System

Abstract

To provide a system for securing the safety of a user by achieving highly accurate and automated countermeasures against fraud.SOLUTION: The system includes means for receiving an incoming call from a communication device and obtaining phone number information, means for matching the obtained phone number information against existing blacklists and whitelists, means for initiating an automated response using generative artificial intelligence for phone numbers not in the lists, means for the generative artificial intelligence to analyze call content in real-time and detect suspected fraudulent keywords or phrases, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a threshold, means for adding the suspected fraudulent phone number to the blacklist, and means for synchronizing the updated blacklist to all relevant communication devices.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] Telephone fraud, particularly "It's me" fraud, remains a serious social problem, with many people falling victim. Elderly people are particularly vulnerable, and they suffer not only financial losses but also significant psychological distress. To solve this problem, efficient and automated methods for detecting and countering fraudulent phone calls are needed. Conventional filtering methods using blacklists and whitelists of phone numbers are prone to falsely rejecting or accepting calls, making them incomplete solutions. To solve this problem, a system with more advanced analytical and learning capabilities is needed. [Means for solving the problem]

[0005] This invention provides a means for acquiring phone number information when an incoming call is received from a communication device and comparing it with existing blacklists and whitelists. For phone numbers not on the list, generative artificial intelligence is used to automatically answer the call and analyze the call content in real time. Keywords and phrases that may be fraudulent are detected from the analyzed call content, and a risk score is calculated based on the results. If the risk score exceeds a certain threshold, the call is terminated and the phone number in question is added to the blacklist. Furthermore, the updated blacklist is synchronized with all related communication devices, and if the call is evaluated as safe, the phone number is added to the whitelist. These means enable highly accurate, automated fraud prevention and ensure user safety.

[0006] "Communication device" refers to any device used for communication, such as a telephone, smartphone, tablet, or PC.

[0007] An "incoming call" refers to an incoming call from an external device to a communication device.

[0008] "Telephone number information" refers to the caller's telephone number obtained when an incoming call is received from a communication device.

[0009] A "blacklist" is a list of phone numbers suspected of fraud that are recorded and used to block future calls from those numbers.

[0010] A "whitelist" is a list of safe and trusted phone numbers that are recorded and used to prioritize and allow calls from those numbers in the future.

[0011] "Generative artificial intelligence" refers to an AI model that uses natural language processing technology to analyze dialogue content and generate appropriate responses.

[0012] "Automatic response" refers to a system that automatically responds to an incoming call on a communication device without requiring human intervention.

[0013] "Call content" refers to what is said by the caller and receiver during a communication.

[0014] "Analysis" refers to the process by which generative artificial intelligence understands the content of a call and extracts important information and keywords.

[0015] "Suspected fraudulent keywords or phrases" refer to specific words or expressions that are deemed to be highly likely to be fraudulent.

[0016] "Risk score" refers to an indicator that quantifies the possibility of fraud based on the analysis of the content of the call.

[0017] The "threshold" refers to a numerical value that serves as a criterion for determining whether a risk score is a fraudulent call or not.

[0018] "Updating" refers to adding new information or data to a blacklist or whitelist and applying it to all relevant communication devices.

[0019] "Synchronization" refers to the process of matching data and settings between multiple communication devices.

[0020] "Evaluation" refers to the process by which generative artificial intelligence analyzes the content of a call and determines its safety and risk. [Brief explanation of the drawings]

[0021] [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

[0022] 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.

[0023] First, the terms used in the following description will be explained.

[0024] 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).

[0025] 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.

[0026] 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.

[0027] 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.

[0028] 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."

[0029] [First embodiment]

[0030] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0031] 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.

[0032] 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).

[0033] 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.

[0034] 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.

[0035] 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.

[0036] 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.

[0037] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0038] 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.

[0039] 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.

[0040] 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.

[0041] 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."

[0042] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking the telephone carrier's server with the user's communication device.

[0043] System configuration

[0044] 1. Server

[0045] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates the risk score, and updates and synchronizes the list.

[0046] 2. Terminal

[0047] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call contents are sent to the server in real time.

[0048] 3. List

[0049] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[0050] Program processing

[0051] Response when receiving a call

[0052] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[0053] Server: The server checks the phone number against a blacklist and whitelist and issues the appropriate action.

[0054] Launch of AI automated response

[0055] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[0056] Server: The server uses generative artificial intelligence to initiate the call and analyze the call content in real time.

[0057] Real-time analysis and risk assessment

[0058] Server: Generative AI converts the call into text data, detects keywords and phrases that may be fraudulent, and calculates an overall risk score for the call.

[0059] Actions based on risk scores

[0060] Server: If the score exceeds the threshold, the server sends an instruction to the device to terminate the call. It also adds the number to the blacklist and synchronizes the updated information to all devices.

[0061] Terminal: Upon receiving the call termination instruction, the terminal immediately terminates the call and saves the updated blacklist in its local database.

[0062] Update the list after the call ends

[0063] Server: Call numbers that are deemed safe are added to a whitelist.

[0064] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[0065] Specific examples

[0066] Example 1: Automatically detect fraudulent calls and terminate them

[0067] User: I get a call from an unknown number.

[0068] Terminal: Sends the phone number to the server.

[0069] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0070] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0071] Server: Generative AI detects the phrase "please prepare money" and assigns a high risk score.

[0072] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[0073] Terminal: The call is ended and the number is added to the blacklist.

[0074] Example 2: Secure Call Whitelisting Scenario

[0075] User: A call is coming from someone you trust.

[0076] Terminal: Sends the phone number to the server.

[0077] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0078] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0079] Server: Generative AI analyzes the content of the call and determines there is no risk of fraud.

[0080] Server: The call is completed successfully and the phone number is added to the whitelist.

[0081] Device: Download the new whitelist.

[0082] In this way, the system utilizes its advanced analytical capabilities to automate responses to fraud and provide users with a safe and secure communications environment.

[0083] The processing flow will be explained below.

[0084] Step 1:

[0085] User: A call comes in on the user's device.

[0086] Terminal: Detects an incoming call and obtains the caller's phone number information.

[0087] Terminal: Sends the acquired phone number information to the server.

[0088] Step 2:

[0089] Server: Checks the received phone number information against existing blacklists and whitelists.

[0090] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[0091] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[0092] Step 3:

[0093] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[0094] Terminal: Initiates an automatic response and transmits the call contents to the server in real time.

[0095] Step 4:

[0096] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[0097] Server: Detects potentially fraudulent keywords and phrases.

[0098] Step 5:

[0099] Server: Calculates the risk score based on the fraud detection results.

[0100] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[0101] Step 6:

[0102] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[0103] Device: Add the caller number of the ended call to the blacklist.

[0104] Step 7:

[0105] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[0106] On your device: Download the new blacklist and update the local database.

[0107] Step 8:

[0108] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[0109] Server: Updates the whitelist and sends instructions to synchronize to all relevant communication devices.

[0110] Device: Download the new whitelist and update the local database.

[0111] Example 1

[0112] 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."

[0113] In recent years, fraudulent activities via communication devices have been on the rise, increasing the risk that users will inadvertently provide personal information or assets to fraudsters. To prevent such fraudulent activities, a system that can detect suspected fraudulent calls in real time and respond automatically is needed. Conventional systems mainly use blacklists and whitelists, which often result in delayed responses to new fraudulent calls. Additionally, manual analysis of call content has the problem of being unable to respond quickly.

[0114] 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.

[0115] In this invention, the server includes means for detecting an incoming call from a communication terminal and acquiring telephone number information, means for comparing the acquired telephone number information with a blacklist and a whitelist in an internal database, means for initiating an automatic response using generative artificial intelligence for telephone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a set threshold, means for adding the suspected fraudulent telephone number to the blacklist, means for synchronizing the updated blacklist with all related communication terminals, and means for converting the content of the call into text data using voice recognition technology, thereby enabling new and existing fraudulent calls to be effectively detected and dealt with in real time.

[0116] 1. "Communications terminal" means a device used by a user to communicate.

[0117] 2. "Incoming call" refers to a call coming into a communication terminal.

[0118] 3. "Telephone number information" refers to data relating to a telephone number obtained when a call is received.

[0119] 4. "Internal database" refers to the system that manages data stored within the server.

[0120] 5. A "blacklist" is a list in which malicious communications are registered in advance.

[0121] 6. A "whitelist" is a list in which safe communications are registered in advance.

[0122] 7. "Generative AI" means AI that has the ability to analyze natural language and generate responses.

[0123] 8. "Automatic response" refers to the function in which the system automatically responds on behalf of the user in response to certain conditions.

[0124] 9. "Real-time analysis" refers to the process of instantly analyzing the contents of a call.

[0125] 10. "Fraud" refers to malicious conduct intended to deceive others.

[0126] 11. "Keywords and phrases" are words or short phrases that have a specific meaning.

[0127] 12. "Risk Score" is a numerical assessment of the likelihood of fraud.

[0128] 13. "Set threshold" means a numerical value that serves as a benchmark for risk assessment.

[0129] 14. "End a call" means to stop making or receiving calls.

[0130] 15. "Updated Blacklist" means the most recent blacklist with new fraud numbers added.

[0131] 16. "Sync to Device" means to share data with all associated devices.

[0132] 17. "Speech recognition technology" refers to technology that converts speech into text data.

[0133] 18. "Text data" means data converted into character information.

[0134] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking a server and the user's communication device.

[0135] System Hardware and Software

[0136] The server acts as the main processing unit and has the following responsibilities:

[0137] 1. Phone number verification when receiving a call

[0138] 2. Start the automatic response

[0139] 3. Real-time analysis of call content

[0140] 4. Calculating the risk score

[0141] 5. Updating and Syncing Lists

[0142] Specific software used includes database management systems (e.g., MySQL or PostgreSQL) and generative artificial intelligence (e.g., GPT-3), as well as voice recognition technologies such as the Google Speech-to-Text API.

[0143] The user's communication terminal has the following roles:

[0144] 1. Receiving a call

[0145] 2. Start of AI automatic response based on instructions from the server

[0146] 3. Sending call content to the server

[0147] 4. Blacklist and Whitelist Updates

[0148] System operation explanation

[0149] Incoming call detection:

[0150] When the user's communication device detects an incoming call, the device sends the phone number information to the server. For example, the device performs a process to call an API that sends the phone number to the server via a network connection.

[0151] Matching the list:

[0152] The server checks the incoming phone number against an internal database of blacklists and whitelists, using efficient database queries, such as SQL select queries:

[0153] Start Auto-Reply:

[0154] If the number is not on either list, the communication terminal switches to automatic response mode and starts an automatic response upon receiving instructions from the server. At this time, the server uses generative artificial intelligence to generate the response content and start the call.

[0155] Real-time analytics:

[0156] Generative AI converts call content into text data in real time, detects keywords and phrases that may be fraudulent, and uses voice recognition technology (e.g., Google Speech-to-Text API) to convert speech to text.

[0157] Risk score calculation:

[0158] Generative AI calculates a risk score for a call, using specific algorithms based on keywords and phrases that are suspected to be fraudulent.

[0159] End the call and refresh the list:

[0160] If the risk score exceeds a set threshold, the server sends an instruction to the communication device to terminate the call and adds the phone number to a blacklist, and the updated blacklist is synchronized with all related communication devices.

[0161] Adding Secure Calls to the whitelist:

[0162] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the new whitelist to all associated communication terminals.

[0163] Specific examples

[0164] Example 1: Automatic detection and termination of fraudulent calls

[0165] A user receives a call from an unknown number on their communication terminal.

[0166] The terminal sends the phone number to the server.

[0167] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[0168] The terminal starts an automatic response and transmits the contents of the call to the server.

