System

Abstract

A system is provided.SOLUTION: A system comprising: means for collecting audio of a call when a user initiates the call; means for converting the collected audio into textual data in real-time; means for sending the converted textual data to a server; means for inputting the textual data received by the server into a generated AI model to assess a likelihood of fraud; and means for sending a notification if the likelihood of fraud is determined to be high.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] Conventional countermeasures have not been effective enough in preventing frauds that primarily target the elderly, particularly "it's me" frauds. Fraudsters use sophisticated methods, and even though awareness of the scams has increased, the number of victims has not decreased. For this reason, there is a need for a system that can prevent fraud by analyzing call content in real time and quickly detecting signs of fraud. This invention aims to solve these problems by more quickly and accurately determining the possibility of fraud and issuing a warning to the subscriber or the designated notification address. [Means for solving the problem]

[0005] In order to solve the above problems, we propose a system that provides the following means.

[0006] When a user initiates a call, the system includes a means for collecting voice data of the call. It also includes a means for converting the collected voice data into text data in real time. It also includes a means for transmitting the converted text data to a server, and a means for the server to input the received text data into a generative AI model and evaluate the likelihood of fraud. By providing a means for sending a notification if it is determined that there is a high likelihood of fraud, it is possible to prevent fraud, particularly targeting the elderly, from occurring.

[0007] "Call voice" refers to voice data emitted by a user via a telephone or voice call application.

[0008] "Voice data" refers to data obtained by converting acoustic signals collected by a microphone during a call into digital form.

[0009] "Collection means" refers to a device or software mechanism for acquiring information such as voice data.

[0010] "Text data" is data that has been converted from voice data into character information, and is expressed as a string of characters.

[0011] A "conversion means" is a device or software mechanism for converting data of one format into data of another format.

[0012] A "server" is a computer system that provides services to other computers over a network.

[0013] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to analyze and generate data.

[0014] "Fraudulence" is the degree to which a call is likely to be fraudulent based on the content of the call.

[0015] "Evaluation means" refers to the equipment or software mechanisms used to analyze data or information and evaluate its content.

[0016] "Notification means" means a device or software mechanism for communicating analysis results or warning information to the user or other designated persons.

[0017] A "secure communication protocol" is a communication method that ensures security when sending and receiving data. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0019] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

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

[0021] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0022] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0023] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0024] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0025] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0026] [First embodiment]

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

[0028] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0029] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0030] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0031] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0032] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0033] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

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

[0035] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0036] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0037] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0038] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0039] The present invention relates to a system that accurately detects potential fraud while a user is on a call and prevents damage before it occurs. The system includes a user terminal, a server, a generative AI model, and a notification means.

[0040] 1. Audio data collection:

[0041] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[0042] 2. Audio to text conversion:

[0043] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[0044] 3. Sending text data:

[0045] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[0046] 4. Performing fraud analysis:

[0047] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[0048] 5. Sending notifications:

[0049] If the server determines that a call is likely to be fraudulent based on the analysis results of the generative AI model, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be fraudulent.

[0050] Examples:

[0051] For example, suppose elderly person A receives a call from someone claiming to be his "son" who says, "I lost my wallet and need money." When the call begins, A's device collects the audio and converts it into text data. This text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud and determines that there is a high possibility of fraud. Based on the results, the server sends a warning notification to A and his family. Upon receiving the notification, A's family can immediately contact A and provide support to prevent the victim from falling victim to fraud.

[0052] In this way, the system of the present invention can detect signs of fraud in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[0053] The processing flow will be explained below.

[0054] Step 1:

[0055] When a user initiates a call using a phone or voice call app, the device automatically begins collecting the call's audio data. The collection occurs in real time, and the user does not need to take any special action.

[0056] Step 2:

[0057] The device inputs the collected voice data into a speech recognition engine in real time. This speech recognition engine converts the voice data into text data using an external service, such as the Google Speech-to-Text API. The converted text data is temporarily stored in the device's memory.

[0058] Step 3:

[0059] The device checks the converted text data for errors, and if any imperfections that affect recognition accuracy are found, the device runs the speech recognition again to improve the accuracy of the text data.

[0060] Step 4:

[0061] The terminal encrypts the text data. This encryption process is necessary to protect the user's privacy, and uses a commonly used encryption algorithm (e.g., AES).

[0062] Step 5:

[0063] The encrypted text data is sent to the server using a secure communications protocol (e.g., HTTPS), with appropriate procedures taken to ensure data integrity and privacy.

[0064] Step 6:

[0065] The server adds the received text data to a queue for analysis, which allows the server to process large amounts of data efficiently.

[0066] Step 7:

[0067] The server takes the text data and inputs it into a generative AI model, which has been pre-trained on a large amount of past fraud case data and is able to assess the likelihood of fraud with a high degree of accuracy.

[0068] Step 8:

[0069] The generative AI model analyzes input text data to detect signs of fraud. The model identifies keywords and context specific to fraud and outputs the results as a numerical score.

[0070] Step 9:

[0071] The server determines whether the likelihood of fraud is high or low based on the numerical score output by the generative AI model. A specific threshold is set, and if the threshold is exceeded, it is determined that there is a "high likelihood of fraud."

[0072] Step 10:

[0073] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined notification recipients via email, SMS, or push notification on a dedicated app.

[0074] Step 11:

[0075] The user or the person who received the notification will check the contents of the notification. Based on the notification, the user can take measures such as double-checking with the other party or reporting the call to the police, thereby preventing fraud from occurring.

[0076] Example 1

[0077] 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."

[0078] In recent years, fraud methods targeting the elderly have become increasingly sophisticated, with the number of fraudulent incidents occurring during phone calls increasing. Therefore, there is a need for a system that allows users to identify potential fraud in real time and prevent it before it occurs. However, existing systems have difficulty in detecting fraud with high accuracy and providing prompt notification, which leaves elderly people and their families unable to take effective countermeasures.

[0079] 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.

[0080] In this invention, the server includes means for collecting voice data, means for converting the voice data into text data in real time, means for transmitting the converted text data to the server, means for adding the text data received by the server to an analysis queue and evaluating the possibility of fraud using a generative AI model, and means for sending a notification when it is determined that there is a high possibility of fraud. This allows users to accurately detect signs of fraud during a call and receive a prompt warning.

[0081] A "means" is a component of an apparatus, method, or system for achieving a particular function.

[0082] "Voice data" refers to data that is a digital recording of a user's voice during a call.

[0083] "Real-time" refers to processing that occurs almost simultaneously with little delay.

[0084] "Text data" is character information converted from voice data.

[0085] A "server" is a computer system that processes and stores data over a network.

[0086] An "analysis queue" is a data structure for sequentially managing data and tasks waiting to be processed.

[0087] A "generative AI model" is a machine learning model that is trained to perform a specific task using artificial intelligence techniques.

[0088] "Probability of fraud" refers to the probability or risk that the content of a call is fraudulent.

[0089] "Evaluation" refers to making a judgment based on data and information according to specific indicators and criteria.

[0090] A "notification" is a message or signal that notifies a user or a third party of specific information.

[0091] The present invention relates to a system that detects and prevents fraudulent acts during phone calls with high accuracy and in real time. This system is composed of a user terminal, a server, etc. Each component and its operation will be described in detail below.

[0092] 1. Collection of audio data

[0093] When a user initiates a call, the device automatically captures the call's audio data using a high-definition microphone, which records the audio in digital form.

[0094] 2. Converting voice data to text

[0095] The collected voice data is converted into text data in real time by the device's voice recognition engine. The voice recognition engine uses common voice recognition services such as the Google Speech-to-Text API. Error checking is performed as needed to improve conversion accuracy.

[0096] 3. Sending text data

[0097] The converted text data is encrypted to protect the user's privacy and sent to the server using a secure communication protocol (e.g., HTTPS). This communication is done in real time and security measures are in place to prevent data leakage.

[0098] 4. Performing fraud analysis

[0099] The server adds the received text data to an analysis queue. The text data added to the analysis queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and detects specific keywords and contextual information to assess the likelihood of fraud. The analysis results are output as a numerical score, with a higher score indicating a higher likelihood of fraud.

[0100] 5. Sending notifications

[0101] If the call is assessed as likely to be fraudulent, the server will send a notification to the subscriber and pre-defined recipients via email, SMS, or push notification via a dedicated app. The notification will include a summary of the call and a warning that "there is a possibility of fraud."

[0102] Specific examples

[0103] For example, suppose an elderly user receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." In this case, when the call begins, the user's device automatically collects voice data and converts it into text data. This text data is then encrypted and sent to a server. The server then analyzes the call using a generative AI model, detects phrases such as "I lost my wallet" and "I need money," and determines that the call is likely a scam. Based on the results, the server then sends a warning notification to the user and their family. The family member who receives the notification can quickly contact the user and provide support to prevent fraud.

[0104] In this way, the system of the present invention can detect fraudulent activity in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[0105] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0106] Step 1:

[0107] When a user initiates a call, the terminal automatically collects the voice data of the call.

[0108] Input: User's voice during a call

[0109] Data processing: Capture audio as digital data using a high-precision microphone

[0110] Output: Digital audio data

[0111] Step 2:

[0112] The collected voice data is converted into text data in real time using the device's voice recognition engine.

[0113] Input: Digital audio data

[0114] Data calculation: Converts voice into text using a voice recognition engine (e.g., Google Speech-to-Text API).

[0115] Output: Text data

[0116] Specific operation: Performs error check and reprocesses if misrecognition is found.

[0117] Step 3:

[0118] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS).

[0119] Input: Text data

[0120] Data processing: Encrypting data using encryption algorithms

[0121] Output: Encrypted text data

[0122] What it does: Sends encrypted data to the server via HTTPS protocol

[0123] Step 4:

[0124] The server adds the received text data to an analysis queue and evaluates the likelihood of fraud using a generative AI model.

[0125] Input: Encrypted text data

[0126] Data processing: Decrypting and adding text data to the analysis queue

[0127] Data calculation: Analyzes data using a generative AI model and calculates a fraud probability score.

[0128] Output: A numerical score indicating the likelihood of fraud.

[0129] What it does: Detects specific keywords and contextual information and makes assessments based on historical fraud data.

[0130] Step 5:

[0131] If the possibility of fraud is assessed to be high, the server will send a notification to the subscriber and to a pre-defined notification address.

[0132] Input: Fraud likelihood score

[0133] Data processing: Generate notification messages and include necessary information

[0134] Output: Information message

[0135] Specific operation: A warning message is sent via email, SMS, or push notification of a dedicated app. The notification includes a summary of the call and a warning message that the call may be fraudulent.

[0136] (Application example 1)

[0137] 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."

[0138] In recent years, there has been an increase in telephone fraud, with elderly people being particularly vulnerable to these attacks. Conventional countermeasures have been limited in their effectiveness in preventing fraud, as they have struggled to detect fraud in real time and issue prompt warnings. Therefore, there is a need for a system that allows users to instantly detect signs of fraud while on the phone and prevent fraud before it occurs.

[0139] 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.

[0140] In this invention, the server includes means for collecting voice data of the call when a user starts a call, means for converting the collected voice data into text data in real time, means for transmitting the converted text data to the server, means for the server to input the received text data into a generative AI model and evaluate the possibility of fraud, means for sending a notification when it is determined that there is a high possibility of fraud, and means for detecting signs of fraud in real time using an application installed on the smartphone and sending a warning, thereby enabling the possibility of fraud to be detected immediately during a call and enabling a prompt response.

[0141] "User" refers to any entity that includes an individual who makes a call using this system.

[0142] "Call voice data" refers to data that is a digital recording of the voice uttered by a user during a call.

[0143] A "collection means" is a combination of hardware and software for capturing audio data during a user's call.

[0144] "Means for converting voice data into text data in real time" refers to the speech recognition engine and associated software used to convert collected voice data into text data on the fly.

[0145] The "transmitting means" is an internet communication means for encrypting the converted text data and transmitting it to a server using a secure communication protocol.

[0146] A "generative AI model" is an artificial intelligence model trained based on a huge amount of past fraud data, and analyzes text data to assess the likelihood of fraud.

[0147] The "assessment means" is a combination of software in which a generative AI model analyzes input text data and outputs a numerical score indicating the likelihood of fraud.

[0148] The "means for sending a notification" is a communication means for sending a warning to the user and a pre-defined notification destination when it is determined that there is a high possibility of fraud.

[0149] "Applications installed on smartphones" are dedicated programs that run on smartphones to execute the functions of this system.

[0150] "Warning method" is a communication method such as email, SMS, or push notification used to notify the user of potential fraud.

[0151] The present invention provides a system for detecting potential fraud with high accuracy while a user is making a call and preventing damage before it occurs. This system includes a user terminal, a server, a generative AI model, and a notification means. Specific embodiments of the system are described below.

[0152] First, when a user starts a call, the device automatically collects voice data. The smartphone's microphone and voice recognition engine (e.g., Google Speech-to-Text API) are used to collect the voice data. The voice recognition engine converts the collected voice data into text data in real time. Appropriate error checking is performed to improve the accuracy of the conversion.

[0153] The converted text data is encrypted to protect user privacy and sent to the server using a secure communication protocol (e.g., HTTPS). The server adds the received text data to an analysis queue and immediately analyzes it using a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and analyzes keywords and contextual information contained in the text data to assess the likelihood of fraud. This assessment is output as a numerical score.

[0154] If the evaluation determines that the call is likely to be fraudulent, the server immediately sends a notification to the user and pre-defined contacts via a dedicated app, SMS, email, etc. The notification will include a summary of the call and a warning that the call may be fraudulent.

[0155] As a concrete example, consider the case where an elderly person receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." When the call begins, the device collects the audio and converts it into text data using a speech recognition engine. The converted text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud, and determines that the call is likely to be fraudulent. As a result, the server sends a warning notification to the elderly person and their family. The family member who receives the notification can immediately contact the elderly person and provide support to prevent them from becoming victims of fraud.

[0156] An example of a prompt might be:

[0157] "Please rate the following calls for signs of fraud.

[0158] It said: 'I lost my wallet and need money.'

[0159] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0160] Step 1:

[0161] When a user initiates a call, the device collects the call's audio data: the smartphone's microphone captures the sound and stores the audio data in digital form.

[0162] Input: Audio during a call

[0163] Output: Digital audio data

[0164] Step 2:

[0165] The device converts the collected voice data into text data in real time using a speech recognition engine (such as the Google Speech-to-Text API). Error checking is also performed simultaneously to improve conversion accuracy.

[0166] Input: Digital audio data

[0167] Output: Text data

[0168] Step 3:

[0169] The converted text data is encrypted and sent to the server using a secure communication protocol (HTTPS). The device performs this encryption and transmission.

[0170] Input: Text data

[0171] Output: Encrypted text data sent to the server

[0172] Step 4:

[0173] The server adds the received text data to an analysis queue. A generative AI model then analyzes the text data and assesses the likelihood of fraud. The model is trained on a vast amount of historical fraud data and analyzes keywords and contextual information.

[0174] Input: Encrypted text data

[0175] Output: A numerical score assessing the likelihood of fraud

[0176] Step 5:

[0177] The server checks the analysis results of the generated AI model, and if it determines that there is a high possibility of fraud, it sends a notification to the user via a dedicated application. The notification includes a summary of the call and a warning that "there is a possibility of fraud."

[0178] Input: A numerical score assessing the likelihood of fraud

[0179] Output: Alert notification to the user and the configured contacts

[0180] 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.

[0181] The present invention relates to a system that can recognize a user's emotions during a call and determine the possibility of fraud with higher accuracy based on the emotions. The system includes a user terminal, a server, a generative AI model, an emotion engine, and a notification means.

[0182] 1. Audio data collection:

[0183] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[0184] 2. Audio to text conversion:

[0185] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[0186] 3. Sending text data:

[0187] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[0188] 4. Performing fraud analysis:

[0189] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[0190] 5. Running the Emotion Engine:

[0191] While collecting voice data, the device uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the tone, speed, and pitch of the voice to assess whether the user is feeling surprised or scared. The output of the emotion engine is sent to the server in real time.

