] Input: Recorded audio data

[1340] Specific operation: When recording is finished, the audio data is automatically sent to the server via the network. Communication to the server is performed using a secure protocol (e.g., HTTPS).

[1341] Output: Audio data sent to the server

[1342] Step 3:

[1343] The server starts a speech recognition engine and converts the speech data into text data.

[1344] Input: Audio data sent to the server

[1345] Specific operation: The server's speech recognition engine (for example, Google Speech-to-Text) analyzes the audio file and converts it into text data as string information.

[1346] Output: Converted text data

[1347] Step 4:

[1348] (Server) analyzes the converted text data using a natural language processing engine.

[1349] Input: Converted text data

[1350] How it works: The server's natural language processing engine (such as spaCy or BERT) performs detailed analysis of the text data to detect specific keywords and context. It extracts important keywords such as "money," "transfer," "account information," and "hurry," and analyzes the context.

[1351] Output: Analysis results (detected keywords and context information)

[1352] Step 5:

[1353] (Server) evaluates the possibility of fraud based on the analysis results.

[1354] Input: Analysis results (detected keywords and context information)

[1355] Specific behavior: Calculate a risk score based on the detected keywords and context. For example, assign a weight to each keyword and evaluate the match with the context to score the overall risk. If the risk score is high, it is determined to be high risk.

[1356] Output: Risk score, risk assessment result

[1357] Step 6:

[1358] The server generates a warning message and sends it to the user terminal.

[1359] Input: Risk assessment result (if high risk)

[1360] Specific operation: If it is determined that there is a high risk of fraud, a warning message is automatically generated and sent to the user's device stating, "This call may be a special fraud. Please be careful."

[1361] Output: Generated warning message

[1362] Step 7:

[1363] (Device) receives the warning message and displays it on the screen.

[1364] Input: The generated warning message

[1365] Specific operation: As soon as the user device receives a warning message, it immediately displays it on the screen to alert the user. In addition to the visual display, additional notification methods such as audio alerts can also be used.

[1366] Output: Display warning message, notify user

[1367] (Application example 1)

[1368] 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."

[1369] In modern society, the methods used by special fraudsters are becoming increasingly sophisticated, and many people are falling victim to fraud. Elderly people and users unfamiliar with the Internet are particularly susceptible to fraud. As a result, not only financial damage but also psychological damage is severe. The present invention aims to prevent such special frauds from occurring. Specifically, the objective of this invention is to provide a system that prevents fraud damage by detecting possible fraud in real time and issuing a warning to the user.

[1370] 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.

[1371] In this invention, the server includes means for recording the conversation, means for converting the recorded conversation from audio data to text data, means for analyzing the text data and detecting specific keywords and context, means for evaluating the possibility of fraud based on the detected keywords and context, means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud, means for displaying the warning message on the user terminal, means for the user to select a response action when the warning message is displayed, and means for analyzing the text data using a generative AI model. This allows for highly accurate detection of the possibility of fraud in real time, enabling the user to take immediate response action, thereby preventing fraud damage.

[1372] A "conversation recording device" is a device or software that has the ability to record a user's telephone or face-to-face conversations in real time and store them in digital form.

[1373] "Means for converting recorded conversation content from voice data to text data" refers to a system that uses voice recognition technology to convert recorded voice files into text format.

[1374] "Means for analyzing text data and detecting specific keywords and contexts" refers to a function that uses a natural language processing engine to analyze important words and phrases contained in text data and their contexts.

[1375] "Means for assessing the likelihood of fraud based on detected keywords and context" refers to algorithms or functions that calculate a fraud risk score and determine the likelihood of fraud based on the analyzed text data.

[1376] "Means for generating a warning message and sending it to a user terminal when there is a high possibility of fraud" refers to a function that generates a warning message and sends it to a user terminal via a network when it is determined that there is a high risk of fraud.

[1377] The "means for displaying a warning message on a user terminal" refers to a display or application for visually displaying to the user the warning message received on the user terminal.

[1378] "Means for the user to select a response action when a warning message is displayed" refers to an interface or function that allows the user to check the warning message and then select an appropriate response action themselves.

[1379] "Means for securely transmitting voice data to a server" refers to a function that uses encryption technology and communication protocols to maintain data integrity and confidentiality when transmitting voice data from a user terminal to a server.