[0169] The server uses generative artificial intelligence to detect the phrase "I need money" and calculates a high risk score.

[0170] Because the risk score exceeds the threshold, the server instructs the call to end.

[0171] The device will end the call and add the phone number to the blacklist.

[0172] Example 2: Adding Secure Calls to the Whitelist

[0173] A user receives a call from a trusted party on their communication terminal.

[0174] The terminal sends the phone number to the server.

[0175] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[0176] The terminal starts an automatic response and transmits the contents of the call to the server.

[0177] The server uses generative artificial intelligence to analyze the content of the call and determine that there is no risk of fraud.

[0178] After the call is successfully completed, the server adds the phone number to the whitelist.

[0179] The device will download the new whitelist.

[0180] Example prompt sentence:

[0181] "I need some money," "This is an urgent call," "There's an investment opportunity."

[0182] In this way, the system leverages powerful generative artificial intelligence and speech recognition technology to provide users with a safe and secure communication experience.

[0183] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0184] Step 1: Detecting an incoming call and obtaining phone number information

[0185] Terminal: The user's communication terminal detects an incoming call. At this point, the terminal obtains the incoming call's phone number information and sends it to the server. The input is the incoming call's phone number, and the output is sending the phone number data to the server. Specifically, the terminal's communication module receives the incoming call signal and sends the phone number as data to the server.

[0186] Step 2: Match list of phone numbers

[0187] Server: Upon receiving phone number information, the server checks the phone number against the blacklist and whitelist in its internal database. The input is the submitted phone number, and the output is the list check result. Specifically, the server searches the blacklist and whitelist using a database query such as SQL to obtain the check result.

[0188] Step 3: Processing the list match results

[0189] Server: Based on the comparison results, if the phone number is on the blacklist, the call is rejected, and if it is on the whitelist, the call is allowed. If it does not match the list, an AI automatic response is initiated. The input is the list comparison result, and the output is instructions for the next process. In terms of specific operation, the internal logic performs conditional branching to select the appropriate process.

[0190] Step 4: Start the AI ​​Automated Response

[0191] Terminal: Upon receiving instructions from the server, the terminal switches to auto-answer mode. At this time, the call is redirected to the server and the generative AI begins to respond. The input is the auto-answer instruction from the server, and the output is the start of the auto-answer. Specifically, the terminal sets up the call redirection, and the AI ​​model begins to generate a response.

[0192] Step 5: Real-time analysis of call content

[0193] Server: Generative AI converts the contents of the call into text data in real time and detects keywords and phrases that may be fraudulent. The input is the audio data of the call, and the output is the text data and detection results. Specifically, speech recognition technology (for example, Google Speech-to-Text API) converts the audio into text, and uses that text to analyze keywords and phrases.

[0194] Step 6: Calculate the risk score

[0195] Server: Calculates fraud risk score based on the analysis results of the call content. The input is the analyzed text data and detected keywords, and the output is the risk score. Specific operations involve quantifying risk using a specific algorithm (e.g., a rule-based scoring system).

[0196] Step 7: End call instructions

[0197] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to end the call. It also adds phone numbers suspected of fraud to a blacklist. The input is the risk score, and the output is an instruction to end the call and an update to the blacklist. Specifically, it makes an API call to issue an end instruction and adds a new blacklist item to the database.

[0198] Step 8: Synchronizing the Blacklist

[0199] Terminal: Receives the call termination instruction from the server, terminates the call, and saves the updated blacklist in the local database. The input is the instruction from the server and the blacklist data, and the output is the end of the call and the updated blacklist. Specific operations include disconnecting the call and downloading and saving the blacklist information locally.

[0200] Step 9: Update and Sync Whitelist

[0201] Server: Adds call numbers that are determined to be free of fraud risk to a whitelist. The new whitelist is then synchronized to all communication devices. The input is the call number, and the output is updating and synchronizing the whitelist. Specific operations include adding the number to the whitelist and sending a notification to distribute it to all devices.

[0202] Step 10: Download the whitelist

[0203] Terminal: Upon receiving notification from the server, downloads the new whitelist and stores it locally. The input is the whitelist update notification from the server, and the output is the locally stored whitelist. The specific operation is to download the new whitelist and store it in the local database.

[0204] (Application example 1)

[0205] 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."

[0206] Currently, countermeasures against fraudulent calls in communication devices are extremely difficult because users must identify signs of fraud themselves. Furthermore, existing systems only check incoming numbers against blacklists or whitelists, and are unable to detect potential fraud in real time and respond immediately. As a result, users are at increased risk of becoming victims of fraud. The objective of this invention is to provide a system that automatically detects fraudulent calls and immediately terminates the call, thereby protecting users from fraud.

[0207] 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.

[0208] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords and phrases that are suspected of being fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a threshold, means for adding phone numbers that are suspected of being fraudulent to the blacklist, means for synchronizing the updated blacklist with all related communication devices, means for detecting and automatically notifying of suspected fraudulent calls in real time, and means for dynamically managing the blacklist and whitelist and synchronizing the latest list information with the communication device. This makes it possible to more quickly and accurately detect fraudulent calls through real-time analysis of call content and dynamic list management, thereby safely protecting users.

[0209] A "communication device" is a device for communicating, and includes mobile devices such as telephones, smartphones, and tablets, as well as landlines.

[0210] "Incoming call" refers to the reception of an external call or message to a communication device.

[0211] "Telephone number information" is data including the telephone number itself and its associated attribute information.

[0212] A "blacklist" is a list of phone numbers that are suspected of fraud or have been found to be involved in fraudulent activities, and when a number is added to this list, it is blocked when it receives a call.

[0213] A "whitelist" is a list of safe phone numbers, and by registering a number on this list, it is given priority when receiving calls.

[0214] "Generative artificial intelligence" is an AI technology that learns on its own based on large data sets and can perform highly accurate analysis and responses to new data, and is particularly good at natural language processing.

[0215] "Automatic response" is a function in which a communication device mechanically responds on behalf of a user.

[0216] "Real-time analysis" is a technology that instantly analyzes the contents of calls and quickly reflects the results.

[0217] "Suspected fraudulent keywords and phrases" are specific words or expressions in the content of a call that suggest fraudulent activity.

[0218] The "risk score" is an evaluation index that quantifies the likelihood of fraud based on the content of the call.

[0219] A "threshold" is a value that indicates a certain standard, and when this value is exceeded, a specific process is executed.

[0220] "Dynamic management" refers to a management method in which the contents of the list are updated in real time or as needed, and the latest information is always synchronized with the entire system.

[0221] To implement the present invention, it is necessary to build a system using a server, a communication device (hereinafter referred to as a "terminal"), and a generative artificial intelligence (hereinafter referred to as a "generative AI model"). Specific embodiments of each function are described below.

[0222] Overview of the entire system

[0223] The system stores blacklists and whitelists on a server, communicates with devices in real time, and processes incoming calls. It uses a generative AI model to analyze the content of calls and detect potential fraud.

[0224] server

[0225] 1. Acquiring and verifying phone numbers

[0226] The system receives incoming phone number information sent from the device and compares it with the blacklist and whitelist stored on the server, instantly identifying known fraud numbers and safe numbers.

[0227] 2. Instructions for starting the automatic response

[0228] Based on the result of the comparison, an instruction is sent to the terminal to initiate an automatic response for numbers that are not on the list.

[0229] 3. Real-time analysis and risk assessment

[0230] Using a generative AI model, the call is transcribed in real time to detect keywords and phrases that may be fraudulent, and a risk score is calculated based on this, instructing the device to immediately terminate the call if a threshold is exceeded.

[0231] 4. Update and sync your list

[0232] Add suspected fraud numbers to a blacklist that is synced in real time to all relevant devices, and add safe calling numbers to a whitelist.

[0233] Terminal

[0234] 1. Receiving a call and sending the number to the server

[0235] When the user's terminal receives an incoming call, the telephone number information is immediately sent to the server.

[0236] 2. Implementing an automatic response

[0237] Upon receiving instructions from the server, the system switches to automatic answering mode, which transmits the contents of the call to the server in real time.

[0238] 3. End the call

[0239] When it receives a call termination instruction from the server, it immediately terminates the call and adds the number to the blacklist, saving the updated list in a local database.

[0240] Generative AI Models

[0241] The generative AI model is used to transcribe the call and analyze it for specific keywords and phrases, such as "please prepare money" which are indicative of fraud, to generate a risk score.

[0242] Specific examples

[0243] Scenario 1: Detecting and terminating fraudulent calls

[0244] 1. The user receives a call from an unknown number.

[0245] 2. The device sends the phone number to the server.

[0246] 3. The server checks the blacklist and whitelist and determines that there is no match.

[0247] 4. The server sends an automatic response instruction to the terminal.

[0248] 5. The call content is analyzed by a generative AI model, and phrases such as "Please prepare some money" are detected.

[0249] 6. Because the risk score exceeds the threshold, the server sends an instruction to end the call to the terminal, and the call ends.

[0250] 7. The number will be added to the blacklist and synced to all your devices.

[0251] Scenario 2: Whitelisting Secure Calls

[0252] 1. You receive a call from a trusted person and the number is not on your list.

[0253] 2. The device sends the phone number to the server.

[0254] 3. The server checks the blacklist and whitelist and determines that there is no match.

[0255] 4. The server sends an automatic response instruction to the terminal.

[0256] 5. The call content is analyzed by a generative AI model and determined to be free of fraud risk.

[0257] 6. The call completes successfully and the number is added to the whitelist.

[0258] Prompt Sentence Examples

[0259] "Please analyze the following call and assess its potential for fraud.

[0260] Text: [insert text of call here]

[0261] Please provide a risk score based on relevant phrases and keywords."

[0262] In this way, the system uses real-time analysis and dynamic list management to quickly and accurately detect fraudulent calls and keep users safe.

[0263] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0264] Step 1:

[0265] The server receives an incoming call from the terminal and obtains the phone number information. The input is the phone number received by the terminal, which is sent to the server. The server receives this phone number information and proceeds to the next step.

[0266] Step 2:

[0267] The server compares the acquired phone number information with existing blacklists and whitelists. The server has blacklists and whitelists stored in a database. The inputs are the received phone number and the blacklist and whitelist, and the comparison results in the output being no match, blacklist, or whitelist.

[0268] Step 3:

[0269] The device switches to automatic response mode based on the server's matching results. In particular, automatic response using a generative AI model is initiated for phone numbers that are not on the list. The server provides response instructions as input, and the device starts automatic response.

[0270] Step 4:

[0271] The generative AI model analyzes the content of the call in real time. The device sends the captured voice data to the server, which then converts it into text data. The input is the voice data of the call, and the output is text data generated based on that data.

[0272] Step 5:

[0273] The server uses the generative AI model to analyze the call content and detect keywords and phrases that may be fraudulent. The input is the generated text data, which is analyzed to extract fraud risk factors. The output is a risk score based on the fraud risk.

[0274] Step 6:

[0275] The server calculates a risk score based on the fraud detection results. The input is data on keywords and phrases that are suspected to be fraudulent, and the risk score is calculated based on this. The output is a specific numerical value for the risk score.

[0276] Step 7:

[0277] If the risk score exceeds the threshold, the server sends an instruction to the terminal to end the call. The inputs are the risk score and threshold information, and the output is an instruction to end the call.

[0278] Step 8:

[0279] The terminal receives a call termination instruction from the server and terminates the call. The input is a call termination instruction from the server, and the output is the end of the call.

[0280] Step 9:

[0281] The server adds suspected fraudulent phone numbers to a blacklist and synchronizes the updated blacklist to all relevant devices. The inputs are fraudulent phone numbers and blacklists, and a new blacklist is created. The output is the new blacklist synchronized to an online database.