[0192] 6. Integrated analysis of emotions and text:

[0193] The server combines the fraud likelihood score output by the generative AI model with the emotion evaluation output by the emotion engine to determine the final fraud likelihood. At this time, the emotion engine's results may be weighted to contribute to the fraud determination.

[0194] 7. Sending Notifications:

[0195] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be a fraud.

[0196] Examples:

[0197] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," when the call begins, A's device collects the audio and converts it into text data. At the same time, the emotion engine analyzes A's tone and pitch to assess whether A is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the server combines the results of the emotion engine and determines that there is a high possibility of fraud, it sends a warning notification to A and his family. A's family can immediately contact A and provide support to prevent fraud.

[0198] In this way, by incorporating the user's emotions into the analysis, the system of the present invention can detect signs of fraud with greater accuracy and send immediate warnings, thereby preventing fraud, particularly targeting the elderly.

[0199] The processing flow will be explained below.

[0200] Step 1:

[0201] A user initiates a call using a phone or voice calling app. Once the call is initiated, the device begins collecting audio data for the call. The audio data is collected in real time, without the user having to take any special action.

[0202] Step 2:

[0203] The device inputs the collected voice data into a speech recognition engine, which uses external services such as the Google Speech-to-Text API to convert the voice data into text data in real time.

[0204] Step 3:

[0205] The device checks the converted text data for errors. If any imperfections that affect the accuracy of the text data are found, the device will perform voice recognition again to improve the accuracy of the text data.

[0206] Step 4:

[0207] The device then inputs the voice data into the emotion engine, which analyzes the tone, speed, and pitch of the voice to assess the user's emotions in real time. The assessment results are output as data to identify emotional states such as surprise, fear, and suspicion.

[0208] Step 5:

[0209] The terminal encrypts the converted text data and the output of the emotion engine using a common encryption algorithm (e.g., AES).

[0210] Step 6:

[0211] The encrypted text data and emotion rating data are sent to the server using a secure communication protocol (e.g., HTTPS), ensuring security and privacy.

[0212] Step 7:

[0213] The server adds the received text data and emotion rating data to a queue for analysis, allowing the server to process large amounts of data efficiently.

[0214] Step 8:

[0215] The server inputs the text data into a generative AI model, which is trained on a large amount of historical fraud case data to accurately assess the likelihood of fraud. The model identifies fraud-specific keywords and contexts and outputs the results as a numerical score.

[0216] Step 9:

[0217] The server analyzes the user's emotional state based on the emotion evaluation data output by the emotion engine. If the user is feeling surprise or fear, this information is taken into account in the final fraud determination.

[0218] Step 10:

[0219] The server combines the fraud score output from the generative AI model with the evaluation results of the emotion engine to determine the final likelihood of fraud. If it determines that there is a high likelihood of fraud, it will give a score above a certain threshold.

[0220] Step 11:

[0221] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined contacts. Notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning message stating that it is a "possible fraud."

[0222] Step 12:

[0223] The user or the person who received the notification will check the contents of the notification. The user can take measures to prevent fraud by reconfirming with the other party based on the notification or reporting it to the police.

[0224] Example 2

[0225] 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."

[0226] There is a need for a system that can accurately detect possible fraud during a call and quickly notify the user. Fraud targeting elderly people is particularly on the rise, and conventional methods have found it difficult to detect fraud based on changes in emotions. Therefore, a system that can simultaneously analyze the user's emotions and more accurately determine the possibility of fraud is needed.

[0227] 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.

[0228] In this invention, the server includes a means for collecting voice data of a call when the user initiates a call, a means for converting the collected voice data into text data in real time, and a means for encrypting the converted text data and transmitting it to the server using a secure communication protocol. This enables highly accurate and rapid detection of signs of fraud occurring during a call and immediate warnings to be sent to the user. In addition, the server includes a means for inputting the text data and the user's emotion evaluation results into a generative AI model and evaluating the likelihood of fraud. This enables integrated analysis that takes user emotions into account, further improving the accuracy of fraud detection.

[0229] "Voice data" refers to information that is a digital recording of the voice that a user makes during a call.

[0230] "Text data" is information obtained by converting voice data into a character string format.

[0231] "Encryption" is a conversion method to protect data from being deciphered by third parties.

[0232] A "secure communications protocol" is a method of transmitting data securely over the Internet. An example is HTTPS.

[0233] A "server" is a computer system that provides services to other computers over a network.

[0234] A "generative AI model" is an artificial intelligence algorithm that performs specific tasks based on large amounts of training data.

[0235] An "emotion engine" is a software module for analyzing voice data and assessing the user's emotional state.

[0236] A "notification" is a message intended to convey specific information to a user or related third party.

[0237] A "speech recognition engine" is a software application for converting voice data into text data.

[0238] A "queue" is a data structure for managing the order of data processing.

[0239] MODE FOR CARRYING OUT THE INVENTION

[0240] The present invention relates to a system that recognizes the emotions of a user during a call in real time and determines the possibility of fraud with high accuracy based on that information. Specific embodiments will be described below.

[0241] First, when a user starts a call, the device automatically collects the call's audio data. The audio data is recorded in high quality using the device's built-in microphone. The collected audio data is then processed in real time.

[0242] The collected voice data is then converted into text data in real time using a voice recognition engine. For example, a common voice recognition service such as the Google Speech-to-Text API is used. During this process, the device performs error checking as needed to improve conversion accuracy. The converted text data is also temporarily stored in a buffer.

[0243] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS). Encryption protects the user's privacy.

[0244] The server queues the received text data for analysis using a generative AI model trained on a large amount of historical fraud data, which analyzes keywords and context within the text data to output a numerical score indicating the likelihood of fraud.

[0245] In parallel, the device analyzes the collected voice data using an emotion engine, which analyzes parameters such as tone, speed, and pitch of the voice to assess whether the user is feeling surprise or fear. The results of the emotion analysis are also sent to the server in real time.

[0246] The server integrates and analyzes the fraud likelihood score output by the generative AI model and the evaluation results of the emotion engine. The data is weighted so that the results of the emotion engine contribute to the fraud judgment. The final fraud likelihood is determined based on the integrated analysis results.

[0247] If it is determined that there is a high possibility of fraud, the server will immediately send a notification. The notification will be sent to the user and to pre-defined recipients (e.g., family and friends). Possible notification methods include email, SMS, and push notifications to a dedicated app. The notification will include a summary of the contents of the call and a warning that it may be a fraud.

[0248] Specific examples

[0249] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," the system of this invention works as follows: When the call begins, A's device collects audio and converts it into text in real time using the Google Speech-to-Text API. At the same time, the emotion engine analyzes A's tone and pitch to assess whether he is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the result indicates a high possibility of fraud, a warning notification is sent to A and his family. The family member who receives the notification can immediately contact A and provide support to prevent fraud.

[0250] Prompt Sentence Examples

[0251] "Imagine a scenario where a senior citizen encounters a scam phone call and explain how AI and an emotion engine can help flag the possibility of a scam."

[0252] The above is a specific embodiment of the system of the present invention. This system can detect signs of fraud with higher accuracy and send immediate warnings by incorporating user sentiment analysis.

[0253] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0254] Program processing steps

[0255] Step 1:

[0256] A user initiates a call

[0257] When a user makes or receives a call, the device automatically starts collecting the call audio. The audio data is converted into digital form in real time and temporarily stored in internal memory. The input is the call initiation trigger, and the output is the digital audio data.

[0258] Step 2:

[0259] The device converts the voice data into text data in real time.

[0260] The collected voice data is converted into text data in real time using the device's built-in speech recognition engine (e.g., Google Speech-to-Text API). The voice data is divided into short segments, each of which is input into the speech recognition engine. The output is converted text data. Specifically, the continuous voice stream is segmented at regular intervals and converted sequentially.

[0261] Step 3:

[0262] The device encrypts the text data and sends it to the server.

[0263] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g. HTTPS). The input is the converted text data, and the output is the encrypted data sent to the server. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[0264] Step 4:

[0265] The server receives the text data and adds it to the analysis queue.

[0266] The server adds the received text data to a queue for analysis. The input is the encrypted and transmitted text data, and the output is the queued data waiting to be analyzed. The specific operation is to dequeue the received data and add it to the data store for analysis.

[0267] Step 5:

[0268] The server uses generative AI models to assess the likelihood of fraud

[0269] The server inputs the text data added to the queue into the generative AI model and performs analysis to evaluate the likelihood of fraud. The input is the text data in the queue, and the output is a score representing the likelihood of fraud. Specifically, the server inputs data into the model, extracts keywords and performs contextual analysis, and scores the likelihood of fraud.

[0270] Step 6:

[0271] The device analyzes the user's emotions in real time

[0272] The emotion engine in the device analyzes the voice data and evaluates the user's emotion (surprise, fear, anxiety, etc.). The input is the collected voice data, and the output is the result of the emotion evaluation. Specifically, it analyzes the voice tone, speed, and pitch to quantify the emotional state.

[0273] Step 7:

[0274] The device sends the emotion evaluation results to the server.

[0275] The emotion evaluation results generated by the emotion engine are encrypted and sent to the server using a secure communication protocol. The input is the emotion evaluation results, and the output is the encrypted transmission data. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[0276] Step 8:

[0277] The server performs integrated analysis of emotion data and text data.

[0278] The server combines the fraud score from the generative AI model and the evaluation results from the emotion engine to determine the final likelihood of fraud. The inputs are the fraud score and the emotion evaluation results, and the output is an integrated score. Specifically, the server weights and analyzes both sets of data to calculate the integrated score.

[0279] Step 9:

[0280] Sends notifications when the server is likely to be fraudulent

[0281] If a fraudulent activity is deemed likely, the server sends a notification to the user and pre-defined contacts. The input is the integrated score, and the output is a notification message. Specifically, the server sends a warning message via email, SMS, or push notification to a dedicated app.

[0282] These are the specific processing steps of the program for this system. This process allows for highly accurate detection of signs of fraud, taking into account the user's emotions during the call, and for immediate notification.

[0283] (Application example 2)

[0284] 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."

[0285] Conventional fraud detection systems generate text data from a user's voice data and evaluate the likelihood of fraud based on that text data. However, there is a need for a system that can more accurately assess the likelihood of fraud by taking into account not only the user's voice but also their emotional information. Furthermore, prompt notification of the results of the assessment is expected to help prevent fraud, especially for the elderly.

[0286] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0287] In this invention, the server includes a means including an emotion engine that recognizes user emotions from voice data, a means for determining the possibility of fraud by integrating the evaluation results of the generative AI model and the results of the emotion engine, and a means for sending a notification when it is determined that there is a high possibility of fraud. This enables highly accurate fraud detection that takes into account both voice data and emotion data, and a rapid response.

[0288] "User" refers to the individual or organization placing the call.

[0289] A "call" refers to a conversation conducted using voice communication means.

[0290] "Voice data" refers to digital information about a user's voice that is recorded during a telephone call.

[0291] "Means of collection" refers to devices or programs used to obtain voice data from calls.

[0292] "Text data" refers to data that has been converted from voice data into text information.

[0293] "Means for converting" refers to a device or program for converting voice data into text data.

[0294] A "server" refers to a computer system that sends, receives, and processes data over a network.

[0295] "Means for transmitting" refers to means for sending collected or generated data to another device or server.

[0296] A "generative AI model" is an artificial intelligence model that is trained on historical data to perform a specific task (in this case, assessing the likelihood of fraud).

[0297] "Means for assessing the likelihood of fraud" refers to devices or programs that use generative AI models to determine the likelihood of fraud based on text data.

[0298] "Means for sending notification" refers to a device or program that sends an alert to the user or a pre-defined notification destination when a fraudulent activity is deemed likely.

[0299] An "emotion engine" refers to a device or software that analyzes voice data and recognizes the user's emotional state.

[0300] "Means for recognizing emotions" refers to means for extracting a user's emotional state from speech data using an emotion engine.

[0301] "Integrated judgment means" refers to the means of combining the evaluation results from the generative AI model with the emotional data from the emotion engine to ultimately determine the possibility of fraud.

[0302] A system for implementing this invention collects voice data from a call when the user initiates a call, converts it into text data in real time, and transmits the text data to a server. It also includes an emotion engine that recognizes the user's emotions from the voice data and integrates the collected data to assess the likelihood of fraud. If it determines that there is a high likelihood of fraud, it sends a notification, thereby protecting the user from fraud.

[0303] Hardware and software used

[0304] 1. Collection of audio data

[0305] Hardware used: Smartphone microphone

[0306] Software used: speech_recognition library

[0307] 2. Converting voice data to text

[0308] Software used: Google Speech-to-Text API

[0309] It is used to convert voice data into text data in real time, and also performs error checking to improve accuracy.

[0310] 3. Sending text data

[0311] Software used: HTTPS protocol

[0312] The converted text data is encrypted and securely sent to the server.

[0313] 4. Performing fraud analysis

[0314] Software used: Generative AI model

[0315] The server analyzes the received text data and assesses the likelihood of fraud. The generative AI model is trained on a large amount of historical fraud case data.

[0316] 5. Emotion Engine

[0317] Hardware used: Smartphone microphone

[0318] Software used: EmotionRecognizer module

[0319] Recognize user emotions from voice data and provide additional information for fraud analysis.

[0320] 6. Pooled Analysis

[0321] Software used: Server-side integrated processing program

[0322] The evaluation results of the generative AI model are combined with the output of the sentiment engine to determine the final likelihood of fraud.

[0323] 7. Sending Notifications

[0324] Software used: notification_service

[0325] If fraud is deemed likely, a warning notification will be sent to the user and emergency contacts.

[0326] Specific examples

[0327] For example, if an elderly person receives a call from someone claiming to be their "son" saying, "I lost my wallet and need money," the smartphone application collects the audio and converts it into text data using the Google Speech-to-Text API. At the same time, the EmotionRecognizer module analyzes the tone and pitch of the elderly person's voice to detect surprise or anxiety. This data is sent to a server, where a generative AI model analyzes the possibility of fraud. The application also integrates the results of the emotion engine, and if it determines that fraud is likely, it sends a warning notification to the elderly person and their family.

[0328] Example prompt sentence:

[0329] Analyze for potential fraud. Input text: "I'm your son. I lost my wallet and need money. Please help me.". Emotion rating: Fear.

[0330] In this way, the system for implementing the invention achieves highly accurate fraud detection that takes into account both voice data and emotional data, and rapid response.

[0331] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0332] Step 1:

[0333] When a user starts a call, the device uses the microphone to collect the audio data of the call. This audio data is collected in RAW format. The input is the audio data of the call, and the output is an audio data file.

[0334] Step 2:

[0335] The collected voice data is converted to text data in real time using the Google Speech-to-Text API on the device. During this conversion process, the voice data is converted to text data and error checking is performed. The input is a voice data file and the output is text data.

[0336] Step 3:

[0337] The converted text data is sent to the server using the HTTPS protocol. At this time, the text data is encrypted and sent to protect the user's privacy. The input is the text data, and the output is an HTTPS request containing the encrypted text data.

[0338] Step 4:

[0339] The server inputs the received text data into a generative AI model to evaluate the likelihood of fraud. This generative AI model is trained based on past fraud case data and outputs the likelihood of fraud as a numerical score. The input is text data, and the output is a score indicating the likelihood of fraud.

[0340] Step 5:

[0341] While collecting voice data, the device uses the EmotionRecognizer module to recognize the user's emotions from the voice data. This module analyzes the tone, speed, and pitch of the voice to evaluate whether the user is feeling surprise or fear. The input is the voice data, and the output is an emotion evaluation score.

[0342] Step 6:

[0343] The server integrates the fraud likelihood score output by the generative AI model and the emotion evaluation output by the emotion engine. By performing an integrated analysis, it determines the final fraud likelihood. The input is the fraud likelihood score and the emotion evaluation score, and the output is the integrated fraud likelihood score.

[0344] Step 7:

[0345] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Notification methods include email, SMS, and push notifications on a dedicated app. The input is the integrated fraud probability score, and the output is a warning notification.

[0346] 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.