[1380] "Means for analyzing text data using a generative AI model" refers to a function that uses a natural language processing engine that incorporates AI technology to analyze text data with high accuracy.

[1381] As an embodiment of the present invention, a system for detecting and warning against special frauds will be described in detail. This system mainly operates using a user terminal, a server, and a communication network. Specific embodiments of this technology are described below.

[1382] User terminal

[1383] The user terminal implements the following main functions:

[1384] 1. Voice recording function

[1385] The user terminal is equipped with a microphone for recording voice, which automatically records the voice during conversation and stores it in digital format.

[1386] 2. Audio file sending function

[1387] The recorded audio data is securely transmitted to the server using encryption technology, which is essential to maintain data integrity and confidentiality.

[1388] 3. Warning display function

[1389] Displays warning messages received from the server to the user. Warning messages are displayed in a popup or notification bar so that the user can see them immediately.

[1390] server

[1391] The server performs the primary processing for analyzing the voice data and assessing the likelihood of fraud.

[1392] 1. Speech Recognition Engine

[1393] The server receives the voice data sent from the user's device and converts it into text data using a voice recognition engine, which uses the Google speech recognition API.

[1394] 2. Text Analysis Engine

[1395] A natural language processing engine using generative AI models analyzes text data, detecting specific keywords (e.g., "money," "transfer," "account") and context to assess the likelihood of fraud.

[1396] 3. Fraud Detection Algorithm

[1397] A fraud risk score is calculated based on the detected keywords and context. If the risk score exceeds a certain threshold, it is determined to be highly likely to be fraud. For example, if a conversation such as "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble" is detected, it is determined to be high risk.

[1398] 4. Warning generation function

[1399] If the risk is determined to be high, a warning message is generated and sent to the user's terminal. The warning message may include a statement such as, "This call may be a special fraud. Please be careful."

[1400] Specific examples

[1401] A specific scenario is shown below.

[1402] Example Scenario

[1403] While on the phone, the user receives the message, "Please transfer 500,000 yen to this account immediately."

[1404] The user terminal records this conversation and sends the audio file to the server.

[1405] The server converts the audio file into text data and detects specific keywords such as "account," "transfer," and "urgent."

[1406] Based on this text data, a generative AI model calculates a fraud risk score and determines it as high risk.

[1407] The server generates a warning message and sends it to the user's terminal stating, "This call may be a special fraud. Please be careful."

[1408] The user terminal displays a warning message and the user takes appropriate action.

[1409] Prompt Sentence Examples

[1410] Below are some example prompts for the generative AI model:

[1411] "Please rate the likelihood of fraud for the following text:

[1412] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[1413] With such detailed embodiments, the present invention is able to detect fraud risks in real time and issue immediate warnings to users.

[1414] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1415] Step 1: Recording audio data

[1416] The user terminal records telephone or face-to-face conversations in real time, and the recorded data is stored in digital format. The input is the user's conversational voice, and the output is the recorded voice data.

[1417] Step 2: Send your recordings

[1418] The device transmits the recorded voice data to the server. The transmission is performed using encryption technology to maintain data integrity and confidentiality. The input is the recorded voice data, and the output is the transmission of encrypted voice data.

[1419] Step 3: Convert audio data to text

[1420] The server converts the received voice data into text data using a voice recognition engine (e.g., Google voice recognition API). The input is encrypted voice data, and the output is text data.

[1421] Step 4: Analyzing the text data

[1422] The server uses a generative AI model to analyze the text data and detect specific keywords and context. Specifically, a natural language processing engine scans the text data and searches for keywords such as "money," "transfer," and "account." The input is the text data, and the output is the analyzed keywords and context information.

[1423] Step 5: Fraud risk assessment

[1424] The server calculates a fraud risk score based on the analyzed keywords and context using a pre-trained fraud detection algorithm, where the input is keywords and context information and the output is a risk score.

[1425] Step 6: Generate a warning message

[1426] Once the risk score is determined, the server automatically generates a warning message if it determines that there is a high possibility of fraud. The warning message includes content such as "This call may be a special fraud. Please be careful." The input is the risk score, and the output is the warning message.

[1427] Step 7: Sending a warning message

[1428] The server sends the generated warning message to the user terminal. The input is the warning message and the output is the transmission to the user terminal.