[0282] Step 10:

[0283] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the updated whitelist to all devices. The inputs are the phone numbers deemed safe and the whitelist, and a new whitelist is created. The output is the new whitelist synchronized to an online database.

[0284] These are the specific processing steps of this system. This series of steps enables fraudulent calls to be detected in real time and the necessary defensive measures to be taken automatically.

[0285] 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.

[0286] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[0287] System configuration

[0288] 1. Server

[0289] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates a risk score, recognizes the user's emotions through an emotion engine, and updates and synchronizes the list.

[0290] 2. Terminal

[0291] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call content and the user's emotional information are sent to the server in real time.

[0292] 3. List

[0293] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[0294] Program processing

[0295] Response when receiving a call

[0296] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[0297] Server: The server checks the phone number against blacklists and whitelists and directs the appropriate action.

[0298] Launch of AI automated response

[0299] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[0300] Server: The server uses generative artificial intelligence to initiate responses and analyze call content in real time.

[0301] Real-time analysis and risk assessment

[0302] Server: Generative AI converts call content into text data and detects keywords and phrases that may indicate fraud.

[0303] Server: The emotion engine recognizes the user's emotions in real time and analyzes them along with the call content.

[0304] Calculating the risk score

[0305] Server: Calculates a risk score based on the fraud detection results and the emotional information recognized by the emotion engine.

[0306] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[0307] End call and update list

[0308] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[0309] Terminal: If the caller number of the terminated call is to be added to the blacklist, it sends the corresponding information to the server.

[0310] List Synchronization

[0311] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[0312] On your device: Download the new blacklist and update the local database.

[0313] Safe Call Rating and Whitelist Updates

[0314] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[0315] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[0316] Specific examples

[0317] Example 1: Automatically detect fraudulent calls and terminate them

[0318] User: I get a call from an unknown number.

[0319] Terminal: Sends the phone number to the server.

[0320] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0321] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0322] Server: The generative AI detects the phrase "please prepare money," and the emotion engine detects the user's tension and anxiety. It then assigns a high risk score.

[0323] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[0324] Terminal: The call is ended and the number is added to the blacklist.

[0325] Example 2: Secure Call Whitelisting Scenario

[0326] User: A call is coming from someone you trust.

[0327] Terminal: Sends the phone number to the server.

[0328] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0329] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0330] Server: Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. It determines that there is no risk of fraud.

[0331] Server: The call is completed successfully and the phone number is added to the whitelist.

[0332] Device: Download the new whitelist.

[0333] In this way, the system of the present invention combines advanced analytical capabilities with user emotion recognition to automate responses to fraud and provide users with a safe and secure communication environment.

[0334] The processing flow will be explained below.

[0335] Step 1:

[0336] User: A call comes in on the user's device.

[0337] Terminal: Detects an incoming call and obtains the caller's phone number information.

[0338] Terminal: Sends the acquired phone number information to the server.

[0339] Step 2:

[0340] Server: Checks the received phone number information against existing blacklists and whitelists.

[0341] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[0342] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[0343] Step 3:

[0344] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[0345] Terminal: Transmits voice data and user emotion information to the server in real time.

[0346] Step 4:

[0347] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[0348] Server: The emotion engine recognizes the user's emotions in real time and analyzes the emotion information along with the call content.

[0349] Step 5:

[0350] Server: Calculates a risk score based on the results of analyzing the call content and the user's emotion recognition.

[0351] Server: If a keyword or phrase is detected that is suspected to be fraudulent, set a higher risk score.

[0352] Step 6:

[0353] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to terminate the call.

[0354] Server: Adds the phone number to the blacklist and syncs the update to all devices.

[0355] Step 7:

[0356] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[0357] Terminal: Apply the received updated blacklist to the local database.

[0358] Step 8:

[0359] Server: If the call is assessed as safe, add the caller ID to a whitelist and manage it.

[0360] Server: Sends instructions to synchronize whitelist updates to all devices.

[0361] On the device: Download the new whitelist and apply it to the local database.

[0362] Step 9:

[0363] Device: Prepare for the next incoming call by quickly assessing its trustworthiness using the latest blacklists and whitelists.

[0364] Example 2

[0365] 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."

[0366] In modern society, fraudulent activities are becoming increasingly sophisticated, making it difficult to protect personal information and property. It is difficult for users to respond immediately to fraudulent calls made through communication devices, and the mental burden is also great. Therefore, there is a need to provide an environment in which users can use communication devices with peace of mind.

[0367] 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.

[0368] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for recognizing the user's emotions and calculating a risk score along with the fraud detection result, means for terminating the call if the risk score exceeds a threshold, and means for adding phone numbers suspected of being fraudulent to the blacklist and synchronizing the updated blacklist with all related communication devices. This makes it possible to automatically detect calls suspected of being fraudulent and prevent users from becoming victims of fraudulent calls.

[0369] A "communication device" is a device through which a user makes voice calls, and generally includes a mobile phone or a landline phone.

[0370] "Telephone number information" is data indicating the telephone number of the caller, and is used by the communication device to identify the caller.

[0371] A "blacklist" is a list of telephone numbers that have previously been the subject of fraudulent or illicit activity, and is used to prevent fraudulent calls.

[0372] A "whitelist" is a list of safe and trusted phone numbers that is used to accurately classify important calls.

[0373] "Generative artificial intelligence" refers to an algorithm or system that uses AI technology to generate natural, human-like dialogue and has the ability to analyze call content in real time.

[0374] An "emotion engine" is a system that uses voice recognition technology to analyze a user's emotional state and acquires that emotion as data.

[0375] A "risk score" is a quantified indicator of risk calculated based on the content of the call and the user's emotional state, and is used to assess the degree of suspected fraud.

[0376] "Synchronizing" refers to matching data between multiple communication devices, and refers to the process of reflecting updates to the blacklist or whitelist in all communication devices.

[0377] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[0378] System configuration

[0379] server

[0380] The server is the main processing unit that performs real-time analysis of call content, calculates risk scores, recognizes user emotions using an emotion engine, and manages blacklists and whitelists. It performs these processes using generative artificial intelligence (AI) and an emotion engine.

[0381] Terminal

[0382] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The content of the call and the user's emotional information are sent to the server in real time.

[0383] list

[0384] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[0385] Program processing

[0386] Response when receiving a call

[0387] When a user receives a call, the number is sent to the server. The server checks the incoming number against a blacklist and a whitelist and instructs the appropriate action. If the number is not on either list, the device switches to auto-answer mode and the server begins responding using generative artificial intelligence. The content of the call is sent to the server in real time.

[0388] Real-time analysis and risk assessment

[0389] The server uses generative artificial intelligence to convert the call content into text data and detect keywords and phrases that may indicate fraud. At the same time, an emotion engine recognizes the user's emotions in real time and analyzes them along with the call content. A risk score is calculated based on the fraud detection results and the emotional information recognized by the emotion engine. If the risk score exceeds a threshold, the server sends an instruction to end the call to the terminal.

[0390] Ending a call and updating the list

[0391] Upon receiving the call termination instruction, the terminal immediately terminates the call and, if the caller number is to be added to the blacklist, sends corresponding information to the server. The server updates the blacklist and sends a synchronization instruction to all related communication devices. The terminal downloads the new blacklist and updates its local database.

[0392] Safe Call Rating and Whitelist Updates

[0393] If the call is deemed safe, the server adds the caller ID to the whitelist, and the device downloads the new whitelist from the server and prepares for the next incoming call.

[0394] Specific examples

[0395] Example 1: Automatically detect fraudulent calls and terminate them

[0396] The user receives a call from an unknown number. The device sends the phone number to the server. Since the server does not match the list, it instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. The generative AI detects the phrase "please prepare some money" and the emotion engine detects the user's tension or anxiety, resulting in a high risk score. If the score exceeds the threshold, the server sends an instruction to end the call to the device. The device ends the call and adds the number to the blacklist.

[0397] Example 2: Secure Call Whitelisting Scenario

[0398] The user receives a call from a trusted party. The device sends the phone number to the server. Since the call does not match the list, the server instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. If it is determined that there is no risk of fraud, the call is terminated normally and the phone number is added to the whitelist. The device downloads the new whitelist and prepares for the next call.

[0399] Prompt Sentence Examples

[0400] "How does this system detect and block suspected fraudulent calls?"

[0401] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0402] Program processing flow

[0403] Step 1: Incoming call detection and number matching

[0404] Terminal: The user receives a call on their terminal. The phone number is sent to the server.

[0405] Input: Incoming call phone number

[0406] Output: Phone number data sent to the server

[0407] Specific operation: The moment a call comes in, the "+Notification" API is called and the phone number is sent to the server.

[0408] Step 2: List Matching

[0409] Server: Immediately checks the received phone number against blacklists and whitelists.

[0410] Input: Phone number data sent to the server

[0411] Output: List match result (existing in blacklist, existing in whitelist, not in either)

[0412] Specific operation: Rapidly search the incoming number against the blacklist and whitelist in the database.

[0413] Step 3: Start the AI ​​Automated Response

[0414] Server: If the number is not on either list, instruct the device to start an AI automatic response.

[0415] Input: List match result (not in either list)

[0416] Output: Instructions to start AI automatic response sent to the device

[0417] Specific operation: Launches the specified generative artificial intelligence (Generative AI) model and generates a response script.

[0418] Step 4: Start Auto-Answer Mode

[0419] Terminal: Switches to AI automatic response mode upon receiving instructions from the server.

[0420] Input: Instruction to start AI automatic response

[0421] Output: Start auto attendant

[0422] What it does: Loads an audio file for the auto attendant and keeps the call active.

[0423] Step 5: Real-time analysis of call content

[0424] Terminal: Sends voice data during a call to the server in real time.

[0425] Server: Generative AI converts the received speech into text data and detects keywords and phrases that may be fraudulent.

[0426] Input: Voice data during a call

[0427] Output: Real-time text data, detection of potentially fraudulent keywords and phrases

[0428] Specific operation:

[0429] Device: Uses WebRTC to send audio data to the server in streaming format.

[0430] Server: Uses the Speech-to-Text API to convert speech to text in real time and leverages natural language processing (NLP) techniques to perform text analysis.

[0431] Step 6: Real-time sentiment analysis

[0432] Server: Analyzes user emotions in real time using an emotion engine.

[0433] Input: Voice data during a call

[0434] Output: Analysis of the user's emotional state

[0435] Specific operation: The emotion engine analyzes the user's emotions from the voice in real time and generates emotion data.

[0436] Step 7: Risk Score and Call Evaluation

[0437] Server: Calculates risk scores based on the results of text analysis and the emotion engine's recognition data.

[0438] Input: Text analysis results, sentiment analysis results

[0439] Output: Risk score

[0440] Specific operation: Calculates a risk score based on an algorithm that takes into account fraud detection results and emotional information.

[0441] Step 8: End call instructions

[0442] Server: If the risk score exceeds the threshold, sends an instruction to end the call to the terminal.

[0443] Input: Risk score

[0444] Output: End call instruction

[0445] Specific operation: If the risk score exceeds a preset threshold, a call termination instruction is sent to the terminal using a RESTful API.

[0446] Step 9: End the call and update the blacklist

[0447] Terminal: Upon receiving the instruction to end the call, it immediately ends the call and reports the caller ID to the server.

[0448] Server: Add the received caller number to the blacklist and store the relevant information in the database.

[0449] Input: End call instruction, ended caller number

[0450] Output: Updated blacklist

[0451] Specific operation:

[0452] Terminal: Upon receiving a call end instruction, immediately execute the Hangup API.

[0453] Server: Updates the database to add the caller number to the blacklist and immediately block the next incoming call.

[0454] Step 10: Synchronizing the Blacklist

[0455] Server: Sends a synchronization instruction to all associated communication devices to update the blacklist.