[0347] 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.

[0348] 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.

[0349] [Second embodiment]

[0350] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0351] 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.

[0352] 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).

[0353] 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.

[0354] 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.

[0355] 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).

[0356] 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.

[0357] 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.

[0358] 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.

[0359] 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.

[0360] 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.

[0361] 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."

[0362] The present invention relates to a system that accurately detects potential fraud while a user is on a call and prevents damage before it occurs. The system includes a user terminal, a server, a generative AI model, and a notification means.

[0363] 1. Audio data collection:

[0364] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[0365] 2. Audio to text conversion:

[0366] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[0367] 3. Sending text data:

[0368] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[0369] 4. Performing fraud analysis:

[0370] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[0371] 5. Sending notifications:

[0372] If the server determines that a call is likely to be fraudulent based on the analysis results of the generative AI model, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be fraudulent.

[0373] Examples:

[0374] For example, suppose elderly person A receives a call from someone claiming to be his "son" who says, "I lost my wallet and need money." When the call begins, A's device collects the audio and converts it into text data. This text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud and determines that there is a high possibility of fraud. Based on the results, the server sends a warning notification to A and his family. Upon receiving the notification, A's family can immediately contact A and provide support to prevent the victim from falling victim to fraud.

[0375] In this way, the system of the present invention can detect signs of fraud in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[0376] The processing flow will be explained below.

[0377] Step 1:

[0378] When a user initiates a call using a phone or voice call app, the device automatically begins collecting the call's audio data. The collection occurs in real time, and the user does not need to take any special action.

[0379] Step 2:

[0380] The device inputs the collected voice data into a speech recognition engine in real time. This speech recognition engine converts the voice data into text data using an external service, such as the Google Speech-to-Text API. The converted text data is temporarily stored in the device's memory.

[0381] Step 3:

[0382] The device checks the converted text data for errors, and if any imperfections that affect recognition accuracy are found, the device runs the speech recognition again to improve the accuracy of the text data.

[0383] Step 4:

[0384] The terminal encrypts the text data. This encryption process is necessary to protect the user's privacy, and uses a commonly used encryption algorithm (e.g., AES).

[0385] Step 5:

[0386] The encrypted text data is sent to the server using a secure communications protocol (e.g., HTTPS), with appropriate procedures taken to ensure data integrity and privacy.

[0387] Step 6:

[0388] The server adds the received text data to a queue for analysis, which allows the server to process large amounts of data efficiently.

[0389] Step 7:

[0390] The server takes the text data and inputs it into a generative AI model, which has been pre-trained on a large amount of past fraud case data and is able to assess the likelihood of fraud with a high degree of accuracy.

[0391] Step 8:

[0392] The generative AI model analyzes input text data to detect signs of fraud. The model identifies keywords and context specific to fraud and outputs the results as a numerical score.

[0393] Step 9:

[0394] The server determines whether the likelihood of fraud is high or low based on the numerical score output by the generative AI model. A specific threshold is set, and if the threshold is exceeded, it is determined that there is a "high likelihood of fraud."

[0395] Step 10:

[0396] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined notification recipients via email, SMS, or push notification on a dedicated app.

[0397] Step 11:

[0398] The user or the person who received the notification will check the contents of the notification. Based on the notification, the user can take measures such as double-checking with the other party or reporting the call to the police, thereby preventing fraud from occurring.

[0399] Example 1

[0400] 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."

[0401] In recent years, fraud methods targeting the elderly have become increasingly sophisticated, with the number of fraudulent incidents occurring during phone calls increasing. Therefore, there is a need for a system that allows users to identify potential fraud in real time and prevent it before it occurs. However, existing systems have difficulty in detecting fraud with high accuracy and providing prompt notification, which leaves elderly people and their families unable to take effective countermeasures.

[0402] 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.

[0403] In this invention, the server includes means for collecting voice data, means for converting the voice data into text data in real time, means for transmitting the converted text data to the server, means for adding the text data received by the server to an analysis queue and evaluating the possibility of fraud using a generative AI model, and means for sending a notification when it is determined that there is a high possibility of fraud. This allows users to accurately detect signs of fraud during a call and receive a prompt warning.

[0404] A "means" is a component of an apparatus, method, or system for achieving a particular function.

[0405] "Voice data" refers to data that is a digital recording of a user's voice during a call.

[0406] "Real-time" refers to processing that occurs almost simultaneously with little delay.

[0407] "Text data" is character information converted from voice data.

[0408] A "server" is a computer system that processes and stores data over a network.

[0409] An "analysis queue" is a data structure for sequentially managing data and tasks waiting to be processed.

[0410] A "generative AI model" is a machine learning model that is trained to perform a specific task using artificial intelligence techniques.

[0411] "Probability of fraud" refers to the probability or risk that the content of a call is fraudulent.

[0412] "Evaluation" refers to making a judgment based on data and information according to specific indicators and criteria.

[0413] A "notification" is a message or signal that notifies a user or a third party of specific information.

[0414] The present invention relates to a system that detects and prevents fraudulent acts during phone calls with high accuracy and in real time. This system is composed of a user terminal, a server, etc. Each component and its operation will be described in detail below.

[0415] 1. Collection of audio data

[0416] When a user initiates a call, the device automatically captures the call's audio data using a high-definition microphone, which records the audio in digital form.

[0417] 2. Converting voice data to text

[0418] The collected voice data is converted into text data in real time by the device's voice recognition engine. The voice recognition engine uses common voice recognition services such as the Google Speech-to-Text API. Error checking is performed as needed to improve conversion accuracy.

[0419] 3. Sending text data

[0420] The converted text data is encrypted to protect the user's privacy and sent to the server using a secure communication protocol (e.g., HTTPS). This communication is done in real time and security measures are in place to prevent data leakage.

[0421] 4. Performing fraud analysis

[0422] The server adds the received text data to an analysis queue. The text data added to the analysis queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and detects specific keywords and contextual information to assess the likelihood of fraud. The analysis results are output as a numerical score, with a higher score indicating a higher likelihood of fraud.

[0423] 5. Sending notifications

[0424] If the call is assessed as likely to be fraudulent, the server will send a notification to the subscriber and pre-defined recipients via email, SMS, or push notification via a dedicated app. The notification will include a summary of the call and a warning that "there is a possibility of fraud."

[0425] Specific examples

[0426] For example, suppose an elderly user receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." In this case, when the call begins, the user's device automatically collects voice data and converts it into text data. This text data is then encrypted and sent to a server. The server then analyzes the call using a generative AI model, detects phrases such as "I lost my wallet" and "I need money," and determines that the call is likely a scam. Based on the results, the server then sends a warning notification to the user and their family. The family member who receives the notification can quickly contact the user and provide support to prevent fraud.

[0427] In this way, the system of the present invention can detect fraudulent activity in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[0428] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0429] Step 1:

[0430] When a user initiates a call, the terminal automatically collects the voice data of the call.

[0431] Input: User's voice during a call

[0432] Data processing: Capture audio as digital data using a high-precision microphone

[0433] Output: Digital audio data

[0434] Step 2:

[0435] The collected voice data is converted into text data in real time using the device's voice recognition engine.

[0436] Input: Digital audio data

[0437] Data calculation: Converts voice into text using a voice recognition engine (e.g., Google Speech-to-Text API).

[0438] Output: Text data

[0439] Specific operation: Performs error check and reprocesses if misrecognition is found.

[0440] Step 3:

[0441] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS).

[0442] Input: Text data

[0443] Data processing: Encrypting data using encryption algorithms

[0444] Output: Encrypted text data

[0445] What it does: Sends encrypted data to the server via HTTPS protocol

[0446] Step 4:

[0447] The server adds the received text data to an analysis queue and evaluates the likelihood of fraud using a generative AI model.

[0448] Input: Encrypted text data

[0449] Data processing: Decrypting and adding text data to the analysis queue

[0450] Data calculation: Analyzes data using a generative AI model and calculates a fraud probability score.

[0451] Output: A numerical score indicating the likelihood of fraud.

[0452] What it does: Detects specific keywords and contextual information and makes assessments based on historical fraud data.

[0453] Step 5:

[0454] If the possibility of fraud is assessed to be high, the server will send a notification to the subscriber and to a pre-defined notification address.

[0455] Input: Fraud likelihood score

[0456] Data processing: Generate notification messages and include necessary information

[0457] Output: Information message

[0458] Specific operation: A warning message is sent via email, SMS, or push notification of a dedicated app. The notification includes a summary of the call and a warning message that the call may be fraudulent.

[0459] (Application example 1)

[0460] 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."

[0461] In recent years, there has been an increase in telephone fraud, with elderly people being particularly vulnerable to these attacks. Conventional countermeasures have been limited in their effectiveness in preventing fraud, as they have struggled to detect fraud in real time and issue prompt warnings. Therefore, there is a need for a system that allows users to instantly detect signs of fraud while on the phone and prevent fraud before it occurs.

[0462] 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.

[0463] In this invention, the server includes means for collecting voice data of the call when a user starts a call, means for converting the collected voice data into text data in real time, means for transmitting the converted text data to the server, means for the server to input the received text data into a generative AI model and evaluate the possibility of fraud, means for sending a notification when it is determined that there is a high possibility of fraud, and means for detecting signs of fraud in real time using an application installed on the smartphone and sending a warning, thereby enabling the possibility of fraud to be detected immediately during a call and enabling a prompt response.

[0464] "User" refers to any entity that includes an individual who makes a call using this system.

[0465] "Call voice data" refers to data that is a digital recording of the voice uttered by a user during a call.

[0466] A "collection means" is a combination of hardware and software for capturing audio data during a user's call.

[0467] "Means for converting voice data into text data in real time" refers to the speech recognition engine and associated software used to convert collected voice data into text data on the fly.

[0468] The "transmitting means" is an internet communication means for encrypting the converted text data and transmitting it to a server using a secure communication protocol.

[0469] A "generative AI model" is an artificial intelligence model trained based on a huge amount of past fraud data, and analyzes text data to assess the likelihood of fraud.

[0470] The "assessment means" is a combination of software in which a generative AI model analyzes input text data and outputs a numerical score indicating the likelihood of fraud.

[0471] The "means for sending a notification" is a communication means for sending a warning to the user and a pre-defined notification destination when it is determined that there is a high possibility of fraud.

[0472] "Applications installed on smartphones" are dedicated programs that run on smartphones to execute the functions of this system.

[0473] "Warning method" is a communication method such as email, SMS, or push notification used to notify the user of potential fraud.

[0474] The present invention provides a system for detecting potential fraud with high accuracy while a user is making a call and preventing damage before it occurs. This system includes a user terminal, a server, a generative AI model, and a notification means. Specific embodiments of the system are described below.

[0475] First, when a user starts a call, the device automatically collects voice data. The smartphone's microphone and voice recognition engine (e.g., Google Speech-to-Text API) are used to collect the voice data. The voice recognition engine converts the collected voice data into text data in real time. Appropriate error checking is performed to improve the accuracy of the conversion.

[0476] The converted text data is encrypted to protect user privacy and sent to the server using a secure communication protocol (e.g., HTTPS). The server adds the received text data to an analysis queue and immediately analyzes it using a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and analyzes keywords and contextual information contained in the text data to assess the likelihood of fraud. This assessment is output as a numerical score.

[0477] If the evaluation determines that the call is likely to be fraudulent, the server immediately sends a notification to the user and pre-defined contacts via a dedicated app, SMS, email, etc. The notification will include a summary of the call and a warning that the call may be fraudulent.

[0478] As a concrete example, consider the case where an elderly person receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." When the call begins, the device collects the audio and converts it into text data using a speech recognition engine. The converted text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud, and determines that the call is likely to be fraudulent. As a result, the server sends a warning notification to the elderly person and their family. The family member who receives the notification can immediately contact the elderly person and provide support to prevent them from becoming victims of fraud.

[0479] An example of a prompt might be:

[0480] "Please rate the following calls for signs of fraud.

[0481] It said: 'I lost my wallet and need money.'

[0482] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0483] Step 1:

[0484] When a user initiates a call, the device collects the call's audio data: the smartphone's microphone captures the sound and stores the audio data in digital form.

[0485] Input: Audio during a call

[0486] Output: Digital audio data

[0487] Step 2:

[0488] The device converts the collected voice data into text data in real time using a speech recognition engine (such as the Google Speech-to-Text API). Error checking is also performed simultaneously to improve conversion accuracy.

[0489] Input: Digital audio data

[0490] Output: Text data

[0491] Step 3:

[0492] The converted text data is encrypted and sent to the server using a secure communication protocol (HTTPS). The device performs this encryption and transmission.

[0493] Input: Text data

[0494] Output: Encrypted text data sent to the server

[0495] Step 4:

[0496] The server adds the received text data to an analysis queue. A generative AI model then analyzes the text data and assesses the likelihood of fraud. The model is trained on a vast amount of historical fraud data and analyzes keywords and contextual information.

[0497] Input: Encrypted text data

[0498] Output: A numerical score assessing the likelihood of fraud

[0499] Step 5:

[0500] The server checks the analysis results of the generated AI model, and if it determines that there is a high possibility of fraud, it sends a notification to the user via a dedicated application. The notification includes a summary of the call and a warning that "there is a possibility of fraud."

[0501] Input: A numerical score assessing the likelihood of fraud

[0502] Output: Alert notification to the user and the configured contacts

[0503] 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.

[0504] The present invention relates to a system that can recognize a user's emotions during a call and determine the possibility of fraud with higher accuracy based on the emotions. The system includes a user terminal, a server, a generative AI model, an emotion engine, and a notification means.

[0505] 1. Audio data collection:

[0506] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[0507] 2. Audio to text conversion:

[0508] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[0509] 3. Sending text data:

[0510] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[0511] 4. Performing fraud analysis:

[0512] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[0513] 5. Running the Emotion Engine:

[0514] While collecting voice data, the device uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the tone, speed, and pitch of the voice to assess whether the user is feeling surprised or scared. The output of the emotion engine is sent to the server in real time.

[0515] 6. Integrated analysis of emotions and text:

[0516] The server combines the fraud likelihood score output by the generative AI model with the emotion evaluation output by the emotion engine to determine the final fraud likelihood. At this time, the emotion engine's results may be weighted to contribute to the fraud determination.

[0517] 7. Sending Notifications:

[0518] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be a fraud.

[0519] Examples:

[0520] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," when the call begins, A's device collects the audio and converts it into text data. At the same time, the emotion engine analyzes A's tone and pitch to assess whether A is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the server combines the results of the emotion engine and determines that there is a high possibility of fraud, it sends a warning notification to A and his family. A's family can immediately contact A and provide support to prevent fraud.

[0521] In this way, by incorporating the user's emotions into the analysis, the system of the present invention can detect signs of fraud with greater accuracy and send immediate warnings, thereby preventing fraud, particularly targeting the elderly.

[0522] The processing flow will be explained below.

[0523] Step 1:

[0524] A user initiates a call using a phone or voice calling app. Once the call is initiated, the device begins collecting audio data for the call. The audio data is collected in real time, without the user having to take any special action.

[0525] Step 2:

[0526] The device inputs the collected voice data into a speech recognition engine, which uses external services such as the Google Speech-to-Text API to convert the voice data into text data in real time.

[0527] Step 3:

[0528] The device checks the converted text data for errors. If any imperfections that affect the accuracy of the text data are found, the device will perform voice recognition again to improve the accuracy of the text data.

[0529] Step 4:

[0530] The device then inputs the voice data into the emotion engine, which analyzes the tone, speed, and pitch of the voice to assess the user's emotions in real time. The assessment results are output as data to identify emotional states such as surprise, fear, and suspicion.

[0531] Step 5:

[0532] The terminal encrypts the converted text data and the output of the emotion engine using a common encryption algorithm (e.g., AES).

[0533] Step 6:

[0534] The encrypted text data and emotion rating data are sent to the server using a secure communication protocol (e.g., HTTPS), ensuring security and privacy.

[0535] Step 7:

[0536] The server adds the received text data and emotion rating data to a queue for analysis, allowing the server to process large amounts of data efficiently.