[1429] Step 8: Displaying warning messages

[1430] The terminal displays the received warning message to the user. A pop-up notification or an alert message is used for display so that the user can immediately check it. The input is the warning message, and the output is the warning displayed on the user's screen.

[1431] Step 9: User Actions

[1432] The user checks the warning message displayed on the terminal and selects an appropriate response action. Based on the warning, the user may end the call or perform additional confirmation. The input is the warning message, and the output is the user's response action.

[1433] 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.

[1434] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[1435] System configuration overview

[1436] User terminal

[1437] Voice Recorder: Provides the ability to record a user's conversations and store them in digital form.

[1438] Audio file sending function: Sends recorded audio data to the server.

[1439] Emotion recognition function: Analyzes the user's conversational voice in real time and recognizes emotions.

[1440] Warning display function: Receives warning messages from the server and displays them on the screen.

[1441] server

[1442] Speech recognition engine: Converts voice data into text data.

[1443] Natural language processing engine: Analyzes text data and detects specific keywords and context.

[1444] Emotion engine: Receives user emotion data and evaluates it in combination with analytical data.

[1445] Fraud detection algorithm: Evaluates the likelihood of fraud based on the analyzed data.

[1446] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[1447] Program processing

[1448] Audio Acquisition

[1449] The device records the user's conversation. The recording is initiated by the user's operation or a specific voice trigger, and is automatically saved in a digital format. The saved file format is WAV, MP3, etc.

[1450] Speech-to-text conversion

[1451] The device sends the recorded audio file to the server, using an encrypted communication protocol (e.g., HTTPS) to ensure security.

[1452] emotion recognition

[1453] The device analyzes the user's speech in real time and recognizes emotions. Emotion recognition technology uses algorithms based on characteristics such as tone, pitch, and speed of speech.

[1454] Text analytics

[1455] The server inputs the received audio file into a speech recognition engine, which analyzes the audio data and converts it into text data. The natural language processing engine then analyzes the text data to detect specific keywords and context.

[1456] Emotional Data Integration

[1457] The server integrates emotional data into the results of text analysis. Emotional data reflects the user's psychological state, and combining this information improves the accuracy of fraud likelihood assessment.

[1458] Determining the likelihood of fraud

[1459] The server evaluates the likelihood of fraud based on the integrated data. Statistical models and machine learning algorithms are used to calculate a risk score. The inclusion of emotional data also enhances the accuracy of the risk assessment.

[1460] Warning

[1461] If the server determines that the risk score is high, it generates a warning message that indicates the likelihood of fraud and prompts the user to take appropriate action. The warning message also includes additional information tailored to the user's psychological state based on emotional data.

[1462] Displaying warnings

[1463] The device displays the received warning message to the user in a pop-up format on the screen, with a visually striking design to attract the user's attention.

[1464] Specific examples

[1465] Consider the example where the user is on a call.

[1466] Call details: "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into problems."

[1467] Processing flow

[1468] 1. (Device) records the call audio and sends it to the server.

[1469] 2. (Server) converts the audio file into text data, and a natural language processing engine detects keywords such as "account," "transfer," and "urgent" and analyzes the context.

[1470] 3. (Device) analyzes the user's emotions in real time and generates emotion data. In this case, it detects that the user is feeling tense.

[1471] 4. (Server) integrates the emotion data and evaluates the likelihood of fraud. The risk score is evaluated as 90, which is determined to be high risk.

[1472] 5. (Server) generates a warning message saying, "This call may be a fraudulent transaction. Please be careful." and sends it to the user. Based on the emotional data, it also includes additional information such as, "You seem nervous, but please remain calm."

[1473] 6. (Device) displays a warning message, and the user can respond appropriately. If necessary, the call can be terminated, preventing fraud before it occurs.

[1474] The present invention simultaneously analyzes the conversation content and user emotions to achieve more accurate fraud detection and user assistance. This system allows users to quickly recognize fraud risks and respond appropriately.

[1475] The processing flow will be explained below.

[1476] Step 1:

[1477] The device records the user's conversation. The recording is initiated by the user's operation or a voice trigger, and is saved in a digital format, such as WAV or MP3.

[1478] Step 2:

[1479] The (device) sends the recorded audio file to the server using an encrypted communication protocol (e.g., HTTPS).