[0456] On your device: Download the new blacklist and update the local database.

[0457] Input: Updated blacklist

[0458] Output: Synchronization instructions, updated local database

[0459] Specific operation:

[0460] Server: Notifies the communication device of blacklist updates using WebSocket.

[0461] Terminal: The new blacklist is retrieved via an API call and reflected in the local database.

[0462] Step 11: Update the whitelist

[0463] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[0464] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[0465] Input: Caller ID after call rated safe

[0466] Output: Updated whitelist, updated local database on device

[0467] Specific operation:

[0468] Server: If the call is assessed as safe, add the number to the whitelist and update the database.

[0469] Device: Download the updated whitelist and apply it the next time you receive a call.

[0470] (Application example 2)

[0471] 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."

[0472] Telephone customer service on online shopping sites is an important means of communication for users. However, the number of calls suspected to be fraudulent is increasing, threatening user safety. To solve this problem, a system that automatically detects suspicious calls and allows only safe calls is needed. In addition, more accurate fraud detection is expected by performing a comprehensive risk assessment that takes user emotions into account.

[0473] 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.

[0474] In this invention, the server includes means for receiving calls from communication devices and acquiring phone number information, means for comparing the calls with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results and the user's emotional information recognized by the emotion engine, means for terminating the call if the risk score exceeds a threshold, means for adding the phone number suspected of fraud to the blacklist, means for synchronizing the updated blacklist with all related communication devices, and means for converting the content of the call into text data in real time, thereby ensuring the security of customer service calls and reducing the risk of fraud.

[0475] "Communication device" is a general term for electronic devices used to make calls and send and receive messages.

[0476] An "incoming call" refers to an incoming call or message from an external device.

[0477] "Telephone number information" is a combination of numbers and codes for identifying a specific communication device.

[0478] A "blacklist" is a list of phone numbers suspected of being fraudulent or of unreliable callers.

[0479] A "whitelist" is a list of safe and trusted sources.

[0480] "Generative AI" is AI that has the ability to generate new information and answers based on input data.

[0481] "Auto-answer" is the ability of a communication device to automatically reply without human intervention.

[0482] "Real-time analysis" is the process of analyzing data as soon as it is received.

[0483] "Suspected fraudulent keywords and phrases" refer to specific words or expressions that suggest fraudulent activity.

[0484] A "risk score" is an evaluation value that quantifies the likelihood of fraud.

[0485] An "emotion engine" is software or an algorithm for analyzing and recognizing the emotions of a user of a communication device.

[0486] Converting to "text data" means converting formal data such as audio and video into text information.

[0487] "Updating" means replacing existing information with new information.

[0488] "Synchronization" is the process of matching data between multiple devices.

[0489] An "online shopping site" is a website for buying and selling products and services over the Internet.

[0490] "Customer service" refers to the support activities provided when customers inquire about products or services via communication devices.

[0491] To realize a fraud prevention system for customer service at a mail-order site, the present invention uses a communication device, a server, and necessary programs.

[0492] 1. System Configuration

[0493] server

[0494] The server performs the following main tasks:

[0495] When a call comes in, the phone number information is received and compared against the blacklist and whitelist.

[0496] If the phone number is not on the list, an automated response is initiated using generative artificial intelligence.

[0497] Calls are converted into text data in real time to detect keywords and phrases that may be fraudulent.

[0498] The emotion engine analyzes the user's emotions and calculates a risk score along with the fraud detection results.

[0499] If the risk score exceeds a threshold, the call is terminated and the suspected fraudulent phone number is added to a blacklist.

[0500] The updated blacklist is synchronized to all associated communication devices.

[0501] Communication equipment (user terminal)

[0502] The user's communication device has the following roles:

[0503] An incoming call is received and the telephone number information is sent to the server.

[0504] Receives instructions from the server and switches to auto-response mode.

[0505] The call contents and user emotional information are sent to the server in real time.

[0506] When an instruction to end the call is received, the call is immediately ended.

[0507] Download new blacklists and whitelists from the server and update the local database.

[0508] 2. Program processing explanation

[0509] The programs in the server and communication device operate using the following hardware and software:

[0510] Hardware:

[0511] Server: A computer system equipped with a high-performance processor

[0512] Communication devices: Smartphones and internet-enabled phones

[0513] software:

[0514] Generative AI: Hugging Face Transformer model (e.g. 'gpt-3')

[0515] Emotion Engine: Emotion recognition algorithm (e.g., EmotionRecognizer)

[0516] Speech Recognition: Speech to text conversion using the speech_recognition package

[0517] HTTP communication: Use the requests library to send and receive data

[0518] The server receives the phone number sent from the user's device and compares it with the blacklist and whitelist. If the phone number is not on the list, it uses generative artificial intelligence to initiate an automatic response and converts the call content into text data in real time. It then detects keywords and phrases that may be fraudulent, while an emotion engine simultaneously analyzes the user's emotional information. Based on this information, it calculates a risk score, and if the score exceeds a threshold, it sends an instruction to the user's device to terminate the call.

[0519] 3. Examples of concrete examples and prompts

[0520] Specific examples

[0521] When an unknown number calls the customer service of an online shopping site, the user's device sends the phone number to the server. The server checks the number and, if it is not in the list, it issues an automatic response. During the automatic response, the server analyzes the content of the call and

[0522] The emotion engine reads the user's emotions and calculates a risk score. If there is a high suspicion of fraud, the call is terminated and the number is added to a blacklist.

[0523] Prompt Sentence Examples

[0524] Prompts that pass call content through a generative AI model to detect fraudulent phrases

[0525] User: "Hello, is this customer service from an online shopping site?"

[0526] AI: "Hi, this is customer service. How can I help you?"

[0527] By inputting this prompt into a generative AI model, the call content begins to be analyzed and fraudulent phrases are detected.

[0528] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0529] Step 1:

[0530] A user receives an incoming call and the communication device acquires the phone number information. The communication device (user terminal) sends the received phone number information to the server. The server receives the phone number information as input and prepares to start processing.

[0531] Step 2:

[0532] The server compares the received phone number information with the existing blacklist and whitelist. The input is the phone number information, blacklist, and whitelist, and the output is the comparison result. Based on the comparison result, it checks whether the phone number is in the list.

[0533] Step 3:

[0534] If the matching result shows that the phone number is not on either the blacklist or the whitelist, the server instructs the communication device to switch to automatic response mode. The input is the matching result, and the output is the start of automatic response mode. The communication device receives this instruction and starts an automatic response using generative artificial intelligence.

[0535] Step 4:

[0536] When the automatic answering starts, the communication device sends the contents of the call to the server in real time. The input of the communication device is the voice data of the call contents, and the output is the voice data sent to the server. The server analyzes the received voice data and converts it from voice to text data.

[0537] Step 5:

[0538] The server then analyzes the converted text data using generative artificial intelligence to detect keywords and phrases that may be fraudulent. The input is text data, and the output is the detection results of keywords and phrases that may be fraudulent. This allows the server to identify potential fraudulent activity.

[0539] Step 6:

[0540] Meanwhile, the server uses an emotion engine to analyze the user's emotional information and determine the user's emotional state. The input is voice data and the output is emotional information. The emotional information serves to detect signs of anxiety or tension in the user.

[0541] Step 7:

[0542] The server combines the analysis results of the generative AI with the emotional information from the emotion engine to calculate a risk score. The input is the keyword and phrase detection results and emotional information, and the output is a risk score. This score quantifies the likelihood of fraud.

[0543] Step 8:

[0544] If the risk score exceeds the threshold, the server sends an instruction to the communication device to end the call. The input is the risk score, and the output is an instruction to end the call. The communication device receives this instruction and immediately ends the call.

[0545] Step 9:

[0546] After the call is completed, the server processes the suspected fraudulent phone number by adding it to the blacklist. The input is the phone number information, and the output is the updated blacklist. The server updates the blacklist and synchronizes it to all relevant communication devices.

[0547] Step 10:

[0548] The communication device downloads a new blacklist from the server and updates the local database. The input is the updated blacklist, and the output is the updated local database. This allows for a more robust response to future incoming calls.

[0549] Prompt Sentence Examples

[0550] Here's a prompt that feeds the call content to a generative AI model to detect scam phrases:

[0551] User: "Hello, is this customer service from an online shopping site?"

[0552] AI: "Hi, this is customer service. How can I help you?"

[0553] By inputting this prompt into a generative AI model, analysis of the call content begins and fraudulent phrases are detected.

[0554] 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.

[0555] 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.

[0556] 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.

[0557] [Second embodiment]

[0558] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0559] 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.

[0560] 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).

[0561] 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.

[0562] 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.

[0563] 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).

[0564] 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.

[0565] 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.

[0566] 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.

[0567] 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.

[0568] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0569] 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."

[0570] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking the telephone carrier's server with the user's communication device.

[0571] System configuration

[0572] 1. Server

[0573] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates the risk score, and updates and synchronizes the list.

[0574] 2. Terminal

[0575] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call contents are sent to the server in real time.

[0576] 3. List

[0577] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[0578] Program processing

[0579] Response when receiving a call

[0580] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[0581] Server: The server checks the phone number against a blacklist and whitelist and issues the appropriate action.

[0582] Launch of AI automated response

[0583] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[0584] Server: The server uses generative artificial intelligence to initiate the call and analyze the call content in real time.

[0585] Real-time analysis and risk assessment

[0586] Server: Generative AI converts the call into text data, detects keywords and phrases that may be fraudulent, and calculates an overall risk score for the call.

[0587] Actions based on risk scores

[0588] Server: If the score exceeds the threshold, the server sends an instruction to the device to terminate the call. It also adds the number to the blacklist and synchronizes the updated information to all devices.

[0589] Terminal: Upon receiving the call termination instruction, the terminal immediately terminates the call and saves the updated blacklist in its local database.

[0590] Update the list after the call ends

[0591] Server: Call numbers that are deemed safe are added to a whitelist.

[0592] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[0593] Specific examples

[0594] Example 1: Automatically detect fraudulent calls and terminate them

[0595] User: I get a call from an unknown number.

[0596] Terminal: Sends the phone number to the server.

[0597] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0598] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0599] Server: Generative AI detects the phrase "please prepare money" and assigns a high risk score.

[0600] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[0601] Terminal: The call is ended and the number is added to the blacklist.

[0602] Example 2: Secure Call Whitelisting Scenario

[0603] User: A call is coming from someone you trust.

[0604] Terminal: Sends the phone number to the server.

[0605] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0606] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0607] Server: Generative AI analyzes the content of the call and determines there is no risk of fraud.

[0608] Server: The call is completed successfully and the phone number is added to the whitelist.

[0609] Device: Download the new whitelist.

[0610] In this way, the system utilizes its advanced analytical capabilities to automate responses to fraud and provide users with a safe and secure communications environment.

[0611] The processing flow will be explained below.

[0612] Step 1:

[0613] User: A call comes in on the user's device.

[0614] Terminal: Detects an incoming call and obtains the caller's phone number information.

[0615] Terminal: Sends the acquired phone number information to the server.

[0616] Step 2:

[0617] Server: Checks the received phone number information against existing blacklists and whitelists.

[0618] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[0619] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[0620] Step 3:

[0621] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[0622] Terminal: Initiates an automatic response and transmits the call contents to the server in real time.

[0623] Step 4:

[0624] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[0625] Server: Detects potentially fraudulent keywords and phrases.

[0626] Step 5:

[0627] Server: Calculates the risk score based on the fraud detection results.

[0628] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[0629] Step 6:

[0630] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[0631] Device: Add the caller number of the ended call to the blacklist.

[0632] Step 7:

[0633] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[0634] On your device: Download the new blacklist and update the local database.