[0537] Step 8:

[0538] The server inputs the text data into a generative AI model, which is trained on a large amount of historical fraud case data to accurately assess the likelihood of fraud. The model identifies fraud-specific keywords and contexts and outputs the results as a numerical score.

[0539] Step 9:

[0540] The server analyzes the user's emotional state based on the emotion evaluation data output by the emotion engine. If the user is feeling surprise or fear, this information is taken into account in the final fraud determination.

[0541] Step 10:

[0542] The server combines the fraud score output from the generative AI model with the evaluation results of the emotion engine to determine the final likelihood of fraud. If it determines that there is a high likelihood of fraud, it will give a score above a certain threshold.

[0543] Step 11:

[0544] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined contacts. Notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning message stating that it is a "possible fraud."

[0545] Step 12:

[0546] The user or the person who received the notification will check the contents of the notification. The user can take measures to prevent fraud by reconfirming with the other party based on the notification or reporting it to the police.

[0547] Example 2

[0548] 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."

[0549] There is a need for a system that can accurately detect possible fraud during a call and quickly notify the user. Fraud targeting elderly people is particularly on the rise, and conventional methods have found it difficult to detect fraud based on changes in emotions. Therefore, a system that can simultaneously analyze the user's emotions and more accurately determine the possibility of fraud is needed.

[0550] 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.

[0551] In this invention, the server includes a means for collecting voice data of a call when the user initiates a call, a means for converting the collected voice data into text data in real time, and a means for encrypting the converted text data and transmitting it to the server using a secure communication protocol. This enables highly accurate and rapid detection of signs of fraud occurring during a call and immediate warnings to be sent to the user. In addition, the server includes a means for inputting the text data and the user's emotion evaluation results into a generative AI model and evaluating the likelihood of fraud. This enables integrated analysis that takes user emotions into account, further improving the accuracy of fraud detection.

[0552] "Voice data" refers to information that is a digital recording of the voice that a user makes during a call.

[0553] "Text data" is information obtained by converting voice data into a character string format.

[0554] "Encryption" is a conversion method to protect data from being deciphered by third parties.

[0555] A "secure communications protocol" is a method of transmitting data securely over the Internet. An example is HTTPS.

[0556] A "server" is a computer system that provides services to other computers over a network.

[0557] A "generative AI model" is an artificial intelligence algorithm that performs specific tasks based on large amounts of training data.

[0558] An "emotion engine" is a software module for analyzing voice data and assessing the user's emotional state.

[0559] A "notification" is a message intended to convey specific information to a user or related third party.

[0560] A "speech recognition engine" is a software application for converting voice data into text data.

[0561] A "queue" is a data structure for managing the order of data processing.

[0562] MODE FOR CARRYING OUT THE INVENTION

[0563] The present invention relates to a system that recognizes the emotions of a user during a call in real time and determines the possibility of fraud with high accuracy based on that information. Specific embodiments will be described below.

[0564] First, when a user starts a call, the device automatically collects the call's audio data. The audio data is recorded in high quality using the device's built-in microphone. The collected audio data is then processed in real time.

[0565] The collected voice data is then converted into text data in real time using a voice recognition engine. For example, a common voice recognition service such as the Google Speech-to-Text API is used. During this process, the device performs error checking as needed to improve conversion accuracy. The converted text data is also temporarily stored in a buffer.

[0566] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS). Encryption protects the user's privacy.

[0567] The server queues the received text data for analysis using a generative AI model trained on a large amount of historical fraud data, which analyzes keywords and context within the text data to output a numerical score indicating the likelihood of fraud.

[0568] In parallel, the device analyzes the collected voice data using an emotion engine, which analyzes parameters such as tone, speed, and pitch of the voice to assess whether the user is feeling surprise or fear. The results of the emotion analysis are also sent to the server in real time.

[0569] The server integrates and analyzes the fraud likelihood score output by the generative AI model and the evaluation results of the emotion engine. The data is weighted so that the results of the emotion engine contribute to the fraud judgment. The final fraud likelihood is determined based on the integrated analysis results.

[0570] If it is determined that there is a high possibility of fraud, the server will immediately send a notification. The notification will be sent to the user and to pre-defined recipients (e.g., family and friends). Possible notification methods include email, SMS, and push notifications to a dedicated app. The notification will include a summary of the contents of the call and a warning that it may be a fraud.

[0571] Specific examples

[0572] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," the system of this invention works as follows: When the call begins, A's device collects audio and converts it into text in real time using the Google Speech-to-Text API. At the same time, the emotion engine analyzes A's tone and pitch to assess whether he is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the result indicates a high possibility of fraud, a warning notification is sent to A and his family. The family member who receives the notification can immediately contact A and provide support to prevent fraud.

[0573] Prompt Sentence Examples

[0574] "Imagine a scenario where a senior citizen encounters a scam phone call and explain how AI and an emotion engine can help flag the possibility of a scam."

[0575] The above is a specific embodiment of the system of the present invention. This system can detect signs of fraud with higher accuracy and send immediate warnings by incorporating user sentiment analysis.

[0576] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0577] Program processing steps

[0578] Step 1:

[0579] A user initiates a call

[0580] When a user makes or receives a call, the device automatically starts collecting the call audio. The audio data is converted into digital form in real time and temporarily stored in internal memory. The input is the call initiation trigger, and the output is the digital audio data.

[0581] Step 2:

[0582] The device converts the voice data into text data in real time.

[0583] The collected voice data is converted into text data in real time using the device's built-in speech recognition engine (e.g., Google Speech-to-Text API). The voice data is divided into short segments, each of which is input into the speech recognition engine. The output is converted text data. Specifically, the continuous voice stream is segmented at regular intervals and converted sequentially.

[0584] Step 3:

[0585] The device encrypts the text data and sends it to the server.

[0586] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g. HTTPS). The input is the converted text data, and the output is the encrypted data sent to the server. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[0587] Step 4:

[0588] The server receives the text data and adds it to the analysis queue.

[0589] The server adds the received text data to a queue for analysis. The input is the encrypted and transmitted text data, and the output is the queued data waiting to be analyzed. The specific operation is to dequeue the received data and add it to the data store for analysis.

[0590] Step 5:

[0591] The server uses generative AI models to assess the likelihood of fraud

[0592] The server inputs the text data added to the queue into the generative AI model and performs analysis to evaluate the likelihood of fraud. The input is the text data in the queue, and the output is a score representing the likelihood of fraud. Specifically, the server inputs data into the model, extracts keywords and performs contextual analysis, and scores the likelihood of fraud.

[0593] Step 6:

[0594] The device analyzes the user's emotions in real time

[0595] The emotion engine in the device analyzes the voice data and evaluates the user's emotion (surprise, fear, anxiety, etc.). The input is the collected voice data, and the output is the result of the emotion evaluation. Specifically, it analyzes the voice tone, speed, and pitch to quantify the emotional state.

[0596] Step 7:

[0597] The device sends the emotion evaluation results to the server.

[0598] The emotion evaluation results generated by the emotion engine are encrypted and sent to the server using a secure communication protocol. The input is the emotion evaluation results, and the output is the encrypted transmission data. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[0599] Step 8:

[0600] The server performs integrated analysis of emotion data and text data.

[0601] The server combines the fraud score from the generative AI model and the evaluation results from the emotion engine to determine the final likelihood of fraud. The inputs are the fraud score and the emotion evaluation results, and the output is an integrated score. Specifically, the server weights and analyzes both sets of data to calculate the integrated score.

[0602] Step 9:

[0603] Sends notifications when the server is likely to be fraudulent

[0604] If a fraudulent activity is deemed likely, the server sends a notification to the user and pre-defined contacts. The input is the integrated score, and the output is a notification message. Specifically, the server sends a warning message via email, SMS, or push notification to a dedicated app.

[0605] These are the specific processing steps of the program for this system. This process allows for highly accurate detection of signs of fraud, taking into account the user's emotions during the call, and for immediate notification.

[0606] (Application example 2)

[0607] 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."

[0608] Conventional fraud detection systems generate text data from a user's voice data and evaluate the likelihood of fraud based on that text data. However, there is a need for a system that can more accurately assess the likelihood of fraud by taking into account not only the user's voice but also their emotional information. Furthermore, prompt notification of the results of the assessment is expected to help prevent fraud, especially for the elderly.

[0609] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0610] In this invention, the server includes a means including an emotion engine that recognizes user emotions from voice data, a means for determining the possibility of fraud by integrating the evaluation results of the generative AI model and the results of the emotion engine, and a means for sending a notification when it is determined that there is a high possibility of fraud. This enables highly accurate fraud detection that takes into account both voice data and emotion data, and a rapid response.

[0611] "User" refers to the individual or organization placing the call.

[0612] A "call" refers to a conversation conducted using voice communication means.

[0613] "Voice data" refers to digital information about a user's voice that is recorded during a telephone call.

[0614] "Means of collection" refers to devices or programs used to obtain voice data from calls.

[0615] "Text data" refers to data that has been converted from voice data into text information.

[0616] "Means for converting" refers to a device or program for converting voice data into text data.

[0617] A "server" refers to a computer system that sends, receives, and processes data over a network.

[0618] "Means for transmitting" refers to means for sending collected or generated data to another device or server.

[0619] A "generative AI model" is an artificial intelligence model that is trained on historical data to perform a specific task (in this case, assessing the likelihood of fraud).

[0620] "Means for assessing the likelihood of fraud" refers to devices or programs that use generative AI models to determine the likelihood of fraud based on text data.

[0621] "Means for sending notification" refers to a device or program that sends an alert to the user or a pre-defined notification destination when a fraudulent activity is deemed likely.

[0622] An "emotion engine" refers to a device or software that analyzes voice data and recognizes the user's emotional state.

[0623] "Means for recognizing emotions" refers to means for extracting a user's emotional state from speech data using an emotion engine.

[0624] "Integrated judgment means" refers to the means of combining the evaluation results from the generative AI model with the emotional data from the emotion engine to ultimately determine the possibility of fraud.

[0625] A system for implementing this invention collects voice data from a call when the user initiates a call, converts it into text data in real time, and transmits the text data to a server. It also includes an emotion engine that recognizes the user's emotions from the voice data and integrates the collected data to assess the likelihood of fraud. If it determines that there is a high likelihood of fraud, it sends a notification, thereby protecting the user from fraud.

[0626] Hardware and software used

[0627] 1. Collection of audio data

[0628] Hardware used: Smartphone microphone

[0629] Software used: speech_recognition library

[0630] 2. Converting voice data to text

[0631] Software used: Google Speech-to-Text API

[0632] It is used to convert voice data into text data in real time, and also performs error checking to improve accuracy.

[0633] 3. Sending text data

[0634] Software used: HTTPS protocol

[0635] The converted text data is encrypted and securely sent to the server.

[0636] 4. Performing fraud analysis

[0637] Software used: Generative AI model

[0638] The server analyzes the received text data and assesses the likelihood of fraud. The generative AI model is trained on a large amount of historical fraud case data.

[0639] 5. Emotion Engine

[0640] Hardware used: Smartphone microphone

[0641] Software used: EmotionRecognizer module

[0642] Recognize user emotions from voice data and provide additional information for fraud analysis.

[0643] 6. Pooled Analysis

[0644] Software used: Server-side integrated processing program

[0645] The evaluation results of the generative AI model are combined with the output of the sentiment engine to determine the final likelihood of fraud.

[0646] 7. Sending Notifications

[0647] Software used: notification_service

[0648] If fraud is deemed likely, a warning notification will be sent to the user and emergency contacts.

[0649] Specific examples

[0650] For example, if an elderly person receives a call from someone claiming to be their "son" saying, "I lost my wallet and need money," the smartphone application collects the audio and converts it into text data using the Google Speech-to-Text API. At the same time, the EmotionRecognizer module analyzes the tone and pitch of the elderly person's voice to detect surprise or anxiety. This data is sent to a server, where a generative AI model analyzes the possibility of fraud. The application also integrates the results of the emotion engine, and if it determines that fraud is likely, it sends a warning notification to the elderly person and their family.

[0651] Example prompt sentence:

[0652] Analyze for potential fraud. Input text: "I'm your son. I lost my wallet and need money. Please help me.". Emotion rating: Fear.

[0653] In this way, the system for implementing the invention achieves highly accurate fraud detection that takes into account both voice data and emotional data, and rapid response.

[0654] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0655] Step 1:

[0656] When a user starts a call, the device uses the microphone to collect the audio data of the call. This audio data is collected in RAW format. The input is the audio data of the call, and the output is an audio data file.

[0657] Step 2:

[0658] The collected voice data is converted to text data in real time using the Google Speech-to-Text API on the device. During this conversion process, the voice data is converted to text data and error checking is performed. The input is a voice data file and the output is text data.

[0659] Step 3:

[0660] The converted text data is sent to the server using the HTTPS protocol. At this time, the text data is encrypted and sent to protect the user's privacy. The input is the text data, and the output is an HTTPS request containing the encrypted text data.

[0661] Step 4:

[0662] The server inputs the received text data into a generative AI model to evaluate the likelihood of fraud. This generative AI model is trained based on past fraud case data and outputs the likelihood of fraud as a numerical score. The input is text data, and the output is a score indicating the likelihood of fraud.

[0663] Step 5:

[0664] While collecting voice data, the device uses the EmotionRecognizer module to recognize the user's emotions from the voice data. This module analyzes the tone, speed, and pitch of the voice to evaluate whether the user is feeling surprise or fear. The input is the voice data, and the output is an emotion evaluation score.

[0665] Step 6:

[0666] The server integrates the fraud likelihood score output by the generative AI model and the emotion evaluation output by the emotion engine. By performing an integrated analysis, it determines the final fraud likelihood. The input is the fraud likelihood score and the emotion evaluation score, and the output is the integrated fraud likelihood score.

[0667] Step 7:

[0668] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Notification methods include email, SMS, and push notifications on a dedicated app. The input is the integrated fraud probability score, and the output is a warning notification.

[0669] 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.

[0670] 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.

[0671] 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.

[0672] [Third embodiment]

[0673] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0674] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0675] 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).

[0676] 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.

[0677] 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.

[0678] 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).

[0679] 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.

[0680] 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.

[0681] 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.

[0682] 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.

[0683] 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.

[0684] 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."

[0685] The present invention relates to a system that accurately detects potential fraud while a user is on a call and prevents damage before it occurs. The system includes a user terminal, a server, a generative AI model, and a notification means.

[0686] 1. Audio data collection:

[0687] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[0688] 2. Audio to text conversion:

[0689] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[0690] 3. Sending text data:

[0691] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[0692] 4. Performing fraud analysis:

[0693] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[0694] 5. Sending notifications:

[0695] If the server determines that a call is likely to be fraudulent based on the analysis results of the generative AI model, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be fraudulent.

[0696] Examples:

[0697] For example, suppose elderly person A receives a call from someone claiming to be his "son" who says, "I lost my wallet and need money." When the call begins, A's device collects the audio and converts it into text data. This text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud and determines that there is a high possibility of fraud. Based on the results, the server sends a warning notification to A and his family. Upon receiving the notification, A's family can immediately contact A and provide support to prevent the victim from falling victim to fraud.

[0698] In this way, the system of the present invention can detect signs of fraud in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[0699] The processing flow will be explained below.

[0700] Step 1:

[0701] When a user initiates a call using a phone or voice call app, the device automatically begins collecting the call's audio data. The collection occurs in real time, and the user does not need to take any special action.

[0702] Step 2:

[0703] The device inputs the collected voice data into a speech recognition engine in real time. This speech recognition engine converts the voice data into text data using an external service, such as the Google Speech-to-Text API. The converted text data is temporarily stored in the device's memory.

[0704] Step 3:

[0705] The device checks the converted text data for errors, and if any imperfections that affect recognition accuracy are found, the device runs the speech recognition again to improve the accuracy of the text data.

[0706] Step 4:

[0707] The terminal encrypts the text data. This encryption process is necessary to protect the user's privacy, and uses a commonly used encryption algorithm (e.g., AES).