[1480] Step 3:

[1481] While recording, the device analyzes the user's speech in real time and uses an emotion recognition engine to detect the user's emotional state, using algorithms based on voice tone, pitch, speed, etc.

[1482] Step 4:

[1483] The server inputs the received audio file into a speech recognition engine, which converts the audio data into text data. The speech recognition engine used is Google Speech-to-Text API or IBM Watson.

[1484] Step 5:

[1485] The server inputs the text data into a natural language processing engine, which analyzes specific keywords (e.g., "transfer," "account," "urgent") and context. Natural language processing is performed using tools such as SpaCy and NLTK.

[1486] Step 6:

[1487] The server detects the context based on the results of text analysis and assesses the likelihood of fraud based on this, using statistical models and machine learning algorithms.

[1488] Step 7:

[1489] The device recognizes the emotion data in real time and sends it to the server. This data may include the user's emotions such as tension, impatience, and fear.

[1490] Step 8:

[1491] The server integrates the results of text analysis with emotion data and reassesssses the likelihood of fraud. Including emotion data improves the accuracy of the assessment.

[1492] Step 9:

[1493] The server calculates a risk score based on the integrated data, and if it is determined to be high risk, it generates a warning message informing the user that it may be a scam and urging them to act calmly.

[1494] Step 10:

[1495] The server generates a warning message and sends it to the user's device. This message is sent using a secure communication protocol.

[1496] Step 11:

[1497] (Device) displays received warning messages to the user. The messages are displayed in a pop-up format and have an attention-grabbing design.

[1498] Step 12:

[1499] The user checks the warning message and takes appropriate action. For example, if it is determined that there is a high possibility of fraud, the user can take action such as terminating the call. The message also includes additional information (e.g., "Please remain calm") tailored to the user's psychological state based on emotional data.

[1500] In this way, the present invention provides a system that analyzes conversation content and emotions in real time, assesses the possibility of fraud with high accuracy, and quickly issues a warning to the user, thereby preventing fraud damage before it occurs.

[1501] Example 2

[1502] 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."

[1503] Special fraud remains a serious social problem, with many people falling victim to it. Conventional fraud detection systems make judgments based solely on the analysis of text data, making it difficult to accurately assess the likelihood of fraud. Furthermore, warnings are issued without taking into account the user's emotions, which can lead to users being unable to respond appropriately. Therefore, there is a need for a new system that can more accurately detect potential fraud and provide warning messages that take the user's emotions into account.

[1504] 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.

[1505] In this invention, the server includes a means for analyzing the user's conversation voice in real time and recognizing emotions, a means for evaluating the possibility of fraud based on detected keywords and context and the recognized emotion data, and a means for generating a warning message and sending it to the user terminal if the possibility of fraud is high. This allows for a more accurate warning by combining text data and emotion data to evaluate the possibility of fraud.

[1506] "Conversation recording means" refers to a method or device for digitally recording and storing a user's voice.

[1507] "Means for converting voice data into text data" refers to technology or devices that analyze recorded voice data and convert it into corresponding text data.

[1508] "Means for analyzing text data and detecting specific keywords and context" refers to technology that analyzes text data converted from audio data and identifies specific words and phrases and the context behind them.

[1509] "Means for analyzing a user's conversational voice in real time and recognizing emotions" refers to a method or device that processes a user's voice in real time and identifies emotions from characteristics of the voice such as tone, pitch, and speed.

[1510] "Means for assessing the likelihood of fraud based on detected keywords and context, and recognized sentiment data" refers to algorithms and technologies that combine analysis results with sentiment data to determine the likelihood of fraud.

[1511] "Means for generating and sending a warning message to a user terminal" refers to technology or equipment that, when a possibility of fraud is determined, creates an appropriate warning message and sends that message to a user terminal.

[1512] "Means for displaying a warning message on a user terminal" refers to a method or device for visually displaying a received warning message on a user terminal.

[1513] "Means for safely transmitting voice data and emotion data to a server" refers to technology for transmitting voice data and emotion data to a server after implementing safety measures such as encryption.

[1514] "Means for the user to select a response action based on the notified warning message and additional information" refers to technology or devices that allow the user to select the most appropriate action based on the received warning message and the provided additional information.

[1515] The system of the present invention detects the possibility of special fraud in real time and improves the accuracy of warnings by recognizing the user's emotions and incorporating them into the analysis. The system operates using a user terminal, a server, an emotion engine, and a communication network.