[0635] Step 8:

[0636] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[0637] Server: Updates the whitelist and sends instructions to synchronize to all relevant communication devices.

[0638] Device: Download the new whitelist and update the local database.

[0639] Example 1

[0640] 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."

[0641] In recent years, fraudulent activities via communication devices have been on the rise, increasing the risk that users will inadvertently provide personal information or assets to fraudsters. To prevent such fraudulent activities, a system that can detect suspected fraudulent calls in real time and respond automatically is needed. Conventional systems mainly use blacklists and whitelists, which often result in delayed responses to new fraudulent calls. Additionally, manual analysis of call content has the problem of being unable to respond quickly.

[0642] 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.

[0643] In this invention, the server includes means for detecting an incoming call from a communication terminal and acquiring telephone number information, means for comparing the acquired telephone number information with a blacklist and a whitelist in an internal database, means for initiating an automatic response using generative artificial intelligence for telephone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a set threshold, means for adding the suspected fraudulent telephone number to the blacklist, means for synchronizing the updated blacklist with all related communication terminals, and means for converting the content of the call into text data using voice recognition technology, thereby enabling new and existing fraudulent calls to be effectively detected and dealt with in real time.

[0644] 1. "Communications terminal" means a device used by a user to communicate.

[0645] 2. "Incoming call" refers to a call coming into a communication terminal.

[0646] 3. "Telephone number information" refers to data relating to a telephone number obtained when a call is received.

[0647] 4. "Internal database" refers to the system that manages data stored within the server.

[0648] 5. A "blacklist" is a list in which malicious communications are registered in advance.

[0649] 6. A "whitelist" is a list in which safe communications are registered in advance.

[0650] 7. "Generative AI" means AI that has the ability to analyze natural language and generate responses.

[0651] 8. "Automatic response" refers to the function in which the system automatically responds on behalf of the user in response to certain conditions.

[0652] 9. "Real-time analysis" refers to the process of instantly analyzing the contents of a call.

[0653] 10. "Fraud" refers to malicious conduct intended to deceive others.

[0654] 11. "Keywords and phrases" are words or short phrases that have a specific meaning.

[0655] 12. "Risk Score" is a numerical assessment of the likelihood of fraud.

[0656] 13. "Set threshold" means a numerical value that serves as a benchmark for risk assessment.

[0657] 14. "End a call" means to stop making or receiving calls.

[0658] 15. "Updated Blacklist" means the most recent blacklist with new fraud numbers added.

[0659] 16. "Sync to Device" means to share data with all associated devices.

[0660] 17. "Speech recognition technology" refers to technology that converts speech into text data.

[0661] 18. "Text data" means data converted into character information.

[0662] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking a server and the user's communication device.

[0663] System Hardware and Software

[0664] The server acts as the main processing unit and has the following responsibilities:

[0665] 1. Phone number verification when receiving a call

[0666] 2. Start the automatic response

[0667] 3. Real-time analysis of call content

[0668] 4. Calculating the risk score

[0669] 5. Updating and Syncing Lists

[0670] Specific software used includes database management systems (e.g., MySQL or PostgreSQL) and generative artificial intelligence (e.g., GPT-3), as well as voice recognition technologies such as the Google Speech-to-Text API.

[0671] The user's communication terminal has the following roles:

[0672] 1. Receiving a call

[0673] 2. Start of AI automatic response based on instructions from the server

[0674] 3. Sending call content to the server

[0675] 4. Blacklist and Whitelist Updates

[0676] System operation explanation

[0677] Incoming call detection:

[0678] When the user's communication device detects an incoming call, the device sends the phone number information to the server. For example, the device performs a process to call an API that sends the phone number to the server via a network connection.

[0679] Matching the list:

[0680] The server checks the incoming phone number against an internal database of blacklists and whitelists, using efficient database queries, such as SQL select queries:

[0681] Start Auto-Reply:

[0682] If the number is not on either list, the communication terminal switches to automatic response mode and starts an automatic response upon receiving instructions from the server. At this time, the server uses generative artificial intelligence to generate the response content and start the call.

[0683] Real-time analytics:

[0684] Generative AI converts call content into text data in real time, detects keywords and phrases that may be fraudulent, and uses voice recognition technology (e.g., Google Speech-to-Text API) to convert speech to text.

[0685] Risk score calculation:

[0686] Generative AI calculates a risk score for a call, using specific algorithms based on keywords and phrases that are suspected to be fraudulent.

[0687] End the call and refresh the list:

[0688] If the risk score exceeds a set threshold, the server sends an instruction to the communication device to terminate the call and adds the phone number to a blacklist, and the updated blacklist is synchronized with all related communication devices.

[0689] Adding Secure Calls to the whitelist:

[0690] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the new whitelist to all associated communication terminals.

[0691] Specific examples

[0692] Example 1: Automatic detection and termination of fraudulent calls

[0693] A user receives a call from an unknown number on their communication terminal.

[0694] The terminal sends the phone number to the server.

[0695] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[0696] The terminal starts an automatic response and transmits the contents of the call to the server.

[0697] The server uses generative artificial intelligence to detect the phrase "I need money" and calculates a high risk score.

[0698] Because the risk score exceeds the threshold, the server instructs the call to end.

[0699] The device will end the call and add the phone number to the blacklist.

[0700] Example 2: Adding Secure Calls to the Whitelist

[0701] A user receives a call from a trusted party on their communication terminal.

[0702] The terminal sends the phone number to the server.

[0703] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[0704] The terminal starts an automatic response and transmits the contents of the call to the server.

[0705] The server uses generative artificial intelligence to analyze the content of the call and determine that there is no risk of fraud.

[0706] After the call is successfully completed, the server adds the phone number to the whitelist.

[0707] The device will download the new whitelist.

[0708] Example prompt sentence:

[0709] "I need some money," "This is an urgent call," "There's an investment opportunity."

[0710] In this way, the system leverages powerful generative artificial intelligence and speech recognition technology to provide users with a safe and secure communication experience.

[0711] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0712] Step 1: Detecting an incoming call and obtaining phone number information

[0713] Terminal: The user's communication terminal detects an incoming call. At this point, the terminal obtains the incoming call's phone number information and sends it to the server. The input is the incoming call's phone number, and the output is sending the phone number data to the server. Specifically, the terminal's communication module receives the incoming call signal and sends the phone number as data to the server.

[0714] Step 2: Match list of phone numbers

[0715] Server: Upon receiving phone number information, the server checks the phone number against the blacklist and whitelist in its internal database. The input is the submitted phone number, and the output is the list check result. Specifically, the server searches the blacklist and whitelist using a database query such as SQL to obtain the check result.

[0716] Step 3: Processing the list match results

[0717] Server: Based on the comparison results, if the phone number is on the blacklist, the call is rejected, and if it is on the whitelist, the call is allowed. If it does not match the list, an AI automatic response is initiated. The input is the list comparison result, and the output is instructions for the next process. In terms of specific operation, the internal logic performs conditional branching to select the appropriate process.

[0718] Step 4: Start the AI ​​Automated Response

[0719] Terminal: Upon receiving instructions from the server, the terminal switches to auto-answer mode. At this time, the call is redirected to the server and the generative AI begins to respond. The input is the auto-answer instruction from the server, and the output is the start of the auto-answer. Specifically, the terminal sets up the call redirection, and the AI ​​model begins to generate a response.

[0720] Step 5: Real-time analysis of call content

[0721] Server: Generative AI converts the contents of the call into text data in real time and detects keywords and phrases that may be fraudulent. The input is the audio data of the call, and the output is the text data and detection results. Specifically, speech recognition technology (for example, Google Speech-to-Text API) converts the audio into text, and uses that text to analyze keywords and phrases.

[0722] Step 6: Calculate the risk score

[0723] Server: Calculates fraud risk score based on the analysis results of the call content. The input is the analyzed text data and detected keywords, and the output is the risk score. Specific operations involve quantifying risk using a specific algorithm (e.g., a rule-based scoring system).

[0724] Step 7: End call instructions

[0725] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to end the call. It also adds phone numbers suspected of fraud to a blacklist. The input is the risk score, and the output is an instruction to end the call and an update to the blacklist. Specifically, it makes an API call to issue an end instruction and adds a new blacklist item to the database.

[0726] Step 8: Synchronizing the Blacklist

[0727] Terminal: Receives the call termination instruction from the server, terminates the call, and saves the updated blacklist in the local database. The input is the instruction from the server and the blacklist data, and the output is the end of the call and the updated blacklist. Specific operations include disconnecting the call and downloading and saving the blacklist information locally.

[0728] Step 9: Update and Sync Whitelist

[0729] Server: Adds call numbers that are determined to be free of fraud risk to a whitelist. The new whitelist is then synchronized to all communication devices. The input is the call number, and the output is updating and synchronizing the whitelist. Specific operations include adding the number to the whitelist and sending a notification to distribute it to all devices.

[0730] Step 10: Download the whitelist

[0731] Terminal: Upon receiving notification from the server, downloads the new whitelist and stores it locally. The input is the whitelist update notification from the server, and the output is the locally stored whitelist. The specific operation is to download the new whitelist and store it in the local database.

[0732] (Application example 1)

[0733] 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."

[0734] Currently, countermeasures against fraudulent calls in communication devices are extremely difficult because users must identify signs of fraud themselves. Furthermore, existing systems only check incoming numbers against blacklists or whitelists, and are unable to detect potential fraud in real time and respond immediately. As a result, users are at increased risk of becoming victims of fraud. The objective of this invention is to provide a system that automatically detects fraudulent calls and immediately terminates the call, thereby protecting users from fraud.

[0735] 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.

[0736] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords and phrases that are suspected of being fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a threshold, means for adding phone numbers that are suspected of being fraudulent to the blacklist, means for synchronizing the updated blacklist with all related communication devices, means for detecting and automatically notifying of suspected fraudulent calls in real time, and means for dynamically managing the blacklist and whitelist and synchronizing the latest list information with the communication device. This makes it possible to more quickly and accurately detect fraudulent calls through real-time analysis of call content and dynamic list management, thereby safely protecting users.

[0737] A "communication device" is a device for communicating, and includes mobile devices such as telephones, smartphones, and tablets, as well as landlines.

[0738] "Incoming call" refers to the reception of an external call or message to a communication device.

[0739] "Telephone number information" is data including the telephone number itself and its associated attribute information.

[0740] A "blacklist" is a list of phone numbers that are suspected of fraud or have been found to be involved in fraudulent activities, and when a number is added to this list, it is blocked when it receives a call.

[0741] A "whitelist" is a list of safe phone numbers, and by registering a number on this list, it is given priority when receiving calls.

[0742] "Generative artificial intelligence" is an AI technology that learns on its own based on large data sets and can perform highly accurate analysis and responses to new data, and is particularly good at natural language processing.

[0743] "Automatic response" is a function in which a communication device mechanically responds on behalf of a user.

[0744] "Real-time analysis" is a technology that instantly analyzes the contents of calls and quickly reflects the results.

[0745] "Suspected fraudulent keywords and phrases" are specific words or expressions in the content of a call that suggest fraudulent activity.

[0746] The "risk score" is an evaluation index that quantifies the likelihood of fraud based on the content of the call.

[0747] A "threshold" is a value that indicates a certain standard, and when this value is exceeded, a specific process is executed.

[0748] "Dynamic management" refers to a management method in which the contents of the list are updated in real time or as needed, and the latest information is always synchronized with the entire system.

[0749] To implement the present invention, it is necessary to build a system using a server, a communication device (hereinafter referred to as a "terminal"), and a generative artificial intelligence (hereinafter referred to as a "generative AI model"). Specific embodiments of each function are described below.