[0708] Step 5:

[0709] The encrypted text data is sent to the server using a secure communications protocol (e.g., HTTPS), with appropriate procedures taken to ensure data integrity and privacy.

[0710] Step 6:

[0711] The server adds the received text data to a queue for analysis, which allows the server to process large amounts of data efficiently.

[0712] Step 7:

[0713] The server takes the text data and inputs it into a generative AI model, which has been pre-trained on a large amount of past fraud case data and is able to assess the likelihood of fraud with a high degree of accuracy.

[0714] Step 8:

[0715] The generative AI model analyzes input text data to detect signs of fraud. The model identifies keywords and context specific to fraud and outputs the results as a numerical score.

[0716] Step 9:

[0717] The server determines whether the likelihood of fraud is high or low based on the numerical score output by the generative AI model. A specific threshold is set, and if the threshold is exceeded, it is determined that there is a "high likelihood of fraud."

[0718] Step 10:

[0719] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined notification recipients via email, SMS, or push notification on a dedicated app.

[0720] Step 11:

[0721] The user or the person who received the notification will check the contents of the notification. Based on the notification, the user can take measures such as double-checking with the other party or reporting the call to the police, thereby preventing fraud from occurring.

[0722] Example 1

[0723] 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."

[0724] In recent years, fraud methods targeting the elderly have become increasingly sophisticated, with the number of fraudulent incidents occurring during phone calls increasing. Therefore, there is a need for a system that allows users to identify potential fraud in real time and prevent it before it occurs. However, existing systems have difficulty in detecting fraud with high accuracy and providing prompt notification, which leaves elderly people and their families unable to take effective countermeasures.

[0725] 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.

[0726] In this invention, the server includes means for collecting voice data, means for converting the voice data into text data in real time, means for transmitting the converted text data to the server, means for adding the text data received by the server to an analysis queue and evaluating the possibility of fraud using a generative AI model, and means for sending a notification when it is determined that there is a high possibility of fraud. This allows users to accurately detect signs of fraud during a call and receive a prompt warning.

[0727] A "means" is a component of an apparatus, method, or system for achieving a particular function.

[0728] "Voice data" refers to data that is a digital recording of a user's voice during a call.

[0729] "Real-time" refers to processing that occurs almost simultaneously with little delay.

[0730] "Text data" is character information converted from voice data.

[0731] A "server" is a computer system that processes and stores data over a network.

[0732] An "analysis queue" is a data structure for sequentially managing data and tasks waiting to be processed.

[0733] A "generative AI model" is a machine learning model that is trained to perform a specific task using artificial intelligence techniques.

[0734] "Probability of fraud" refers to the probability or risk that the content of a call is fraudulent.

[0735] "Evaluation" refers to making a judgment based on data and information according to specific indicators and criteria.

[0736] A "notification" is a message or signal that notifies a user or a third party of specific information.

[0737] The present invention relates to a system that detects and prevents fraudulent acts during phone calls with high accuracy and in real time. This system is composed of a user terminal, a server, etc. Each component and its operation will be described in detail below.

[0738] 1. Collection of audio data

[0739] When a user initiates a call, the device automatically captures the call's audio data using a high-definition microphone, which records the audio in digital form.

[0740] 2. Converting voice data to text

[0741] The collected voice data is converted into text data in real time by the device's voice recognition engine. The voice recognition engine uses common voice recognition services such as the Google Speech-to-Text API. Error checking is performed as needed to improve conversion accuracy.

[0742] 3. Sending text data

[0743] The converted text data is encrypted to protect the user's privacy and sent to the server using a secure communication protocol (e.g., HTTPS). This communication is done in real time and security measures are in place to prevent data leakage.

[0744] 4. Performing fraud analysis

[0745] The server adds the received text data to an analysis queue. The text data added to the analysis queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and detects specific keywords and contextual information to assess the likelihood of fraud. The analysis results are output as a numerical score, with a higher score indicating a higher likelihood of fraud.

[0746] 5. Sending notifications

[0747] If the call is assessed as likely to be fraudulent, the server will send a notification to the subscriber and pre-defined recipients via email, SMS, or push notification via a dedicated app. The notification will include a summary of the call and a warning that "there is a possibility of fraud."

[0748] Specific examples

[0749] For example, suppose an elderly user receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." In this case, when the call begins, the user's device automatically collects voice data and converts it into text data. This text data is then encrypted and sent to a server. The server then analyzes the call using a generative AI model, detects phrases such as "I lost my wallet" and "I need money," and determines that the call is likely a scam. Based on the results, the server then sends a warning notification to the user and their family. The family member who receives the notification can quickly contact the user and provide support to prevent fraud.

[0750] In this way, the system of the present invention can detect fraudulent activity in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[0751] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0752] Step 1:

[0753] When a user initiates a call, the terminal automatically collects the voice data of the call.

[0754] Input: User's voice during a call

[0755] Data processing: Capture audio as digital data using a high-precision microphone

[0756] Output: Digital audio data

[0757] Step 2:

[0758] The collected voice data is converted into text data in real time using the device's voice recognition engine.

[0759] Input: Digital audio data

[0760] Data calculation: Converts voice into text using a voice recognition engine (e.g., Google Speech-to-Text API).

[0761] Output: Text data

[0762] Specific operation: Performs error check and reprocesses if misrecognition is found.

[0763] Step 3:

[0764] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS).

[0765] Input: Text data

[0766] Data processing: Encrypting data using encryption algorithms

[0767] Output: Encrypted text data

[0768] What it does: Sends encrypted data to the server via HTTPS protocol

[0769] Step 4:

[0770] The server adds the received text data to an analysis queue and evaluates the likelihood of fraud using a generative AI model.

[0771] Input: Encrypted text data

[0772] Data processing: Decrypting and adding text data to the analysis queue

[0773] Data calculation: Analyzes data using a generative AI model and calculates a fraud probability score.

[0774] Output: A numerical score indicating the likelihood of fraud.

[0775] What it does: Detects specific keywords and contextual information and makes assessments based on historical fraud data.

[0776] Step 5:

[0777] If the possibility of fraud is assessed to be high, the server will send a notification to the subscriber and to a pre-defined notification address.

[0778] Input: Fraud likelihood score

[0779] Data processing: Generate notification messages and include necessary information

[0780] Output: Information message

[0781] Specific operation: A warning message is sent via email, SMS, or push notification of a dedicated app. The notification includes a summary of the call and a warning message that the call may be fraudulent.

[0782] (Application example 1)

[0783] 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."

[0784] In recent years, there has been an increase in telephone fraud, with elderly people being particularly vulnerable to these attacks. Conventional countermeasures have been limited in their effectiveness in preventing fraud, as they have struggled to detect fraud in real time and issue prompt warnings. Therefore, there is a need for a system that allows users to instantly detect signs of fraud while on the phone and prevent fraud before it occurs.

[0785] 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.

[0786] In this invention, the server includes means for collecting voice data of the call when a user starts a call, means for converting the collected voice data into text data in real time, means for transmitting the converted text data to the server, means for the server to input the received text data into a generative AI model and evaluate the possibility of fraud, means for sending a notification when it is determined that there is a high possibility of fraud, and means for detecting signs of fraud in real time using an application installed on the smartphone and sending a warning, thereby enabling the possibility of fraud to be detected immediately during a call and enabling a prompt response.

[0787] "User" refers to any entity that includes an individual who makes a call using this system.

[0788] "Call voice data" refers to data that is a digital recording of the voice uttered by a user during a call.

[0789] A "collection means" is a combination of hardware and software for capturing audio data during a user's call.

[0790] "Means for converting voice data into text data in real time" refers to the speech recognition engine and associated software used to convert collected voice data into text data on the fly.

[0791] The "transmitting means" is an internet communication means for encrypting the converted text data and transmitting it to a server using a secure communication protocol.

[0792] A "generative AI model" is an artificial intelligence model trained based on a huge amount of past fraud data, and analyzes text data to assess the likelihood of fraud.

[0793] The "assessment means" is a combination of software in which a generative AI model analyzes input text data and outputs a numerical score indicating the likelihood of fraud.

[0794] The "means for sending a notification" is a communication means for sending a warning to the user and a pre-defined notification destination when it is determined that there is a high possibility of fraud.

[0795] "Applications installed on smartphones" are dedicated programs that run on smartphones to execute the functions of this system.

[0796] "Warning method" is a communication method such as email, SMS, or push notification used to notify the user of potential fraud.

[0797] The present invention provides a system for detecting potential fraud with high accuracy while a user is making a call and preventing damage before it occurs. This system includes a user terminal, a server, a generative AI model, and a notification means. Specific embodiments of the system are described below.

[0798] First, when a user starts a call, the device automatically collects voice data. The smartphone's microphone and voice recognition engine (e.g., Google Speech-to-Text API) are used to collect the voice data. The voice recognition engine converts the collected voice data into text data in real time. Appropriate error checking is performed to improve the accuracy of the conversion.

[0799] The converted text data is encrypted to protect user privacy and sent to the server using a secure communication protocol (e.g., HTTPS). The server adds the received text data to an analysis queue and immediately analyzes it using a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and analyzes keywords and contextual information contained in the text data to assess the likelihood of fraud. This assessment is output as a numerical score.

[0800] If the evaluation determines that the call is likely to be fraudulent, the server immediately sends a notification to the user and pre-defined contacts via a dedicated app, SMS, email, etc. The notification will include a summary of the call and a warning that the call may be fraudulent.

[0801] As a concrete example, consider the case where an elderly person receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." When the call begins, the device collects the audio and converts it into text data using a speech recognition engine. The converted text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud, and determines that the call is likely to be fraudulent. As a result, the server sends a warning notification to the elderly person and their family. The family member who receives the notification can immediately contact the elderly person and provide support to prevent them from becoming victims of fraud.

[0802] An example of a prompt might be:

[0803] "Please rate the following calls for signs of fraud.

[0804] It said: 'I lost my wallet and need money.'

[0805] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0806] Step 1:

[0807] When a user initiates a call, the device collects the call's audio data: the smartphone's microphone captures the sound and stores the audio data in digital form.

[0808] Input: Audio during a call

[0809] Output: Digital audio data

[0810] Step 2:

[0811] The device converts the collected voice data into text data in real time using a speech recognition engine (such as the Google Speech-to-Text API). Error checking is also performed simultaneously to improve conversion accuracy.

[0812] Input: Digital audio data

[0813] Output: Text data

[0814] Step 3:

[0815] The converted text data is encrypted and sent to the server using a secure communication protocol (HTTPS). The device performs this encryption and transmission.

[0816] Input: Text data

[0817] Output: Encrypted text data sent to the server

[0818] Step 4:

[0819] The server adds the received text data to an analysis queue. A generative AI model then analyzes the text data and assesses the likelihood of fraud. The model is trained on a vast amount of historical fraud data and analyzes keywords and contextual information.

[0820] Input: Encrypted text data

[0821] Output: A numerical score assessing the likelihood of fraud

[0822] Step 5:

[0823] The server checks the analysis results of the generated AI model, and if it determines that there is a high possibility of fraud, it sends a notification to the user via a dedicated application. The notification includes a summary of the call and a warning that "there is a possibility of fraud."

[0824] Input: A numerical score assessing the likelihood of fraud

[0825] Output: Alert notification to the user and the configured contacts

[0826] 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.

[0827] The present invention relates to a system that can recognize a user's emotions during a call and determine the possibility of fraud with higher accuracy based on the emotions. The system includes a user terminal, a server, a generative AI model, an emotion engine, and a notification means.

[0828] 1. Audio data collection:

[0829] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[0830] 2. Audio to text conversion:

[0831] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[0832] 3. Sending text data:

[0833] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[0834] 4. Performing fraud analysis:

[0835] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[0836] 5. Running the Emotion Engine:

[0837] While collecting voice data, the device uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the tone, speed, and pitch of the voice to assess whether the user is feeling surprised or scared. The output of the emotion engine is sent to the server in real time.

[0838] 6. Integrated analysis of emotions and text:

[0839] The server combines the fraud likelihood score output by the generative AI model with the emotion evaluation output by the emotion engine to determine the final fraud likelihood. At this time, the emotion engine's results may be weighted to contribute to the fraud determination.

[0840] 7. Sending Notifications:

[0841] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be a fraud.

[0842] Examples:

[0843] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," when the call begins, A's device collects the audio and converts it into text data. At the same time, the emotion engine analyzes A's tone and pitch to assess whether A is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the server combines the results of the emotion engine and determines that there is a high possibility of fraud, it sends a warning notification to A and his family. A's family can immediately contact A and provide support to prevent fraud.

[0844] In this way, by incorporating the user's emotions into the analysis, the system of the present invention can detect signs of fraud with greater accuracy and send immediate warnings, thereby preventing fraud, particularly targeting the elderly.

[0845] The processing flow will be explained below.

[0846] Step 1:

[0847] A user initiates a call using a phone or voice calling app. Once the call is initiated, the device begins collecting audio data for the call. The audio data is collected in real time, without the user having to take any special action.

[0848] Step 2:

[0849] The device inputs the collected voice data into a speech recognition engine, which uses external services such as the Google Speech-to-Text API to convert the voice data into text data in real time.

[0850] Step 3:

[0851] The device checks the converted text data for errors. If any imperfections that affect the accuracy of the text data are found, the device will perform voice recognition again to improve the accuracy of the text data.

[0852] Step 4:

[0853] The device then inputs the voice data into the emotion engine, which analyzes the tone, speed, and pitch of the voice to assess the user's emotions in real time. The assessment results are output as data to identify emotional states such as surprise, fear, and suspicion.

[0854] Step 5:

[0855] The terminal encrypts the converted text data and the output of the emotion engine using a common encryption algorithm (e.g., AES).

[0856] Step 6:

[0857] The encrypted text data and emotion rating data are sent to the server using a secure communication protocol (e.g., HTTPS), ensuring security and privacy.

[0858] Step 7:

[0859] The server adds the received text data and emotion rating data to a queue for analysis, allowing the server to process large amounts of data efficiently.

[0860] Step 8:

[0861] The server inputs the text data into a generative AI model, which is trained on a large amount of historical fraud case data to accurately assess the likelihood of fraud. The model identifies fraud-specific keywords and contexts and outputs the results as a numerical score.

[0862] Step 9:

[0863] The server analyzes the user's emotional state based on the emotion evaluation data output by the emotion engine. If the user is feeling surprise or fear, this information is taken into account in the final fraud determination.

[0864] Step 10:

[0865] The server combines the fraud score output from the generative AI model with the evaluation results of the emotion engine to determine the final likelihood of fraud. If it determines that there is a high likelihood of fraud, it will give a score above a certain threshold.

[0866] Step 11:

[0867] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined contacts. Notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning message stating that it is a "possible fraud."

[0868] Step 12:

[0869] The user or the person who received the notification will check the contents of the notification. The user can take measures to prevent fraud by reconfirming with the other party based on the notification or reporting it to the police.

[0870] Example 2

[0871] 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."

[0872] There is a need for a system that can accurately detect possible fraud during a call and quickly notify the user. Fraud targeting elderly people is particularly on the rise, and conventional methods have found it difficult to detect fraud based on changes in emotions. Therefore, a system that can simultaneously analyze the user's emotions and more accurately determine the possibility of fraud is needed.

[0873] 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.

[0874] In this invention, the server includes a means for collecting voice data of a call when the user initiates a call, a means for converting the collected voice data into text data in real time, and a means for encrypting the converted text data and transmitting it to the server using a secure communication protocol. This enables highly accurate and rapid detection of signs of fraud occurring during a call and immediate warnings to be sent to the user. In addition, the server includes a means for inputting the text data and the user's emotion evaluation results into a generative AI model and evaluating the likelihood of fraud. This enables integrated analysis that takes user emotions into account, further improving the accuracy of fraud detection.

[0875] "Voice data" refers to information that is a digital recording of the voice that a user makes during a call.

[0876] "Text data" is information obtained by converting voice data into a character string format.

[0877] "Encryption" is a conversion method to protect data from being deciphered by third parties.

[0878] A "secure communications protocol" is a method of transmitting data securely over the Internet. An example is HTTPS.