[1516] System configuration overview

[1517] User terminal

[1518] Audio Recording Device: A device that has the ability to record a user's conversation and store it digitally. Examples include smartphones and dedicated recording devices.

[1519] Audio file transmission function: Recorded audio data is sent to the server using an encrypted communication protocol (e.g. HTTPS).

[1520] Emotion Recognition: Analyzes the user's speech in real time to recognize emotions. Emotion recognition technology uses algorithms based on voice characteristics such as tone, pitch, and speed.

[1521] Warning display function: Receives warning messages from the server and displays them on the screen. For example, a warning is displayed in pop-up format on a smartphone screen.

[1522] server

[1523] Speech recognition engine: Converts voice data into text data. Specifically, Google Cloud Speech-to-Text API is used.

[1524] Natural language processing engines: Analyze text data to detect specific keywords and context. Examples include OpenAI GPT-3.

[1525] Emotion Engine: Receives user emotion data, combines it with analytical data, evaluates it, and uses emotion recognition algorithms to analyze the user's mental state.

[1526] Fraud detection algorithms: Evaluate the likelihood of fraud based on the analyzed data. Statistical models and machine learning algorithms are used to calculate a risk score.

[1527] Warning generation function: If there is a high possibility of fraud, a warning message containing additional information is generated and sent to the user's device.

[1528] Specific examples

[1529] Consider an example where a user is on a call. For example, if the user receives a scam call saying, "Please transfer 500,000 yen to this account immediately. If you don't hurry, you will be in trouble," the system will perform the following process.

[1530] 1. (Device) records the call audio and sends it to the server. The recording is saved in a digital format such as WAV or MP3 and transmitted securely over HTTPS.

[1531] 2. (Server) receives the audio file and converts it to text data using the Google Cloud Speech-to-Text API.

[1532] 3. (Server) analyzes the text data using a natural language processing engine such as OpenAI GPT-3 to detect keywords and context such as "account," "transfer," and "urgent."

[1533] 4. (Device) analyzes the user's emotions in real time and generates emotional data such as "tension."

[1534] 5. (Server) integrates the emotion data and evaluates the likelihood of fraud. As a result of the evaluation, the risk score is 90, which is determined to be high risk.

[1535] 6. (Server) generates a warning message saying, "This call may be a special fraud. Please be careful." and sends it to the user's terminal, along with additional information saying, "You seem nervous, but please remain calm."

[1536] 7. (Device) displays the received warning message to the user, who can then take appropriate action. For example, the user can end the call if necessary, preventing fraud before it occurs.

[1537] Prompt Sentence Examples

[1538] "Please transfer 500,000 yen to this account immediately. If you don't hurry, you'll run into trouble."

[1539] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1540] Step 1: Start recording audio

[1541] (Device) records the user's conversation. When the user starts a call, the device's audio recording device automatically starts and records the conversation in digital format. The input is the user's voice, and the output is a digital audio file (WAV or MP3).

[1542] Step 2: Send the audio file

[1543] The device sends the recorded audio file to the server. The audio file is sent securely using an encrypted communication protocol such as HTTPS. The input is the recorded audio file, and the output is the audio data sent to the server.

[1544] Step 3: Convert speech to text

[1545] The server inputs the received audio file into a speech recognition engine such as the Google Cloud Speech-to-Text API. The speech recognition engine analyzes the audio data and converts it into text data. The input is the audio file, and the output is the converted text data.

[1546] Step 4: Natural Language Processing Analysis

[1547] The server inputs the text data into a natural language processing engine such as OpenAI GPT-3. The engine analyzes the text data and detects specific keywords and contexts such as "account," "transfer," and "urgent." The input is the text data, and the output is the analyzed keywords and context information.

[1548] Step 5: Start Emotion Recognition

[1549] The device analyzes the user's speech in real time and recognizes emotions. It uses emotion recognition algorithms to identify the user's emotions from features such as voice tone, pitch, and speed. The input is real-time voice data, and the output is recognized emotion data.

[1550] Step 6: Sending Emotion Data

[1551] (Device) sends the recognized emotion data to the server. The input is the emotion data, and the output is the emotion data sent to the server.