[0750] Overview of the entire system

[0751] The system stores blacklists and whitelists on a server, communicates with devices in real time, and processes incoming calls. It uses a generative AI model to analyze the content of calls and detect potential fraud.

[0752] server

[0753] 1. Acquiring and verifying phone numbers

[0754] The system receives incoming phone number information sent from the device and compares it with the blacklist and whitelist stored on the server, instantly identifying known fraud numbers and safe numbers.

[0755] 2. Instructions for starting the automatic response

[0756] Based on the result of the comparison, an instruction is sent to the terminal to initiate an automatic response for numbers that are not on the list.

[0757] 3. Real-time analysis and risk assessment

[0758] Using a generative AI model, the call is transcribed in real time to detect keywords and phrases that may be fraudulent, and a risk score is calculated based on this, instructing the device to immediately terminate the call if a threshold is exceeded.

[0759] 4. Update and sync your list

[0760] Add suspected fraud numbers to a blacklist that is synced in real time to all relevant devices, and add safe calling numbers to a whitelist.

[0761] Terminal

[0762] 1. Receiving a call and sending the number to the server

[0763] When the user's terminal receives an incoming call, the telephone number information is immediately sent to the server.

[0764] 2. Implementing an automatic response

[0765] Upon receiving instructions from the server, the system switches to automatic answering mode, which transmits the contents of the call to the server in real time.

[0766] 3. End the call

[0767] When it receives a call termination instruction from the server, it immediately terminates the call and adds the number to the blacklist, saving the updated list in a local database.

[0768] Generative AI Models

[0769] The generative AI model is used to transcribe the call and analyze it for specific keywords and phrases, such as "please prepare money" which are indicative of fraud, to generate a risk score.

[0770] Specific examples

[0771] Scenario 1: Detecting and terminating fraudulent calls

[0772] 1. The user receives a call from an unknown number.

[0773] 2. The device sends the phone number to the server.

[0774] 3. The server checks the blacklist and whitelist and determines that there is no match.

[0775] 4. The server sends an automatic response instruction to the terminal.

[0776] 5. The call content is analyzed by a generative AI model, and phrases such as "Please prepare some money" are detected.

[0777] 6. Because the risk score exceeds the threshold, the server sends an instruction to end the call to the terminal, and the call ends.

[0778] 7. The number will be added to the blacklist and synced to all your devices.

[0779] Scenario 2: Whitelisting Secure Calls

[0780] 1. You receive a call from a trusted person and the number is not on your list.

[0781] 2. The device sends the phone number to the server.

[0782] 3. The server checks the blacklist and whitelist and determines that there is no match.

[0783] 4. The server sends an automatic response instruction to the terminal.

[0784] 5. The call content is analyzed by a generative AI model and determined to be free of fraud risk.

[0785] 6. The call completes successfully and the number is added to the whitelist.

[0786] Prompt Sentence Examples

[0787] "Please analyze the following call and assess its potential for fraud.

[0788] Text: [insert text of call here]

[0789] Please provide a risk score based on relevant phrases and keywords."

[0790] In this way, the system uses real-time analysis and dynamic list management to quickly and accurately detect fraudulent calls and keep users safe.

[0791] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0792] Step 1:

[0793] The server receives an incoming call from the terminal and obtains the phone number information. The input is the phone number received by the terminal, which is sent to the server. The server receives this phone number information and proceeds to the next step.

[0794] Step 2:

[0795] The server compares the acquired phone number information with existing blacklists and whitelists. The server has blacklists and whitelists stored in a database. The inputs are the received phone number and the blacklist and whitelist, and the comparison results in the output being no match, blacklist, or whitelist.

[0796] Step 3:

[0797] The device switches to automatic response mode based on the server's matching results. In particular, automatic response using a generative AI model is initiated for phone numbers that are not on the list. The server provides response instructions as input, and the device starts automatic response.

[0798] Step 4:

[0799] The generative AI model analyzes the content of the call in real time. The device sends the captured voice data to the server, which then converts it into text data. The input is the voice data of the call, and the output is text data generated based on that data.

[0800] Step 5:

[0801] The server uses the generative AI model to analyze the call content and detect keywords and phrases that may be fraudulent. The input is the generated text data, which is analyzed to extract fraud risk factors. The output is a risk score based on the fraud risk.

[0802] Step 6:

[0803] The server calculates a risk score based on the fraud detection results. The input is data on keywords and phrases that are suspected to be fraudulent, and the risk score is calculated based on this. The output is a specific numerical value for the risk score.

[0804] Step 7:

[0805] If the risk score exceeds the threshold, the server sends an instruction to the terminal to end the call. The inputs are the risk score and threshold information, and the output is an instruction to end the call.

[0806] Step 8:

[0807] The terminal receives a call termination instruction from the server and terminates the call. The input is a call termination instruction from the server, and the output is the end of the call.

[0808] Step 9:

[0809] The server adds suspected fraudulent phone numbers to a blacklist and synchronizes the updated blacklist to all relevant devices. The inputs are fraudulent phone numbers and blacklists, and a new blacklist is created. The output is the new blacklist synchronized to an online database.

[0810] Step 10:

[0811] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the updated whitelist to all devices. The inputs are the phone numbers deemed safe and the whitelist, and a new whitelist is created. The output is the new whitelist synchronized to an online database.

[0812] These are the specific processing steps of this system. This series of steps enables fraudulent calls to be detected in real time and the necessary defensive measures to be taken automatically.

[0813] 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.

[0814] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[0815] System configuration

[0816] 1. Server

[0817] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates a risk score, recognizes the user's emotions through an emotion engine, and updates and synchronizes the list.

[0818] 2. Terminal

[0819] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call content and the user's emotional information are sent to the server in real time.

[0820] 3. List

[0821] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[0822] Program processing

[0823] Response when receiving a call

[0824] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[0825] Server: The server checks the phone number against blacklists and whitelists and directs the appropriate action.

[0826] Launch of AI automated response

[0827] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[0828] Server: The server uses generative artificial intelligence to initiate responses and analyze call content in real time.

[0829] Real-time analysis and risk assessment

[0830] Server: Generative AI converts call content into text data and detects keywords and phrases that may indicate fraud.

[0831] Server: The emotion engine recognizes the user's emotions in real time and analyzes them along with the call content.

[0832] Calculating the risk score

[0833] Server: Calculates a risk score based on the fraud detection results and the emotional information recognized by the emotion engine.

[0834] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[0835] End call and update list

[0836] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[0837] Terminal: If the caller number of the terminated call is to be added to the blacklist, it sends the corresponding information to the server.

[0838] List Synchronization

[0839] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[0840] On your device: Download the new blacklist and update the local database.

[0841] Safe Call Rating and Whitelist Updates

[0842] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[0843] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[0844] Specific examples

[0845] Example 1: Automatically detect fraudulent calls and terminate them

[0846] User: I get a call from an unknown number.

[0847] Terminal: Sends the phone number to the server.

[0848] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0849] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0850] Server: The generative AI detects the phrase "please prepare money," and the emotion engine detects the user's tension and anxiety. It then assigns a high risk score.

[0851] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[0852] Terminal: The call is ended and the number is added to the blacklist.

[0853] Example 2: Secure Call Whitelisting Scenario

[0854] User: A call is coming from someone you trust.

[0855] Terminal: Sends the phone number to the server.

[0856] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[0857] Terminal: An automatic response is initiated and the call contents are sent to the server.

[0858] Server: Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. It determines that there is no risk of fraud.

[0859] Server: The call is completed successfully and the phone number is added to the whitelist.

[0860] Device: Download the new whitelist.

[0861] In this way, the system of the present invention combines advanced analytical capabilities with user emotion recognition to automate responses to fraud and provide users with a safe and secure communication environment.

[0862] The processing flow will be explained below.

[0863] Step 1:

[0864] User: A call comes in on the user's device.

[0865] Terminal: Detects an incoming call and obtains the caller's phone number information.

[0866] Terminal: Sends the acquired phone number information to the server.

[0867] Step 2:

[0868] Server: Checks the received phone number information against existing blacklists and whitelists.

[0869] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[0870] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[0871] Step 3:

[0872] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[0873] Terminal: Transmits voice data and user emotion information to the server in real time.

[0874] Step 4:

[0875] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[0876] Server: The emotion engine recognizes the user's emotions in real time and analyzes the emotion information along with the call content.

[0877] Step 5:

[0878] Server: Calculates a risk score based on the results of analyzing the call content and the user's emotion recognition.

[0879] Server: If a keyword or phrase is detected that is suspected to be fraudulent, set a higher risk score.

[0880] Step 6:

[0881] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to terminate the call.

[0882] Server: Adds the phone number to the blacklist and syncs the update to all devices.

[0883] Step 7:

[0884] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[0885] Terminal: Apply the received updated blacklist to the local database.

[0886] Step 8:

[0887] Server: If the call is assessed as safe, add the caller ID to a whitelist and manage it.

[0888] Server: Sends instructions to synchronize whitelist updates to all devices.

[0889] On the device: Download the new whitelist and apply it to the local database.

[0890] Step 9:

[0891] Device: Prepare for the next incoming call by quickly assessing its trustworthiness using the latest blacklists and whitelists.

[0892] Example 2

[0893] 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."

[0894] In modern society, fraudulent activities are becoming increasingly sophisticated, making it difficult to protect personal information and property. It is difficult for users to respond immediately to fraudulent calls made through communication devices, and the mental burden is also great. Therefore, there is a need to provide an environment in which users can use communication devices with peace of mind.

[0895] 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.

[0896] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for recognizing the user's emotions and calculating a risk score along with the fraud detection result, means for terminating the call if the risk score exceeds a threshold, and means for adding phone numbers suspected of being fraudulent to the blacklist and synchronizing the updated blacklist with all related communication devices. This makes it possible to automatically detect calls suspected of being fraudulent and prevent users from becoming victims of fraudulent calls.

[0897] A "communication device" is a device through which a user makes voice calls, and generally includes a mobile phone or a landline phone.

[0898] "Telephone number information" is data indicating the telephone number of the caller, and is used by the communication device to identify the caller.

[0899] A "blacklist" is a list of telephone numbers that have previously been the subject of fraudulent or illicit activity, and is used to prevent fraudulent calls.

[0900] A "whitelist" is a list of safe and trusted phone numbers that is used to accurately classify important calls.

[0901] "Generative artificial intelligence" refers to an algorithm or system that uses AI technology to generate natural, human-like dialogue and has the ability to analyze call content in real time.

[0902] An "emotion engine" is a system that uses voice recognition technology to analyze a user's emotional state and acquires that emotion as data.

[0903] A "risk score" is a quantified indicator of risk calculated based on the content of the call and the user's emotional state, and is used to assess the degree of suspected fraud.

[0904] "Synchronizing" refers to matching data between multiple communication devices, and refers to the process of reflecting updates to the blacklist or whitelist in all communication devices.

[0905] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[0906] System configuration

[0907] server

[0908] The server is the main processing unit that performs real-time analysis of call content, calculates risk scores, recognizes user emotions using an emotion engine, and manages blacklists and whitelists. It performs these processes using generative artificial intelligence (AI) and an emotion engine.

[0909] Terminal

[0910] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The content of the call and the user's emotional information are sent to the server in real time.

[0911] list

[0912] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[0913] Program processing

[0914] Response when receiving a call

[0915] When a user receives a call, the number is sent to the server. The server checks the incoming number against a blacklist and a whitelist and instructs the appropriate action. If the number is not on either list, the device switches to auto-answer mode and the server begins responding using generative artificial intelligence. The content of the call is sent to the server in real time.