[0879] A "server" is a computer system that provides services to other computers over a network.

[0880] A "generative AI model" is an artificial intelligence algorithm that performs specific tasks based on large amounts of training data.

[0881] An "emotion engine" is a software module for analyzing voice data and assessing the user's emotional state.

[0882] A "notification" is a message intended to convey specific information to a user or related third party.

[0883] A "speech recognition engine" is a software application for converting voice data into text data.

[0884] A "queue" is a data structure for managing the order of data processing.

[0885] MODE FOR CARRYING OUT THE INVENTION

[0886] The present invention relates to a system that recognizes the emotions of a user during a call in real time and determines the possibility of fraud with high accuracy based on that information. Specific embodiments will be described below.

[0887] First, when a user starts a call, the device automatically collects the call's audio data. The audio data is recorded in high quality using the device's built-in microphone. The collected audio data is then processed in real time.

[0888] The collected voice data is then converted into text data in real time using a voice recognition engine. For example, a common voice recognition service such as the Google Speech-to-Text API is used. During this process, the device performs error checking as needed to improve conversion accuracy. The converted text data is also temporarily stored in a buffer.

[0889] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS). Encryption protects the user's privacy.

[0890] The server queues the received text data for analysis using a generative AI model trained on a large amount of historical fraud data, which analyzes keywords and context within the text data to output a numerical score indicating the likelihood of fraud.

[0891] In parallel, the device analyzes the collected voice data using an emotion engine, which analyzes parameters such as tone, speed, and pitch of the voice to assess whether the user is feeling surprise or fear. The results of the emotion analysis are also sent to the server in real time.

[0892] The server integrates and analyzes the fraud likelihood score output by the generative AI model and the evaluation results of the emotion engine. The data is weighted so that the results of the emotion engine contribute to the fraud judgment. The final fraud likelihood is determined based on the integrated analysis results.

[0893] If it is determined that there is a high possibility of fraud, the server will immediately send a notification. The notification will be sent to the user and to pre-defined recipients (e.g., family and friends). Possible notification methods include email, SMS, and push notifications to a dedicated app. The notification will include a summary of the contents of the call and a warning that it may be a fraud.

[0894] Specific examples

[0895] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," the system of this invention works as follows: When the call begins, A's device collects audio and converts it into text in real time using the Google Speech-to-Text API. At the same time, the emotion engine analyzes A's tone and pitch to assess whether he is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the result indicates a high possibility of fraud, a warning notification is sent to A and his family. The family member who receives the notification can immediately contact A and provide support to prevent fraud.

[0896] Prompt Sentence Examples

[0897] "Imagine a scenario where a senior citizen encounters a scam phone call and explain how AI and an emotion engine can help flag the possibility of a scam."

[0898] The above is a specific embodiment of the system of the present invention. This system can detect signs of fraud with higher accuracy and send immediate warnings by incorporating user sentiment analysis.

[0899] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0900] Program processing steps

[0901] Step 1:

[0902] A user initiates a call

[0903] When a user makes or receives a call, the device automatically starts collecting the call audio. The audio data is converted into digital form in real time and temporarily stored in internal memory. The input is the call initiation trigger, and the output is the digital audio data.

[0904] Step 2:

[0905] The device converts the voice data into text data in real time.

[0906] The collected voice data is converted into text data in real time using the device's built-in speech recognition engine (e.g., Google Speech-to-Text API). The voice data is divided into short segments, each of which is input into the speech recognition engine. The output is converted text data. Specifically, the continuous voice stream is segmented at regular intervals and converted sequentially.

[0907] Step 3:

[0908] The device encrypts the text data and sends it to the server.

[0909] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g. HTTPS). The input is the converted text data, and the output is the encrypted data sent to the server. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[0910] Step 4:

[0911] The server receives the text data and adds it to the analysis queue.

[0912] The server adds the received text data to a queue for analysis. The input is the encrypted and transmitted text data, and the output is the queued data waiting to be analyzed. The specific operation is to dequeue the received data and add it to the data store for analysis.

[0913] Step 5:

[0914] The server uses generative AI models to assess the likelihood of fraud

[0915] The server inputs the text data added to the queue into the generative AI model and performs analysis to evaluate the likelihood of fraud. The input is the text data in the queue, and the output is a score representing the likelihood of fraud. Specifically, the server inputs data into the model, extracts keywords and performs contextual analysis, and scores the likelihood of fraud.

[0916] Step 6:

[0917] The device analyzes the user's emotions in real time

[0918] The emotion engine in the device analyzes the voice data and evaluates the user's emotion (surprise, fear, anxiety, etc.). The input is the collected voice data, and the output is the result of the emotion evaluation. Specifically, it analyzes the voice tone, speed, and pitch to quantify the emotional state.

[0919] Step 7:

[0920] The device sends the emotion evaluation results to the server.

[0921] The emotion evaluation results generated by the emotion engine are encrypted and sent to the server using a secure communication protocol. The input is the emotion evaluation results, and the output is the encrypted transmission data. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[0922] Step 8:

[0923] The server performs integrated analysis of emotion data and text data.

[0924] The server combines the fraud score from the generative AI model and the evaluation results from the emotion engine to determine the final likelihood of fraud. The inputs are the fraud score and the emotion evaluation results, and the output is an integrated score. Specifically, the server weights and analyzes both sets of data to calculate the integrated score.

[0925] Step 9:

[0926] Sends notifications when the server is likely to be fraudulent

[0927] If a fraudulent activity is deemed likely, the server sends a notification to the user and pre-defined contacts. The input is the integrated score, and the output is a notification message. Specifically, the server sends a warning message via email, SMS, or push notification to a dedicated app.

[0928] These are the specific processing steps of the program for this system. This process allows for highly accurate detection of signs of fraud, taking into account the user's emotions during the call, and for immediate notification.

[0929] (Application example 2)

[0930] 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."

[0931] Conventional fraud detection systems generate text data from a user's voice data and evaluate the likelihood of fraud based on that text data. However, there is a need for a system that can more accurately assess the likelihood of fraud by taking into account not only the user's voice but also their emotional information. Furthermore, prompt notification of the results of the assessment is expected to help prevent fraud, especially for the elderly.

[0932] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0933] In this invention, the server includes a means including an emotion engine that recognizes user emotions from voice data, a means for determining the possibility of fraud by integrating the evaluation results of the generative AI model and the results of the emotion engine, and a means for sending a notification when it is determined that there is a high possibility of fraud. This enables highly accurate fraud detection that takes into account both voice data and emotion data, and a rapid response.

[0934] "User" refers to the individual or organization placing the call.

[0935] A "call" refers to a conversation conducted using voice communication means.

[0936] "Voice data" refers to digital information about a user's voice that is recorded during a telephone call.

[0937] "Means of collection" refers to devices or programs used to obtain voice data from calls.

[0938] "Text data" refers to data that has been converted from voice data into text information.

[0939] "Means for converting" refers to a device or program for converting voice data into text data.

[0940] A "server" refers to a computer system that sends, receives, and processes data over a network.

[0941] "Means for transmitting" refers to means for sending collected or generated data to another device or server.

[0942] A "generative AI model" is an artificial intelligence model that is trained on historical data to perform a specific task (in this case, assessing the likelihood of fraud).

[0943] "Means for assessing the likelihood of fraud" refers to devices or programs that use generative AI models to determine the likelihood of fraud based on text data.

[0944] "Means for sending notification" refers to a device or program that sends an alert to the user or a pre-defined notification destination when a fraudulent activity is deemed likely.

[0945] An "emotion engine" refers to a device or software that analyzes voice data and recognizes the user's emotional state.

[0946] "Means for recognizing emotions" refers to means for extracting a user's emotional state from speech data using an emotion engine.

[0947] "Integrated judgment means" refers to the means of combining the evaluation results from the generative AI model with the emotional data from the emotion engine to ultimately determine the possibility of fraud.

[0948] A system for implementing this invention collects voice data from a call when the user initiates a call, converts it into text data in real time, and transmits the text data to a server. It also includes an emotion engine that recognizes the user's emotions from the voice data and integrates the collected data to assess the likelihood of fraud. If it determines that there is a high likelihood of fraud, it sends a notification, thereby protecting the user from fraud.

[0949] Hardware and software used

[0950] 1. Collection of audio data

[0951] Hardware used: Smartphone microphone

[0952] Software used: speech_recognition library

[0953] 2. Converting voice data to text

[0954] Software used: Google Speech-to-Text API

[0955] It is used to convert voice data into text data in real time, and also performs error checking to improve accuracy.

[0956] 3. Sending text data

[0957] Software used: HTTPS protocol

[0958] The converted text data is encrypted and securely sent to the server.

[0959] 4. Performing fraud analysis

[0960] Software used: Generative AI model

[0961] The server analyzes the received text data and assesses the likelihood of fraud. The generative AI model is trained on a large amount of historical fraud case data.

[0962] 5. Emotion Engine

[0963] Hardware used: Smartphone microphone

[0964] Software used: EmotionRecognizer module

[0965] Recognize user emotions from voice data and provide additional information for fraud analysis.

[0966] 6. Pooled Analysis

[0967] Software used: Server-side integrated processing program

[0968] The evaluation results of the generative AI model are combined with the output of the sentiment engine to determine the final likelihood of fraud.

[0969] 7. Sending Notifications

[0970] Software used: notification_service

[0971] If fraud is deemed likely, a warning notification will be sent to the user and emergency contacts.

[0972] Specific examples

[0973] For example, if an elderly person receives a call from someone claiming to be their "son" saying, "I lost my wallet and need money," the smartphone application collects the audio and converts it into text data using the Google Speech-to-Text API. At the same time, the EmotionRecognizer module analyzes the tone and pitch of the elderly person's voice to detect surprise or anxiety. This data is sent to a server, where a generative AI model analyzes the possibility of fraud. The application also integrates the results of the emotion engine, and if it determines that fraud is likely, it sends a warning notification to the elderly person and their family.

[0974] Example prompt sentence:

[0975] Analyze for potential fraud. Input text: "I'm your son. I lost my wallet and need money. Please help me.". Emotion rating: Fear.

[0976] In this way, the system for implementing the invention achieves highly accurate fraud detection that takes into account both voice data and emotional data, and rapid response.

[0977] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0978] Step 1:

[0979] When a user starts a call, the device uses the microphone to collect the audio data of the call. This audio data is collected in RAW format. The input is the audio data of the call, and the output is an audio data file.

[0980] Step 2:

[0981] The collected voice data is converted to text data in real time using the Google Speech-to-Text API on the device. During this conversion process, the voice data is converted to text data and error checking is performed. The input is a voice data file and the output is text data.

[0982] Step 3:

[0983] The converted text data is sent to the server using the HTTPS protocol. At this time, the text data is encrypted and sent to protect the user's privacy. The input is the text data, and the output is an HTTPS request containing the encrypted text data.

[0984] Step 4:

[0985] The server inputs the received text data into a generative AI model to evaluate the likelihood of fraud. This generative AI model is trained based on past fraud case data and outputs the likelihood of fraud as a numerical score. The input is text data, and the output is a score indicating the likelihood of fraud.

[0986] Step 5:

[0987] While collecting voice data, the device uses the EmotionRecognizer module to recognize the user's emotions from the voice data. This module analyzes the tone, speed, and pitch of the voice to evaluate whether the user is feeling surprise or fear. The input is the voice data, and the output is an emotion evaluation score.

[0988] Step 6:

[0989] The server integrates the fraud likelihood score output by the generative AI model and the emotion evaluation output by the emotion engine. By performing an integrated analysis, it determines the final fraud likelihood. The input is the fraud likelihood score and the emotion evaluation score, and the output is the integrated fraud likelihood score.

[0990] Step 7:

[0991] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Notification methods include email, SMS, and push notifications on a dedicated app. The input is the integrated fraud probability score, and the output is a warning notification.

[0992] 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.

[0993] 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.

[0994] 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.

[0995] [Fourth embodiment]

[0996] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0997] 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.

[0998] 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).

[0999] 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.

[1000] 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.

[1001] 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).

[1002] 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.

[1003] 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.

[1004] 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.

[1005] 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.

[1006] 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.

[1007] 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.

[1008] 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."

[1009] The present invention relates to a system that accurately detects potential fraud while a user is on a call and prevents damage before it occurs. The system includes a user terminal, a server, a generative AI model, and a notification means.

[1010] 1. Audio data collection:

[1011] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[1012] 2. Audio to text conversion:

[1013] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[1014] 3. Sending text data:

[1015] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[1016] 4. Performing fraud analysis:

[1017] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[1018] 5. Sending notifications:

[1019] If the server determines that a call is likely to be fraudulent based on the analysis results of the generative AI model, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be fraudulent.

[1020] Examples:

[1021] For example, suppose elderly person A receives a call from someone claiming to be his "son" who says, "I lost my wallet and need money." When the call begins, A's device collects the audio and converts it into text data. This text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud and determines that there is a high possibility of fraud. Based on the results, the server sends a warning notification to A and his family. Upon receiving the notification, A's family can immediately contact A and provide support to prevent the victim from falling victim to fraud.

[1022] In this way, the system of the present invention can detect signs of fraud in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[1023] The processing flow will be explained below.

[1024] Step 1:

[1025] When a user initiates a call using a phone or voice call app, the device automatically begins collecting the call's audio data. The collection occurs in real time, and the user does not need to take any special action.

[1026] Step 2:

[1027] The device inputs the collected voice data into a speech recognition engine in real time. This speech recognition engine converts the voice data into text data using an external service, such as the Google Speech-to-Text API. The converted text data is temporarily stored in the device's memory.

[1028] Step 3:

[1029] The device checks the converted text data for errors, and if any imperfections that affect recognition accuracy are found, the device runs the speech recognition again to improve the accuracy of the text data.

[1030] Step 4:

[1031] The terminal encrypts the text data. This encryption process is necessary to protect the user's privacy, and uses a commonly used encryption algorithm (e.g., AES).

[1032] Step 5:

[1033] The encrypted text data is sent to the server using a secure communications protocol (e.g., HTTPS), with appropriate procedures taken to ensure data integrity and privacy.

[1034] Step 6:

[1035] The server adds the received text data to a queue for analysis, which allows the server to process large amounts of data efficiently.

[1036] Step 7:

[1037] The server takes the text data and inputs it into a generative AI model, which has been pre-trained on a large amount of past fraud case data and is able to assess the likelihood of fraud with a high degree of accuracy.

[1038] Step 8:

[1039] The generative AI model analyzes input text data to detect signs of fraud. The model identifies keywords and context specific to fraud and outputs the results as a numerical score.

[1040] Step 9:

[1041] The server determines whether the likelihood of fraud is high or low based on the numerical score output by the generative AI model. A specific threshold is set, and if the threshold is exceeded, it is determined that there is a "high likelihood of fraud."

[1042] Step 10:

[1043] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined notification recipients via email, SMS, or push notification on a dedicated app.

[1044] Step 11:

[1045] The user or the person who received the notification will check the contents of the notification. Based on the notification, the user can take measures such as double-checking with the other party or reporting the call to the police, thereby preventing fraud from occurring.

[1046] Example 1

[1047] 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."

[1048] In recent years, fraud methods targeting the elderly have become increasingly sophisticated, with the number of fraudulent incidents occurring during phone calls increasing. Therefore, there is a need for a system that allows users to identify potential fraud in real time and prevent it before it occurs. However, existing systems have difficulty in detecting fraud with high accuracy and providing prompt notification, which leaves elderly people and their families unable to take effective countermeasures.

[1049] 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.

[1050] In this invention, the server includes means for collecting voice data, means for converting the voice data into text data in real time, means for transmitting the converted text data to the server, means for adding the text data received by the server to an analysis queue and evaluating the possibility of fraud using a generative AI model, and means for sending a notification when it is determined that there is a high possibility of fraud. This allows users to accurately detect signs of fraud during a call and receive a prompt warning.

[1051] A "means" is a component of an apparatus, method, or system for achieving a particular function.

[1052] "Voice data" refers to data that is a digital recording of a user's voice during a call.