[1552] Step 7: Data synthesis and evaluation

[1553] The (server) integrates the analyzed keyword, context, and sentiment data. Based on the integrated data, it assesses the likelihood of fraud. It calculates a risk score using statistical models and machine learning algorithms. The input is keyword and context information and sentiment data, and the output is a risk score.

[1554] Step 8: Generate a warning message

[1555] If the server determines the risk score is high, it generates a warning message that includes information indicating the likelihood of fraud and additional information based on the user's sentiment. The input is the risk score and the output is the generated warning message.

[1556] Step 9: Sending a warning message

[1557] (Server) generates and sends the generated warning message to the user terminal. The input is the warning message, and the output is the warning message sent to the user terminal.

[1558] Step 10: Displaying warning messages

[1559] Displays to the user the warning message received by (the terminal). The warning message appears on the screen in a pop-up format and is designed to grab the user's attention. The input is the warning message and the output is the warning message displayed to the user.

[1560] (Application example 2)

[1561] 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."

[1562] In recent years, the methods of special fraud have become more sophisticated, and the damage caused by them has been increasing. Conventional fraud detection systems assess the possibility of fraud solely by analyzing the content of conversations, which limits their accuracy and leaves users at a high risk of falling victim to fraud. Therefore, there is a need for a more accurate fraud detection system that takes into account not only the content of conversations but also the user's emotional state.

[1563] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for recording the content of the conversation, means for converting the recorded content of the conversation from voice data to text data, means for analyzing the text data and detecting specific keywords and context, means for analyzing the user's conversation voice in real time and recognizing emotions, means for integrating the emotion data into the text analysis results and evaluating the possibility of fraud, means for generating a warning message and sending it to the user terminal if there is a high possibility of fraud, and means for displaying the warning message on the user terminal. This makes it possible to simultaneously consider the content of the conversation and the user's emotional state and more accurately evaluate the possibility of fraud.

[1564] A "conversation recording device" is a device or system that has the ability to digitally record and store a user's calls and voice messages.

[1565] The "means for converting voice data into text data" is a voice recognition engine that has the function of analyzing recorded voice data and converting it into corresponding text data.

[1566] The "means for analyzing text data and detecting specific keywords and context" is a natural language processing engine for identifying important words and phrases within the text data and analyzing their context.

[1567] The "means for analyzing the user's conversation voice in real time and recognizing emotions" is an emotion recognition engine for analyzing emotions in real time based on the user's voice data.

[1568] The "means for integrating emotion data with text analysis results to assess the likelihood of fraud" is an algorithm for combining emotion recognition results and text analysis results to assess the likelihood of fraud.

[1569] "Means for generating a warning message and sending it to the user terminal when there is a high possibility of fraud" refers to a system that has the function of creating a warning message and notifying the user terminal when it is determined that there is a high risk of fraud.

[1570] The "means for displaying a warning message on a user terminal" refers to a device or software that has a function for visually displaying a warning message on the screen of a user terminal.

[1571] 1. Overall system configuration

[1572] This system consists of a user terminal, a server, and a communication network. The main components are a user terminal that includes a means for recording conversations and sending the audio data to the server, and the server that includes a means for analyzing the received audio data and evaluating the possibility of fraud.

[1573] 2. User Device

[1574] The user terminal has the following functions:

[1575] Voice Recorder: A device that has the ability to record user calls and voice messages. The recorded voice data is saved in a digital format (e.g., WAV or MP3).

[1576] Audio data transmission function: Recorded audio data is encrypted (e.g., AES encrypted) and then securely transmitted to the server. HTTPS is used as the communication protocol.

[1577] Emotion recognition function: Analyzes the user's speech in real time and recognizes emotions. IBM Watson Tone Analyzer and other engines are used as emotion recognition engines.

[1578] Warning display function: Displays the warning message sent from the server on the screen of the user's terminal.

[1579] 3. Server

[1580] The server has the following functions:

[1581] Speech recognition engine: An engine for converting voice data into text data, such as the Google Cloud Speech-to-Text API.

[1582] Natural language processing engines: SpaCy and NLTK are used to analyze text data and detect specific keywords and context.

[1583] Emotion recognition engine: Receives emotion data sent from the user's device and integrates it into the text analysis results. IBM Watson Tone Analyzer and other technologies are used.