[0916] Real-time analysis and risk assessment

[0917] The server uses generative artificial intelligence to convert the call content into text data and detect keywords and phrases that may indicate fraud. At the same time, an emotion engine recognizes the user's emotions in real time and analyzes them along with the call content. A risk score is calculated based on the fraud detection results and the emotional information recognized by the emotion engine. If the risk score exceeds a threshold, the server sends an instruction to end the call to the terminal.

[0918] Ending a call and updating the list

[0919] Upon receiving the call termination instruction, the terminal immediately terminates the call and, if the caller number is to be added to the blacklist, sends corresponding information to the server. The server updates the blacklist and sends a synchronization instruction to all related communication devices. The terminal downloads the new blacklist and updates its local database.

[0920] Safe Call Rating and Whitelist Updates

[0921] If the call is deemed safe, the server adds the caller ID to the whitelist, and the device downloads the new whitelist from the server and prepares for the next incoming call.

[0922] Specific examples

[0923] Example 1: Automatically detect fraudulent calls and terminate them

[0924] The user receives a call from an unknown number. The device sends the phone number to the server. Since the server does not match the list, it instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. The generative AI detects the phrase "please prepare some money" and the emotion engine detects the user's tension or anxiety, resulting in a high risk score. If the score exceeds the threshold, the server sends an instruction to end the call to the device. The device ends the call and adds the number to the blacklist.

[0925] Example 2: Secure Call Whitelisting Scenario

[0926] The user receives a call from a trusted party. The device sends the phone number to the server. Since the call does not match the list, the server instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. If it is determined that there is no risk of fraud, the call is terminated normally and the phone number is added to the whitelist. The device downloads the new whitelist and prepares for the next call.

[0927] Prompt Sentence Examples

[0928] "How does this system detect and block suspected fraudulent calls?"

[0929] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0930] Program processing flow

[0931] Step 1: Incoming call detection and number matching

[0932] Terminal: The user receives a call on their terminal. The phone number is sent to the server.

[0933] Input: Incoming call phone number

[0934] Output: Phone number data sent to the server

[0935] Specific operation: The moment a call comes in, the "+Notification" API is called and the phone number is sent to the server.

[0936] Step 2: List Matching

[0937] Server: Immediately checks the received phone number against blacklists and whitelists.

[0938] Input: Phone number data sent to the server

[0939] Output: List match result (existing in blacklist, existing in whitelist, not in either)

[0940] Specific operation: Rapidly search the incoming number against the blacklist and whitelist in the database.

[0941] Step 3: Start the AI ​​Automated Response

[0942] Server: If the number is not on either list, instruct the device to start an AI automatic response.

[0943] Input: List match result (not in either list)

[0944] Output: Instructions to start AI automatic response sent to the device

[0945] Specific operation: Launches the specified generative artificial intelligence (Generative AI) model and generates a response script.

[0946] Step 4: Start Auto-Answer Mode

[0947] Terminal: Switches to AI automatic response mode upon receiving instructions from the server.

[0948] Input: Instruction to start AI automatic response

[0949] Output: Start auto attendant

[0950] What it does: Loads an audio file for the auto attendant and keeps the call active.

[0951] Step 5: Real-time analysis of call content

[0952] Terminal: Sends voice data during a call to the server in real time.

[0953] Server: Generative AI converts the received speech into text data and detects keywords and phrases that may be fraudulent.

[0954] Input: Voice data during a call

[0955] Output: Real-time text data, detection of potentially fraudulent keywords and phrases

[0956] Specific operation:

[0957] Device: Uses WebRTC to send audio data to the server in streaming format.

[0958] Server: Uses the Speech-to-Text API to convert speech to text in real time and leverages natural language processing (NLP) techniques to perform text analysis.

[0959] Step 6: Real-time sentiment analysis

[0960] Server: Analyzes user emotions in real time using an emotion engine.

[0961] Input: Voice data during a call

[0962] Output: Analysis of the user's emotional state

[0963] Specific operation: The emotion engine analyzes the user's emotions from the voice in real time and generates emotion data.

[0964] Step 7: Risk Score and Call Evaluation

[0965] Server: Calculates risk scores based on the results of text analysis and the emotion engine's recognition data.

[0966] Input: Text analysis results, sentiment analysis results

[0967] Output: Risk score

[0968] Specific operation: Calculates a risk score based on an algorithm that takes into account fraud detection results and emotional information.

[0969] Step 8: End call instructions

[0970] Server: If the risk score exceeds the threshold, sends an instruction to end the call to the terminal.

[0971] Input: Risk score

[0972] Output: End call instruction

[0973] Specific operation: If the risk score exceeds a preset threshold, a call termination instruction is sent to the terminal using a RESTful API.

[0974] Step 9: End the call and update the blacklist

[0975] Terminal: Upon receiving the instruction to end the call, it immediately ends the call and reports the caller ID to the server.

[0976] Server: Add the received caller number to the blacklist and store the relevant information in the database.

[0977] Input: End call instruction, ended caller number

[0978] Output: Updated blacklist

[0979] Specific operation:

[0980] Terminal: Upon receiving a call end instruction, immediately execute the Hangup API.

[0981] Server: Updates the database to add the caller number to the blacklist and immediately block the next incoming call.

[0982] Step 10: Synchronizing the Blacklist

[0983] Server: Sends a synchronization instruction to all associated communication devices to update the blacklist.

[0984] On your device: Download the new blacklist and update the local database.

[0985] Input: Updated blacklist

[0986] Output: Synchronization instructions, updated local database

[0987] Specific operation:

[0988] Server: Notifies the communication device of blacklist updates using WebSocket.

[0989] Terminal: The new blacklist is retrieved via an API call and reflected in the local database.

[0990] Step 11: Update the whitelist

[0991] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[0992] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[0993] Input: Caller ID after call rated safe

[0994] Output: Updated whitelist, updated local database on device

[0995] Specific operation:

[0996] Server: If the call is assessed as safe, add the number to the whitelist and update the database.

[0997] Device: Download the updated whitelist and apply it the next time you receive a call.

[0998] (Application example 2)

[0999] 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."

[1000] Telephone customer service on online shopping sites is an important means of communication for users. However, the number of calls suspected to be fraudulent is increasing, threatening user safety. To solve this problem, a system that automatically detects suspicious calls and allows only safe calls is needed. In addition, more accurate fraud detection is expected by performing a comprehensive risk assessment that takes user emotions into account.

[1001] 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.

[1002] In this invention, the server includes means for receiving calls from communication devices and acquiring phone number information, means for comparing the calls with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results and the user's emotional information recognized by the emotion engine, means for terminating the call if the risk score exceeds a threshold, means for adding the phone number suspected of fraud to the blacklist, means for synchronizing the updated blacklist with all related communication devices, and means for converting the content of the call into text data in real time, thereby ensuring the security of customer service calls and reducing the risk of fraud.

[1003] "Communication device" is a general term for electronic devices used to make calls and send and receive messages.

[1004] An "incoming call" refers to an incoming call or message from an external device.

[1005] "Telephone number information" is a combination of numbers and codes for identifying a specific communication device.

[1006] A "blacklist" is a list of phone numbers suspected of being fraudulent or of unreliable callers.

[1007] A "whitelist" is a list of safe and trusted sources.

[1008] "Generative AI" is AI that has the ability to generate new information and answers based on input data.

[1009] "Auto-answer" is the ability of a communication device to automatically reply without human intervention.

[1010] "Real-time analysis" is the process of analyzing data as soon as it is received.

[1011] "Suspected fraudulent keywords and phrases" refer to specific words or expressions that suggest fraudulent activity.

[1012] A "risk score" is an evaluation value that quantifies the likelihood of fraud.

[1013] An "emotion engine" is software or an algorithm for analyzing and recognizing the emotions of a user of a communication device.

[1014] Converting to "text data" means converting formal data such as audio and video into text information.

[1015] "Updating" means replacing existing information with new information.

[1016] "Synchronization" is the process of matching data between multiple devices.

[1017] An "online shopping site" is a website for buying and selling products and services over the Internet.

[1018] "Customer service" refers to the support activities provided when customers inquire about products or services via communication devices.

[1019] To realize a fraud prevention system for customer service at a mail-order site, the present invention uses a communication device, a server, and necessary programs.

[1020] 1. System Configuration

[1021] server

[1022] The server performs the following main tasks:

[1023] When a call comes in, the phone number information is received and compared against the blacklist and whitelist.

[1024] If the phone number is not on the list, an automated response is initiated using generative artificial intelligence.

[1025] Calls are converted into text data in real time to detect keywords and phrases that may be fraudulent.

[1026] The emotion engine analyzes the user's emotions and calculates a risk score along with the fraud detection results.

[1027] If the risk score exceeds a threshold, the call is terminated and the suspected fraudulent phone number is added to a blacklist.

[1028] The updated blacklist is synchronized to all associated communication devices.

[1029] Communication equipment (user terminal)

[1030] The user's communication device has the following roles:

[1031] An incoming call is received and the telephone number information is sent to the server.

[1032] Receives instructions from the server and switches to auto-response mode.

[1033] The call contents and user emotional information are sent to the server in real time.

[1034] When an instruction to end the call is received, the call is immediately ended.

[1035] Download new blacklists and whitelists from the server and update the local database.

[1036] 2. Program processing explanation

[1037] The programs in the server and communication device operate using the following hardware and software:

[1038] Hardware:

[1039] Server: A computer system equipped with a high-performance processor

[1040] Communication devices: Smartphones and internet-enabled phones

[1041] software:

[1042] Generative AI: Hugging Face Transformer model (e.g. 'gpt-3')

[1043] Emotion Engine: Emotion recognition algorithm (e.g., EmotionRecognizer)

[1044] Speech Recognition: Speech to text conversion using the speech_recognition package

[1045] HTTP communication: Use the requests library to send and receive data

[1046] The server receives the phone number sent from the user's device and compares it with the blacklist and whitelist. If the phone number is not on the list, it uses generative artificial intelligence to initiate an automatic response and converts the call content into text data in real time. It then detects keywords and phrases that may be fraudulent, while an emotion engine simultaneously analyzes the user's emotional information. Based on this information, it calculates a risk score, and if the score exceeds a threshold, it sends an instruction to the user's device to terminate the call.

[1047] 3. Examples of concrete examples and prompts

[1048] Specific examples

[1049] When an unknown number calls the customer service of an online shopping site, the user's device sends the phone number to the server. The server checks the number and, if it is not in the list, it issues an automatic response. During the automatic response, the server analyzes the content of the call and

[1050] The emotion engine reads the user's emotions and calculates a risk score. If there is a high suspicion of fraud, the call is terminated and the number is added to a blacklist.

[1051] Prompt Sentence Examples

[1052] Prompts that pass call content through a generative AI model to detect fraudulent phrases

[1053] User: "Hello, is this customer service from an online shopping site?"

[1054] AI: "Hi, this is customer service. How can I help you?"

[1055] By inputting this prompt into a generative AI model, the call content begins to be analyzed and fraudulent phrases are detected.

[1056] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1057] Step 1:

[1058] A user receives an incoming call and the communication device acquires the phone number information. The communication device (user terminal) sends the received phone number information to the server. The server receives the phone number information as input and prepares to start processing.

[1059] Step 2:

[1060] The server compares the received phone number information with the existing blacklist and whitelist. The input is the phone number information, blacklist, and whitelist, and the output is the comparison result. Based on the comparison result, it checks whether the phone number is in the list.

[1061] Step 3:

[1062] If the matching result shows that the phone number is not on either the blacklist or the whitelist, the server instructs the communication device to switch to automatic response mode. The input is the matching result, and the output is the start of automatic response mode. The communication device receives this instruction and starts an automatic response using generative artificial intelligence.

[1063] Step 4:

[1064] When the automatic answering starts, the communication device sends the contents of the call to the server in real time. The input of the communication device is the voice data of the call contents, and the output is the voice data sent to the server. The server analyzes the received voice data and converts it from voice to text data.