[1053] "Real-time" refers to processing that occurs almost simultaneously with little delay.

[1054] "Text data" is character information converted from voice data.

[1055] A "server" is a computer system that processes and stores data over a network.

[1056] An "analysis queue" is a data structure for sequentially managing data and tasks waiting to be processed.

[1057] A "generative AI model" is a machine learning model that is trained to perform a specific task using artificial intelligence techniques.

[1058] "Probability of fraud" refers to the probability or risk that the content of a call is fraudulent.

[1059] "Evaluation" refers to making a judgment based on data and information according to specific indicators and criteria.

[1060] A "notification" is a message or signal that notifies a user or a third party of specific information.

[1061] The present invention relates to a system that detects and prevents fraudulent acts during phone calls with high accuracy and in real time. This system is composed of a user terminal, a server, etc. Each component and its operation will be described in detail below.

[1062] 1. Collection of audio data

[1063] When a user initiates a call, the device automatically captures the call's audio data using a high-definition microphone, which records the audio in digital form.

[1064] 2. Converting voice data to text

[1065] The collected voice data is converted into text data in real time by the device's voice recognition engine. The voice recognition engine uses common voice recognition services such as the Google Speech-to-Text API. Error checking is performed as needed to improve conversion accuracy.

[1066] 3. Sending text data

[1067] The converted text data is encrypted to protect the user's privacy and sent to the server using a secure communication protocol (e.g., HTTPS). This communication is done in real time and security measures are in place to prevent data leakage.

[1068] 4. Performing fraud analysis

[1069] The server adds the received text data to an analysis queue. The text data added to the analysis queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and detects specific keywords and contextual information to assess the likelihood of fraud. The analysis results are output as a numerical score, with a higher score indicating a higher likelihood of fraud.

[1070] 5. Sending notifications

[1071] If the call is assessed as likely to be fraudulent, the server will send a notification to the subscriber and pre-defined recipients via email, SMS, or push notification via a dedicated app. The notification will include a summary of the call and a warning that "there is a possibility of fraud."

[1072] Specific examples

[1073] For example, suppose an elderly user receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." In this case, when the call begins, the user's device automatically collects voice data and converts it into text data. This text data is then encrypted and sent to a server. The server then analyzes the call using a generative AI model, detects phrases such as "I lost my wallet" and "I need money," and determines that the call is likely a scam. Based on the results, the server then sends a warning notification to the user and their family. The family member who receives the notification can quickly contact the user and provide support to prevent fraud.

[1074] In this way, the system of the present invention can detect fraudulent activity in real time while the user is on a call and send an immediate warning, thereby preventing fraud, especially targeting the elderly.

[1075] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1076] Step 1:

[1077] When a user initiates a call, the terminal automatically collects the voice data of the call.

[1078] Input: User's voice during a call

[1079] Data processing: Capture audio as digital data using a high-precision microphone

[1080] Output: Digital audio data

[1081] Step 2:

[1082] The collected voice data is converted into text data in real time using the device's voice recognition engine.

[1083] Input: Digital audio data

[1084] Data calculation: Converts voice into text using a voice recognition engine (e.g., Google Speech-to-Text API).

[1085] Output: Text data

[1086] Specific operation: Performs error check and reprocesses if misrecognition is found.

[1087] Step 3:

[1088] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS).

[1089] Input: Text data

[1090] Data processing: Encrypting data using encryption algorithms

[1091] Output: Encrypted text data

[1092] What it does: Sends encrypted data to the server via HTTPS protocol

[1093] Step 4:

[1094] The server adds the received text data to an analysis queue and evaluates the likelihood of fraud using a generative AI model.

[1095] Input: Encrypted text data

[1096] Data processing: Decrypting and adding text data to the analysis queue

[1097] Data calculation: Analyzes data using a generative AI model and calculates a fraud probability score.

[1098] Output: A numerical score indicating the likelihood of fraud.

[1099] What it does: Detects specific keywords and contextual information and makes assessments based on historical fraud data.

[1100] Step 5:

[1101] If the possibility of fraud is assessed to be high, the server will send a notification to the subscriber and to a pre-defined notification address.

[1102] Input: Fraud likelihood score

[1103] Data processing: Generate notification messages and include necessary information

[1104] Output: Information message

[1105] Specific operation: A warning message is sent via email, SMS, or push notification of a dedicated app. The notification includes a summary of the call and a warning message that the call may be fraudulent.

[1106] (Application example 1)

[1107] 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."

[1108] In recent years, there has been an increase in telephone fraud, with elderly people being particularly vulnerable to these attacks. Conventional countermeasures have been limited in their effectiveness in preventing fraud, as they have struggled to detect fraud in real time and issue prompt warnings. Therefore, there is a need for a system that allows users to instantly detect signs of fraud while on the phone and prevent fraud before it occurs.

[1109] 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.

[1110] In this invention, the server includes means for collecting voice data of the call when a user starts a call, means for converting the collected voice data into text data in real time, means for transmitting the converted text data to the server, means for the server to input the received text data into a generative AI model and evaluate the possibility of fraud, means for sending a notification when it is determined that there is a high possibility of fraud, and means for detecting signs of fraud in real time using an application installed on the smartphone and sending a warning, thereby enabling the possibility of fraud to be detected immediately during a call and enabling a prompt response.

[1111] "User" refers to any entity that includes an individual who makes a call using this system.

[1112] "Call voice data" refers to data that is a digital recording of the voice uttered by a user during a call.

[1113] A "collection means" is a combination of hardware and software for capturing audio data during a user's call.

[1114] "Means for converting voice data into text data in real time" refers to the speech recognition engine and associated software used to convert collected voice data into text data on the fly.

[1115] The "transmitting means" is an internet communication means for encrypting the converted text data and transmitting it to a server using a secure communication protocol.

[1116] A "generative AI model" is an artificial intelligence model trained based on a huge amount of past fraud data, and analyzes text data to assess the likelihood of fraud.

[1117] The "assessment means" is a combination of software in which a generative AI model analyzes input text data and outputs a numerical score indicating the likelihood of fraud.

[1118] The "means for sending a notification" is a communication means for sending a warning to the user and a pre-defined notification destination when it is determined that there is a high possibility of fraud.

[1119] "Applications installed on smartphones" are dedicated programs that run on smartphones to execute the functions of this system.

[1120] "Warning method" is a communication method such as email, SMS, or push notification used to notify the user of potential fraud.

[1121] The present invention provides a system for detecting potential fraud with high accuracy while a user is making a call and preventing damage before it occurs. This system includes a user terminal, a server, a generative AI model, and a notification means. Specific embodiments of the system are described below.

[1122] First, when a user starts a call, the device automatically collects voice data. The smartphone's microphone and voice recognition engine (e.g., Google Speech-to-Text API) are used to collect the voice data. The voice recognition engine converts the collected voice data into text data in real time. Appropriate error checking is performed to improve the accuracy of the conversion.

[1123] The converted text data is encrypted to protect user privacy and sent to the server using a secure communication protocol (e.g., HTTPS). The server adds the received text data to an analysis queue and immediately analyzes it using a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and analyzes keywords and contextual information contained in the text data to assess the likelihood of fraud. This assessment is output as a numerical score.

[1124] If the evaluation determines that the call is likely to be fraudulent, the server immediately sends a notification to the user and pre-defined contacts via a dedicated app, SMS, email, etc. The notification will include a summary of the call and a warning that the call may be fraudulent.

[1125] As a concrete example, consider the case where an elderly person receives a call from someone claiming to be their "son" who says, "I lost my wallet and need money." When the call begins, the device collects the audio and converts it into text data using a speech recognition engine. The converted text data is sent to a server and analyzed by a generative AI model. The model detects phrases such as "I lost my wallet" and "I need money" as signs of fraud, and determines that the call is likely to be fraudulent. As a result, the server sends a warning notification to the elderly person and their family. The family member who receives the notification can immediately contact the elderly person and provide support to prevent them from becoming victims of fraud.

[1126] An example of a prompt might be:

[1127] "Please rate the following calls for signs of fraud.

[1128] It said: 'I lost my wallet and need money.'

[1129] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1130] Step 1:

[1131] When a user initiates a call, the device collects the call's audio data: the smartphone's microphone captures the sound and stores the audio data in digital form.

[1132] Input: Audio during a call

[1133] Output: Digital audio data

[1134] Step 2:

[1135] The device converts the collected voice data into text data in real time using a speech recognition engine (such as the Google Speech-to-Text API). Error checking is also performed simultaneously to improve conversion accuracy.

[1136] Input: Digital audio data

[1137] Output: Text data

[1138] Step 3:

[1139] The converted text data is encrypted and sent to the server using a secure communication protocol (HTTPS). The device performs this encryption and transmission.

[1140] Input: Text data

[1141] Output: Encrypted text data sent to the server

[1142] Step 4:

[1143] The server adds the received text data to an analysis queue. A generative AI model then analyzes the text data and assesses the likelihood of fraud. The model is trained on a vast amount of historical fraud data and analyzes keywords and contextual information.

[1144] Input: Encrypted text data

[1145] Output: A numerical score assessing the likelihood of fraud

[1146] Step 5:

[1147] The server checks the analysis results of the generated AI model, and if it determines that there is a high possibility of fraud, it sends a notification to the user via a dedicated application. The notification includes a summary of the call and a warning that "there is a possibility of fraud."

[1148] Input: A numerical score assessing the likelihood of fraud

[1149] Output: Alert notification to the user and the configured contacts

[1150] 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.

[1151] The present invention relates to a system that can recognize a user's emotions during a call and determine the possibility of fraud with higher accuracy based on the emotions. The system includes a user terminal, a server, a generative AI model, an emotion engine, and a notification means.

[1152] 1. Audio data collection:

[1153] When a user initiates a call, the device automatically collects the call's audio data, which is then processed in real time.

[1154] 2. Audio to text conversion:

[1155] The collected voice data is converted into text data in real time using the device's voice recognition engine. The voice recognition engine uses a common voice recognition service, such as the Google Speech-to-Text API. During this process, the device performs appropriate error checking to improve conversion accuracy.

[1156] 3. Sending text data:

[1157] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS), which is designed to protect the user's privacy.

[1158] 4. Performing fraud analysis:

[1159] The server adds the received text data to a queue for analysis. The text data added to the queue is immediately analyzed by a generative AI model. The generative AI model is trained based on a large amount of past fraud case data and is able to assess the likelihood of fraud. The model analyzes keywords and contextual information contained in the text data and outputs a numerical score indicating the likelihood of fraud.

[1160] 5. Running the Emotion Engine:

[1161] While collecting voice data, the device uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the tone, speed, and pitch of the voice to assess whether the user is feeling surprised or scared. The output of the emotion engine is sent to the server in real time.

[1162] 6. Integrated analysis of emotions and text:

[1163] The server combines the fraud likelihood score output by the generative AI model with the emotion evaluation output by the emotion engine to determine the final fraud likelihood. At this time, the emotion engine's results may be weighted to contribute to the fraud determination.

[1164] 7. Sending Notifications:

[1165] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Possible notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning that it may be a fraud.

[1166] Examples:

[1167] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," when the call begins, A's device collects the audio and converts it into text data. At the same time, the emotion engine analyzes A's tone and pitch to assess whether A is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the server combines the results of the emotion engine and determines that there is a high possibility of fraud, it sends a warning notification to A and his family. A's family can immediately contact A and provide support to prevent fraud.

[1168] In this way, by incorporating the user's emotions into the analysis, the system of the present invention can detect signs of fraud with greater accuracy and send immediate warnings, thereby preventing fraud, particularly targeting the elderly.

[1169] The processing flow will be explained below.

[1170] Step 1:

[1171] A user initiates a call using a phone or voice calling app. Once the call is initiated, the device begins collecting audio data for the call. The audio data is collected in real time, without the user having to take any special action.

[1172] Step 2:

[1173] The device inputs the collected voice data into a speech recognition engine, which uses external services such as the Google Speech-to-Text API to convert the voice data into text data in real time.

[1174] Step 3:

[1175] The device checks the converted text data for errors. If any imperfections that affect the accuracy of the text data are found, the device will perform voice recognition again to improve the accuracy of the text data.

[1176] Step 4:

[1177] The device then inputs the voice data into the emotion engine, which analyzes the tone, speed, and pitch of the voice to assess the user's emotions in real time. The assessment results are output as data to identify emotional states such as surprise, fear, and suspicion.

[1178] Step 5:

[1179] The terminal encrypts the converted text data and the output of the emotion engine using a common encryption algorithm (e.g., AES).

[1180] Step 6:

[1181] The encrypted text data and emotion rating data are sent to the server using a secure communication protocol (e.g., HTTPS), ensuring security and privacy.

[1182] Step 7:

[1183] The server adds the received text data and emotion rating data to a queue for analysis, allowing the server to process large amounts of data efficiently.

[1184] Step 8:

[1185] The server inputs the text data into a generative AI model, which is trained on a large amount of historical fraud case data to accurately assess the likelihood of fraud. The model identifies fraud-specific keywords and contexts and outputs the results as a numerical score.

[1186] Step 9:

[1187] The server analyzes the user's emotional state based on the emotion evaluation data output by the emotion engine. If the user is feeling surprise or fear, this information is taken into account in the final fraud determination.

[1188] Step 10:

[1189] The server combines the fraud score output from the generative AI model with the evaluation results of the emotion engine to determine the final likelihood of fraud. If it determines that there is a high likelihood of fraud, it will give a score above a certain threshold.

[1190] Step 11:

[1191] If it is determined that there is a high possibility of fraud, the server will send a warning to the subscriber and pre-defined contacts. Notification methods include email, SMS, and push notifications on a dedicated app. The notification will include a summary of the call and a warning message stating that it is a "possible fraud."

[1192] Step 12:

[1193] The user or the person who received the notification will check the contents of the notification. The user can take measures to prevent fraud by reconfirming with the other party based on the notification or reporting it to the police.

[1194] Example 2

[1195] 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."

[1196] There is a need for a system that can accurately detect possible fraud during a call and quickly notify the user. Fraud targeting elderly people is particularly on the rise, and conventional methods have found it difficult to detect fraud based on changes in emotions. Therefore, a system that can simultaneously analyze the user's emotions and more accurately determine the possibility of fraud is needed.

[1197] 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.

[1198] In this invention, the server includes a means for collecting voice data of a call when the user initiates a call, a means for converting the collected voice data into text data in real time, and a means for encrypting the converted text data and transmitting it to the server using a secure communication protocol. This enables highly accurate and rapid detection of signs of fraud occurring during a call and immediate warnings to be sent to the user. In addition, the server includes a means for inputting the text data and the user's emotion evaluation results into a generative AI model and evaluating the likelihood of fraud. This enables integrated analysis that takes user emotions into account, further improving the accuracy of fraud detection.

[1199] "Voice data" refers to information that is a digital recording of the voice that a user makes during a call.

[1200] "Text data" is information obtained by converting voice data into a character string format.

[1201] "Encryption" is a conversion method to protect data from being deciphered by third parties.

[1202] A "secure communications protocol" is a method of transmitting data securely over the Internet. An example is HTTPS.

[1203] A "server" is a computer system that provides services to other computers over a network.

[1204] A "generative AI model" is an artificial intelligence algorithm that performs specific tasks based on large amounts of training data.

[1205] An "emotion engine" is a software module for analyzing voice data and assessing the user's emotional state.

[1206] A "notification" is a message intended to convey specific information to a user or related third party.

[1207] A "speech recognition engine" is a software application for converting voice data into text data.

[1208] A "queue" is a data structure for managing the order of data processing.

[1209] MODE FOR CARRYING OUT THE INVENTION

[1210] The present invention relates to a system that recognizes the emotions of a user during a call in real time and determines the possibility of fraud with high accuracy based on that information. Specific embodiments will be described below.

[1211] First, when a user starts a call, the device automatically collects the call's audio data. The audio data is recorded in high quality using the device's built-in microphone. The collected audio data is then processed in real time.