[1584] Fraud detection algorithm: An algorithm that assesses the likelihood of fraud based on text analysis and sentiment data. It uses Scikit-learn and TensorFlow models.

[1585] Warning generation function: If there is a high risk of fraud, a warning message is generated and sent to the user's terminal.

[1586] 4. Data Analysis and Alert Generation

[1587] The server inputs the received voice data into a voice recognition engine to generate text data. The generated text data is then analyzed by a natural language processing engine to detect specific keywords and context. At the same time, an emotion recognition engine analyzes the user's emotional data and integrates it into the text analysis results. Finally, a fraud detection algorithm evaluates the data and calculates a risk score. If the risk is assessed as high, a warning message is generated and sent to the user's device.

[1588] 5. Specific usage examples

[1589] Suppose a user is on a call and says something like, "Please transfer 500,000 yen to this account right away. If we don't hurry, we'll be in big trouble." In this case, the following process will occur:

[1590] 1. The user device records the call audio and sends it to the server.

[1591] 2. The server converts the audio file into text data using a speech recognition engine and detects important keywords such as "transfer" and "urgent."

[1592] 3. The emotion recognition engine detects high levels of tension in the user's voice.

[1593] 4. The fraud detection algorithm integrates the emotional data and evaluates the risk score as 90, determining it to be high risk.

[1594] 5. The server generates a warning message stating, "This call may be a special fraud. Please remain calm." and sends it to the user's terminal.

[1595] 6. The user terminal displays a warning message on the screen, allowing the user to take appropriate action.

[1596] Example prompt sentence:

[1597] "If a transfer request is made during a call, analyze the data in real time and assess the risk of special fraud."

[1598] "Please also integrate the results of user sentiment analysis and generate warning messages based on the evaluation results."

[1599] In this way, the present invention simultaneously analyzes the conversation content and user emotions, achieving more accurate fraud detection and user assistance.

[1600] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1601] Step 1:

[1602] When a user initiates a call, the device records the conversation in real time. The recording is saved in a digital format (e.g., WAV or MP3). The input is the user's voice, and the output is the recorded audio file.

[1603] Step 2:

[1604] The device encrypts the recorded audio file (e.g., AES encryption) and sends it to the server. The input is the recorded audio file, and the output is the encrypted audio file.

[1605] Step 3:

[1606] The server inputs the received audio file into a speech recognition engine (e.g., Google Cloud Speech-to-Text API) and converts the audio data into text data. The input is an encrypted audio file, and the output is text data.

[1607] Step 4:

[1608] The server inputs the generated text data into a natural language processing engine (e.g., spaCy or NLTK) to detect specific keywords and contexts. The input is the text data, and the output is the detected keywords and context information.

[1609] Step 5:

[1610] The device analyzes the user's speech in real time and generates emotion data using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The input is the user's speech data, and the output is emotion data.

[1611] Step 6:

[1612] The server integrates the text analysis results and sentiment data and evaluates the likelihood of fraud using a fraud detection algorithm (e.g., Scikit-learn or TensorFlow model). The inputs are the text analysis results and sentiment data, and the output is a risk score.

[1613] Step 7:

[1614] If the risk score is high, the server uses the warning generation function to generate a warning message and send it to the user terminal. The input is the risk score, and the output is the warning message.

[1615] Step 8:

[1616] The user terminal displays the received warning message on the screen. The input is the warning message, and the output is the warning message displayed on the screen.

[1617] The above is a detailed description of the processing flow of the system program that realizes the application example.

[1618] 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.

[1619] 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.

[1620] 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.

[1621] 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.

[1622] 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.

[1623] 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.

[1624] 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).

[1625] 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.

[1626] 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."

[1627] 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.

[1628] 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).

[1629] 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.

[1630] 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.

[1631] 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.

[1632] 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.

[1633] 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.

[1634] 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.

[1635] 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.

[1636] 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.

[1637] 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.

[1638] 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.

[1639] The following is further disclosed regarding the above embodiment.

[1640] (Claim 1)

[1641] A means for recording the conversation;

[1642] A means for converting the recorded conversation content from audio data to text data;

[1643] means for analyzing the text data to detect specific keywords and contexts;

[1644] a means for assessing the likelihood of fraud based on the detected keywords and context;

[1645] means for generating and transmitting a warning message to a user terminal when there is a high possibility of fraud;

[1646] means for displaying a warning message on a user terminal;

[1647] A system including:

[1648] (Claim 2)

[1649] 10. The system of claim 1, further comprising means for securely transmitting the audio data to the server.