[1212] The collected voice data is then converted into text data in real time using a voice recognition engine. For example, a common voice recognition service such as the Google Speech-to-Text API is used. During this process, the device performs error checking as needed to improve conversion accuracy. The converted text data is also temporarily stored in a buffer.

[1213] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g., HTTPS). Encryption protects the user's privacy.

[1214] The server queues the received text data for analysis using a generative AI model trained on a large amount of historical fraud data, which analyzes keywords and context within the text data to output a numerical score indicating the likelihood of fraud.

[1215] In parallel, the device analyzes the collected voice data using an emotion engine, which analyzes parameters such as tone, speed, and pitch of the voice to assess whether the user is feeling surprise or fear. The results of the emotion analysis are also sent to the server in real time.

[1216] The server integrates and analyzes the fraud likelihood score output by the generative AI model and the evaluation results of the emotion engine. The data is weighted so that the results of the emotion engine contribute to the fraud judgment. The final fraud likelihood is determined based on the integrated analysis results.

[1217] If it is determined that there is a high possibility of fraud, the server will immediately send a notification. The notification will be sent to the user and to pre-defined recipients (e.g., family and friends). Possible notification methods include email, SMS, and push notifications to a dedicated app. The notification will include a summary of the contents of the call and a warning that it may be a fraud.

[1218] Specific examples

[1219] For example, if elderly person A receives a call from someone claiming to be his "son" saying, "I lost my wallet and need money," the system of this invention works as follows: When the call begins, A's device collects audio and converts it into text in real time using the Google Speech-to-Text API. At the same time, the emotion engine analyzes A's tone and pitch to assess whether he is feeling surprised or anxious. The text data and emotion evaluation data are sent to the server, and the generative AI model analyzes the possibility of fraud. If the result indicates a high possibility of fraud, a warning notification is sent to A and his family. The family member who receives the notification can immediately contact A and provide support to prevent fraud.

[1220] Prompt Sentence Examples

[1221] "Imagine a scenario where a senior citizen encounters a scam phone call and explain how AI and an emotion engine can help flag the possibility of a scam."

[1222] The above is a specific embodiment of the system of the present invention. This system can detect signs of fraud with higher accuracy and send immediate warnings by incorporating user sentiment analysis.

[1223] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1224] Program processing steps

[1225] Step 1:

[1226] A user initiates a call

[1227] When a user makes or receives a call, the device automatically starts collecting the call audio. The audio data is converted into digital form in real time and temporarily stored in internal memory. The input is the call initiation trigger, and the output is the digital audio data.

[1228] Step 2:

[1229] The device converts the voice data into text data in real time.

[1230] The collected voice data is converted into text data in real time using the device's built-in speech recognition engine (e.g., Google Speech-to-Text API). The voice data is divided into short segments, each of which is input into the speech recognition engine. The output is converted text data. Specifically, the continuous voice stream is segmented at regular intervals and converted sequentially.

[1231] Step 3:

[1232] The device encrypts the text data and sends it to the server.

[1233] The converted text data is encrypted and sent to the server using a secure communication protocol (e.g. HTTPS). The input is the converted text data, and the output is the encrypted data sent to the server. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[1234] Step 4:

[1235] The server receives the text data and adds it to the analysis queue.

[1236] The server adds the received text data to a queue for analysis. The input is the encrypted and transmitted text data, and the output is the queued data waiting to be analyzed. The specific operation is to dequeue the received data and add it to the data store for analysis.

[1237] Step 5:

[1238] The server uses generative AI models to assess the likelihood of fraud

[1239] The server inputs the text data added to the queue into the generative AI model and performs analysis to evaluate the likelihood of fraud. The input is the text data in the queue, and the output is a score representing the likelihood of fraud. Specifically, the server inputs data into the model, extracts keywords and performs contextual analysis, and scores the likelihood of fraud.

[1240] Step 6:

[1241] The device analyzes the user's emotions in real time

[1242] The emotion engine in the device analyzes the voice data and evaluates the user's emotion (surprise, fear, anxiety, etc.). The input is the collected voice data, and the output is the result of the emotion evaluation. Specifically, it analyzes the voice tone, speed, and pitch to quantify the emotional state.

[1243] Step 7:

[1244] The device sends the emotion evaluation results to the server.

[1245] The emotion evaluation results generated by the emotion engine are encrypted and sent to the server using a secure communication protocol. The input is the emotion evaluation results, and the output is the encrypted transmission data. Specifically, the data is protected with an encryption algorithm and sent using the HTTPS protocol.

[1246] Step 8:

[1247] The server performs integrated analysis of emotion data and text data.

[1248] The server combines the fraud score from the generative AI model and the evaluation results from the emotion engine to determine the final likelihood of fraud. The inputs are the fraud score and the emotion evaluation results, and the output is an integrated score. Specifically, the server weights and analyzes both sets of data to calculate the integrated score.

[1249] Step 9:

[1250] Sends notifications when the server is likely to be fraudulent

[1251] If a fraudulent activity is deemed likely, the server sends a notification to the user and pre-defined contacts. The input is the integrated score, and the output is a notification message. Specifically, the server sends a warning message via email, SMS, or push notification to a dedicated app.

[1252] These are the specific processing steps of the program for this system. This process allows for highly accurate detection of signs of fraud, taking into account the user's emotions during the call, and for immediate notification.

[1253] (Application example 2)

[1254] 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."

[1255] Conventional fraud detection systems generate text data from a user's voice data and evaluate the likelihood of fraud based on that text data. However, there is a need for a system that can more accurately assess the likelihood of fraud by taking into account not only the user's voice but also their emotional information. Furthermore, prompt notification of the results of the assessment is expected to help prevent fraud, especially for the elderly.

[1256] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1257] In this invention, the server includes a means including an emotion engine that recognizes user emotions from voice data, a means for determining the possibility of fraud by integrating the evaluation results of the generative AI model and the results of the emotion engine, and a means for sending a notification when it is determined that there is a high possibility of fraud. This enables highly accurate fraud detection that takes into account both voice data and emotion data, and a rapid response.

[1258] "User" refers to the individual or organization placing the call.

[1259] A "call" refers to a conversation conducted using voice communication means.

[1260] "Voice data" refers to digital information about a user's voice that is recorded during a telephone call.

[1261] "Means of collection" refers to devices or programs used to obtain voice data from calls.

[1262] "Text data" refers to data that has been converted from voice data into text information.

[1263] "Means for converting" refers to a device or program for converting voice data into text data.

[1264] A "server" refers to a computer system that sends, receives, and processes data over a network.

[1265] "Means for transmitting" refers to means for sending collected or generated data to another device or server.

[1266] A "generative AI model" is an artificial intelligence model that is trained on historical data to perform a specific task (in this case, assessing the likelihood of fraud).

[1267] "Means for assessing the likelihood of fraud" refers to devices or programs that use generative AI models to determine the likelihood of fraud based on text data.

[1268] "Means for sending notification" refers to a device or program that sends an alert to the user or a pre-defined notification destination when a fraudulent activity is deemed likely.

[1269] An "emotion engine" refers to a device or software that analyzes voice data and recognizes the user's emotional state.

[1270] "Means for recognizing emotions" refers to means for extracting a user's emotional state from speech data using an emotion engine.

[1271] "Integrated judgment means" refers to the means of combining the evaluation results from the generative AI model with the emotional data from the emotion engine to ultimately determine the possibility of fraud.

[1272] A system for implementing this invention collects voice data from a call when the user initiates a call, converts it into text data in real time, and transmits the text data to a server. It also includes an emotion engine that recognizes the user's emotions from the voice data and integrates the collected data to assess the likelihood of fraud. If it determines that there is a high likelihood of fraud, it sends a notification, thereby protecting the user from fraud.

[1273] Hardware and software used

[1274] 1. Collection of audio data

[1275] Hardware used: Smartphone microphone

[1276] Software used: speech_recognition library

[1277] 2. Converting voice data to text

[1278] Software used: Google Speech-to-Text API

[1279] It is used to convert voice data into text data in real time, and also performs error checking to improve accuracy.

[1280] 3. Sending text data

[1281] Software used: HTTPS protocol

[1282] The converted text data is encrypted and securely sent to the server.

[1283] 4. Performing fraud analysis

[1284] Software used: Generative AI model

[1285] The server analyzes the received text data and assesses the likelihood of fraud. The generative AI model is trained on a large amount of historical fraud case data.

[1286] 5. Emotion Engine

[1287] Hardware used: Smartphone microphone

[1288] Software used: EmotionRecognizer module

[1289] Recognize user emotions from voice data and provide additional information for fraud analysis.

[1290] 6. Pooled Analysis

[1291] Software used: Server-side integrated processing program

[1292] The evaluation results of the generative AI model are combined with the output of the sentiment engine to determine the final likelihood of fraud.

[1293] 7. Sending Notifications

[1294] Software used: notification_service

[1295] If fraud is deemed likely, a warning notification will be sent to the user and emergency contacts.

[1296] Specific examples

[1297] For example, if an elderly person receives a call from someone claiming to be their "son" saying, "I lost my wallet and need money," the smartphone application collects the audio and converts it into text data using the Google Speech-to-Text API. At the same time, the EmotionRecognizer module analyzes the tone and pitch of the elderly person's voice to detect surprise or anxiety. This data is sent to a server, where a generative AI model analyzes the possibility of fraud. The application also integrates the results of the emotion engine, and if it determines that fraud is likely, it sends a warning notification to the elderly person and their family.

[1298] Example prompt sentence:

[1299] Analyze for potential fraud. Input text: "I'm your son. I lost my wallet and need money. Please help me.". Emotion rating: Fear.

[1300] In this way, the system for implementing the invention achieves highly accurate fraud detection that takes into account both voice data and emotional data, and rapid response.

[1301] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1302] Step 1:

[1303] When a user starts a call, the device uses the microphone to collect the audio data of the call. This audio data is collected in RAW format. The input is the audio data of the call, and the output is an audio data file.

[1304] Step 2:

[1305] The collected voice data is converted to text data in real time using the Google Speech-to-Text API on the device. During this conversion process, the voice data is converted to text data and error checking is performed. The input is a voice data file and the output is text data.

[1306] Step 3:

[1307] The converted text data is sent to the server using the HTTPS protocol. At this time, the text data is encrypted and sent to protect the user's privacy. The input is the text data, and the output is an HTTPS request containing the encrypted text data.

[1308] Step 4:

[1309] The server inputs the received text data into a generative AI model to evaluate the likelihood of fraud. This generative AI model is trained based on past fraud case data and outputs the likelihood of fraud as a numerical score. The input is text data, and the output is a score indicating the likelihood of fraud.

[1310] Step 5:

[1311] While collecting voice data, the device uses the EmotionRecognizer module to recognize the user's emotions from the voice data. This module analyzes the tone, speed, and pitch of the voice to evaluate whether the user is feeling surprise or fear. The input is the voice data, and the output is an emotion evaluation score.

[1312] Step 6:

[1313] The server integrates the fraud likelihood score output by the generative AI model and the emotion evaluation output by the emotion engine. By performing an integrated analysis, it determines the final fraud likelihood. The input is the fraud likelihood score and the emotion evaluation score, and the output is the integrated fraud likelihood score.

[1314] Step 7:

[1315] Based on the analysis results of the generative AI model and emotion engine, if the server determines that there is a high possibility of fraud, it will send a notification to the subscriber and pre-defined notification recipients. Notification methods include email, SMS, and push notifications on a dedicated app. The input is the integrated fraud probability score, and the output is a warning notification.

[1316] 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.

[1317] 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.

[1318] 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.

[1319] 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.

[1320] 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.

[1321] 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.

[1322] 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).

[1323] 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.

[1324] 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."

[1325] 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.

[1326] 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).

[1327] 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.

[1328] 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.

[1329] 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.

[1330] 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.

[1331] 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.

[1332] 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.

[1333] 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.

[1334] 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.

[1335] 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.

[1336] 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.

[1337] The following is further disclosed regarding the above embodiment.

[1338] (Claim 1)

[1339] means for collecting audio data of a call when the user initiates the call;

[1340] A means for converting the collected voice data into text data in real time;

[1341] means for transmitting the converted text data to a server;

[1342] a means for the server to input the received text data into a generative AI model to assess the likelihood of fraud;

[1343] A means of sending notification if fraud is deemed likely;

[1344] A system including:

[1345] (Claim 2)

[1346] 2. The system according to claim 1, wherein the means for converting voice data into text data in real time uses a voice recognition engine.

[1347] (Claim 3)

[1348] 2. The system according to claim 1, further comprising means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol.

[1349] "Example 1"

[1350] (Claim 1)

[1351] means for collecting audio data of a call when the user initiates the call;

[1352] A means for converting the collected voice data into text data in real time;

[1353] means for transmitting the converted text data to a server;

[1354] A means for the server to add the received text data to an analysis queue and evaluate the likelihood of fraud using a generative AI model;

[1355] A means of sending notification if fraud is deemed likely;

[1356] A system including:

[1357] (Claim 2)

[1358] 2. The system according to claim 1, wherein the means for converting voice data into text data in real time uses a voice recognition engine.

[1359] (Claim 3)

[1360] 2. The system according to claim 1, further comprising means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol.

[1361] "Application Example 1"

[1362] (Claim 1)

[1363] means for collecting audio data of a call when the user initiates the call;

[1364] A means for converting the collected voice data into text data in real time;

[1365] means for transmitting the converted text data to a server;

[1366] a means for the server to input the received text data into a generative AI model to assess the likelihood of fraud;

[1367] A means of sending notification if fraud is deemed likely;

[1368] A means for detecting signs of fraud in real time and sending warnings using an application installed on a smartphone;

[1369] A system including:

[1370] (Claim 2)

[1371] 2. The system according to claim 1, wherein the means for converting voice data into text data in real time uses a voice recognition engine.

[1372] (Claim 3)

[1373] 2. The system according to claim 1, further comprising means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol.

[1374] "Example 2: Combining Emotion Engines"

[1375] (Claim 1)

[1376] means for collecting audio data of a call when the user initiates the call;

[1377] A means for converting the collected voice data into text data in real time;

[1378] means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol;

[1379] a means for the server to input the received text data into a generative AI model to assess the likelihood of fraud;

[1380] means for sending a notification to the user and contacts if fraud is determined to be likely;

[1381] means for analyzing user sentiment in real time during a call;

[1382] A means for integrating the user's emotion assessment into the generative AI model based on the analysis results;

[1383] A means for integrating the user's emotion evaluation and the results of text data analysis to determine the final possibility of fraud;

[1384] A system including:

[1385] (Claim 2)

[1386] 2. The system according to claim 1, wherein the means for converting voice data into text data in real time uses a voice recognition engine.

[1387] (Claim 3)

[1388] 2. The system according to claim 1, further comprising means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol.

[1389] "Application example 2 when combining emotion engines"

[1390] (Claim 1)

[1391] means for collecting audio data of a call when the user initiates the call;

[1392] A means for converting the collected voice data into text data in real time;

[1393] means for transmitting the converted text data to a server;

[1394] a means for the server to input the received text data into a generative AI model to assess the likelihood of fraud;

[1395] A means of sending notification if fraud is deemed likely;

[1396] means including an emotion engine for recognizing a user's emotion from voice data;

[1397] A means of integrating the results of the generative AI model's evaluation with the results of the sentiment engine to determine the likelihood of fraud; and

[1398] A system including:

[1399] (Claim 2)

[1400] 2. The system according to claim 1, wherein the means for converting voice data into text data in real time uses a voice recognition engine.

[1401] (Claim 3)

[1402] 2. The system according to claim 1, further comprising means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol. [Explanation of symbols]

[1403] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for collecting audio data of a call when the user initiates the call; A means for converting the collected voice data into text data in real time; means for transmitting the converted text data to a server; a means for the server to input the received text data into a generative AI model to assess the likelihood of fraud; A means of sending notification if fraud is deemed likely; A system including:

2. 2. The system according to claim 1, wherein the means for converting voice data into text data in real time uses a voice recognition engine.

3. 2. The system according to claim 1, further comprising means for encrypting the converted text data and transmitting the data to a server using a secure communication protocol.

Images