[1650] (Claim 3)

[1651] 2. The system according to claim 1, further comprising means for allowing a user to select a corresponding action based on the notified warning message.

[1652] "Example 1"

[1653] (Claim 1)

[1654] A means for recording the conversation;

[1655] A means for converting the recorded conversation content from audio data to text data;

[1656] means for analyzing the text data to detect specific keywords and contexts;

[1657] a means for assessing the likelihood of fraud based on the detected keywords and context;

[1658] means for generating and transmitting a warning message to a user terminal when there is a high possibility of fraud;

[1659] means for displaying a warning message on a user terminal;

[1660] means for transmitting the audio data to a server via a network;

[1661] means for calculating a risk score in the analysis to determine a fraud risk level;

[1662] A system including:

[1663] (Claim 2)

[1664] 10. The system of claim 1, further comprising means for securely transmitting the audio data to the server.

[1665] (Claim 3)

[1666] 2. The system according to claim 1, further comprising means for allowing a user to select a corresponding action based on the notified warning message.

[1667] "Application Example 1"

[1668] (Claim 1)

[1669] A means for recording the conversation;

[1670] A means for converting the recorded conversation content from audio data to text data;

[1671] means for analyzing the text data to detect specific keywords and contexts;

[1672] a means for assessing the likelihood of fraud based on the detected keywords and context;

[1673] means for generating and transmitting a warning message to a user terminal when there is a high possibility of fraud;

[1674] means for displaying a warning message on a user terminal;

[1675] a means for a user to select a corresponding action when a warning message is displayed;

[1676] A system including:

[1677] (Claim 2)

[1678] 10. The system of claim 1, further comprising means for securely transmitting the audio data to the server.

[1679] (Claim 3)

[1680] 10. The system of claim 1, further comprising means for analyzing text data using a generative AI model.

[1681] "Example 2: Combining Emotion Engines"

[1682] (Claim 1)

[1683] A means for recording the conversation;

[1684] A means for converting the recorded conversation content from audio data to text data;

[1685] means for analyzing the text data to detect specific keywords and contexts;

[1686] A means for analyzing a user's conversation voice in real time and recognizing emotions;

[1687] a means for assessing the likelihood of fraud based on the detected keywords and context and the recognized sentiment data;

[1688] means for generating and transmitting a warning message to a user terminal when there is a high possibility of fraud;

[1689] means for displaying a warning message on a user terminal;

[1690] A system including:

[1691] (Claim 2)

[1692] 10. The system of claim 1, further comprising means for securely transmitting the voice data and emotion data to a server.

[1693] (Claim 3)

[1694] 2. The system according to claim 1, further comprising means for allowing a user to select a corresponding action based on the notified warning message and additional information.

[1695] "Application example 2 when combining emotion engines"

[1696] (Claim 1)

[1697] A means for recording the conversation;

[1698] A means for converting the recorded conversation content from audio data to text data;

[1699] means for analyzing the text data to detect specific keywords and contexts;

[1700] A means for analyzing a user's conversation voice in real time and recognizing emotions;

[1701] A means of integrating sentiment data with text analysis results to assess the likelihood of fraud; and

[1702] means for generating and transmitting a warning message to a user terminal when there is a high possibility of fraud;

[1703] means for displaying a warning message on a user terminal;

[1704] A system including:

[1705] (Claim 2)

[1706] 10. The system of claim 1, further comprising means for securely transmitting the audio data to the server.

[1707] (Claim 3)

[1708] 2. The system according to claim 1, further comprising means for allowing a user to select a corresponding action based on the notified warning message. [Explanation of symbols]

[1709] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for recording the conversation; A means for converting the recorded conversation content from audio data to text data; means for analyzing the text data to detect specific keywords and contexts; a means for assessing the likelihood of fraud based on the detected keywords and context; means for generating and transmitting a warning message to a user terminal when there is a high possibility of fraud; means for displaying a warning message on a user terminal; A system including:

2. 10. The system of claim 1, further comprising means for securely transmitting the audio data to the server.

3. 2. The system according to claim 1, further comprising means for allowing a user to select a corresponding action based on the notified warning message.

Images