[1065] Step 5:

[1066] The server then analyzes the converted text data using generative artificial intelligence to detect keywords and phrases that may be fraudulent. The input is text data, and the output is the detection results of keywords and phrases that may be fraudulent. This allows the server to identify potential fraudulent activity.

[1067] Step 6:

[1068] Meanwhile, the server uses an emotion engine to analyze the user's emotional information and determine the user's emotional state. The input is voice data and the output is emotional information. The emotional information serves to detect signs of anxiety or tension in the user.

[1069] Step 7:

[1070] The server combines the analysis results of the generative AI with the emotional information from the emotion engine to calculate a risk score. The input is the keyword and phrase detection results and emotional information, and the output is a risk score. This score quantifies the likelihood of fraud.

[1071] Step 8:

[1072] If the risk score exceeds the threshold, the server sends an instruction to the communication device to end the call. The input is the risk score, and the output is an instruction to end the call. The communication device receives this instruction and immediately ends the call.

[1073] Step 9:

[1074] After the call is completed, the server processes the suspected fraudulent phone number by adding it to the blacklist. The input is the phone number information, and the output is the updated blacklist. The server updates the blacklist and synchronizes it to all relevant communication devices.

[1075] Step 10:

[1076] The communication device downloads a new blacklist from the server and updates the local database. The input is the updated blacklist, and the output is the updated local database. This allows for a more robust response to future incoming calls.

[1077] Prompt Sentence Examples

[1078] Here's a prompt that feeds the call content to a generative AI model to detect scam phrases:

[1079] User: "Hello, is this customer service from an online shopping site?"

[1080] AI: "Hi, this is customer service. How can I help you?"

[1081] By inputting this prompt into a generative AI model, analysis of the call content begins and fraudulent phrases are detected.

[1082] 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.

[1083] 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.

[1084] 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.

[1085] [Third embodiment]

[1086] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1087] 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.

[1088] 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).

[1089] 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.

[1090] 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.

[1091] 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).

[1092] 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.

[1093] 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.

[1094] 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.

[1095] 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.

[1096] 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.

[1097] 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."

[1098] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking the telephone carrier's server with the user's communication device.

[1099] System configuration

[1100] 1. Server

[1101] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates the risk score, and updates and synchronizes the list.

[1102] 2. Terminal

[1103] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call contents are sent to the server in real time.

[1104] 3. List

[1105] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[1106] Program processing

[1107] Response when receiving a call

[1108] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[1109] Server: The server checks the phone number against a blacklist and whitelist and issues the appropriate action.

[1110] Launch of AI automated response

[1111] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[1112] Server: The server uses generative artificial intelligence to initiate the call and analyze the call content in real time.

[1113] Real-time analysis and risk assessment

[1114] Server: Generative AI converts the call into text data, detects keywords and phrases that may be fraudulent, and calculates an overall risk score for the call.

[1115] Actions based on risk scores

[1116] Server: If the score exceeds the threshold, the server sends an instruction to the device to terminate the call. It also adds the number to the blacklist and synchronizes the updated information to all devices.

[1117] Terminal: Upon receiving the call termination instruction, the terminal immediately terminates the call and saves the updated blacklist in its local database.

[1118] Update the list after the call ends

[1119] Server: Call numbers that are deemed safe are added to a whitelist.

[1120] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[1121] Specific examples

[1122] Example 1: Automatically detect fraudulent calls and terminate them

[1123] User: I get a call from an unknown number.

[1124] Terminal: Sends the phone number to the server.

[1125] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1126] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1127] Server: Generative AI detects the phrase "please prepare money" and assigns a high risk score.

[1128] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[1129] Terminal: The call is ended and the number is added to the blacklist.

[1130] Example 2: Secure Call Whitelisting Scenario

[1131] User: A call is coming from someone you trust.

[1132] Terminal: Sends the phone number to the server.

[1133] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1134] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1135] Server: Generative AI analyzes the content of the call and determines there is no risk of fraud.

[1136] Server: The call is completed successfully and the phone number is added to the whitelist.

[1137] Device: Download the new whitelist.

[1138] In this way, the system utilizes its advanced analytical capabilities to automate responses to fraud and provide users with a safe and secure communications environment.

[1139] The processing flow will be explained below.

[1140] Step 1:

[1141] User: A call comes in on the user's device.

[1142] Terminal: Detects an incoming call and obtains the caller's phone number information.

[1143] Terminal: Sends the acquired phone number information to the server.

[1144] Step 2:

[1145] Server: Checks the received phone number information against existing blacklists and whitelists.

[1146] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[1147] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[1148] Step 3:

[1149] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[1150] Terminal: Initiates an automatic response and transmits the call contents to the server in real time.

[1151] Step 4:

[1152] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[1153] Server: Detects potentially fraudulent keywords and phrases.

[1154] Step 5:

[1155] Server: Calculates the risk score based on the fraud detection results.

[1156] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[1157] Step 6:

[1158] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[1159] Device: Add the caller number of the ended call to the blacklist.

[1160] Step 7:

[1161] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[1162] On your device: Download the new blacklist and update the local database.

[1163] Step 8:

[1164] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[1165] Server: Updates the whitelist and sends instructions to synchronize to all relevant communication devices.

[1166] Device: Download the new whitelist and update the local database.

[1167] Example 1

[1168] 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."

[1169] In recent years, fraudulent activities via communication devices have been on the rise, increasing the risk that users will inadvertently provide personal information or assets to fraudsters. To prevent such fraudulent activities, a system that can detect suspected fraudulent calls in real time and respond automatically is needed. Conventional systems mainly use blacklists and whitelists, which often result in delayed responses to new fraudulent calls. Additionally, manual analysis of call content has the problem of being unable to respond quickly.

[1170] 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.

[1171] In this invention, the server includes means for detecting an incoming call from a communication terminal and acquiring telephone number information, means for comparing the acquired telephone number information with a blacklist and a whitelist in an internal database, means for initiating an automatic response using generative artificial intelligence for telephone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a set threshold, means for adding the suspected fraudulent telephone number to the blacklist, means for synchronizing the updated blacklist with all related communication terminals, and means for converting the content of the call into text data using voice recognition technology, thereby enabling new and existing fraudulent calls to be effectively detected and dealt with in real time.

[1172] 1. "Communications terminal" means a device used by a user to communicate.

[1173] 2. "Incoming call" refers to a call coming into a communication terminal.

[1174] 3. "Telephone number information" refers to data relating to a telephone number obtained when a call is received.

[1175] 4. "Internal database" refers to the system that manages data stored within the server.

[1176] 5. A "blacklist" is a list in which malicious communications are registered in advance.

[1177] 6. A "whitelist" is a list in which safe communications are registered in advance.

[1178] 7. "Generative AI" means AI that has the ability to analyze natural language and generate responses.

[1179] 8. "Automatic response" refers to the function in which the system automatically responds on behalf of the user in response to certain conditions.

[1180] 9. "Real-time analysis" refers to the process of instantly analyzing the contents of a call.

[1181] 10. "Fraud" refers to malicious conduct intended to deceive others.

[1182] 11. "Keywords and phrases" are words or short phrases that have a specific meaning.

[1183] 12. "Risk Score" is a numerical assessment of the likelihood of fraud.

[1184] 13. "Set threshold" means a numerical value that serves as a benchmark for risk assessment.

[1185] 14. "End a call" means to stop making or receiving calls.

[1186] 15. "Updated Blacklist" means the most recent blacklist with new fraud numbers added.

[1187] 16. "Sync to Device" means to share data with all associated devices.

[1188] 17. "Speech recognition technology" refers to technology that converts speech into text data.

[1189] 18. "Text data" means data converted into character information.

[1190] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking a server and the user's communication device.

[1191] System Hardware and Software

[1192] The server acts as the main processing unit and has the following responsibilities:

[1193] 1. Phone number verification when receiving a call

[1194] 2. Start the automatic response

[1195] 3. Real-time analysis of call content

[1196] 4. Calculating the risk score

[1197] 5. Updating and Syncing Lists

[1198] Specific software used includes database management systems (e.g., MySQL or PostgreSQL) and generative artificial intelligence (e.g., GPT-3), as well as voice recognition technologies such as the Google Speech-to-Text API.

[1199] The user's communication terminal has the following roles:

[1200] 1. Receiving a call

[1201] 2. Start of AI automatic response based on instructions from the server

[1202] 3. Sending call content to the server

[1203] 4. Blacklist and Whitelist Updates

[1204] System operation explanation

[1205] Incoming call detection:

[1206] When the user's communication device detects an incoming call, the device sends the phone number information to the server. For example, the device performs a process to call an API that sends the phone number to the server via a network connection.

[1207] Matching the list:

[1208] The server checks the incoming phone number against an internal database of blacklists and whitelists, using efficient database queries, such as SQL select queries:

[1209] Start Auto-Reply:

[1210] If the number is not on either list, the communication terminal switches to automatic response mode and starts an automatic response upon receiving instructions from the server. At this time, the server uses generative artificial intelligence to generate the response content and start the call.

[1211] Real-time analytics:

[1212] Generative AI converts call content into text data in real time, detects keywords and phrases that may be fraudulent, and uses voice recognition technology (e.g., Google Speech-to-Text API) to convert speech to text.

[1213] Risk score calculation:

[1214] Generative AI calculates a risk score for a call, using specific algorithms based on keywords and phrases that are suspected to be fraudulent.

[1215] End the call and refresh the list:

[1216] If the risk score exceeds a set threshold, the server sends an instruction to the communication device to terminate the call and adds the phone number to a blacklist, and the updated blacklist is synchronized with all related communication devices.

[1217] Adding Secure Calls to the whitelist:

[1218] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the new whitelist to all associated communication terminals.

[1219] Specific examples

[1220] Example 1: Automatic detection and termination of fraudulent calls

[1221] A user receives a call from an unknown number on their communication terminal.

[1222] The terminal sends the phone number to the server.

[1223] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[1224] The terminal starts an automatic response and transmits the contents of the call to the server.

[1225] The server uses generative artificial intelligence to detect the phrase "I need money" and calculates a high risk score.

[1226] Because the risk score exceeds the threshold, the server instructs the call to end.

[1227] The device will end the call and add the phone number to the blacklist.

[1228] Example 2: Adding Secure Calls to the Whitelist

[1229] A user receives a call from a trusted party on their communication terminal.

[1230] The terminal sends the phone number to the server.

[1231] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[1232] The terminal starts an automatic response and transmits the contents of the call to the server.

[1233] The server uses generative artificial intelligence to analyze the content of the call and determine that there is no risk of fraud.

[1234] After the call is successfully completed, the server adds the phone number to the whitelist.

[1235] The device will download the new whitelist.

[1236] Example prompt sentence:

[1237] "I need some money," "This is an urgent call," "There's an investment opportunity."

[1238] In this way, the system leverages powerful generative artificial intelligence and speech recognition technology to provide users with a safe and secure communication experience.

[1239] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1240] Step 1: Detecting an incoming call and obtaining phone number information

[1241] Terminal: The user's communication terminal detects an incoming call. At this point, the terminal obtains the incoming call's phone number information and sends it to the server. The input is the incoming call's phone number, and the output is sending the phone number data to the server. Specifically, the terminal's communication module receives the incoming call signal and sends the phone number as data to the server.

[1242] Step 2: Match list of phone numbers

[1243] Server: Upon receiving phone number information, the server checks the phone number against the blacklist and whitelist in its internal database. The input is the submitted phone number, and the output is the list check result. Specifically, the server searches the blacklist and whitelist using a database query such as SQL to obtain the check result.

[1244] Step 3: Processing the list match results

[1245] Server: Based on the comparison results, if the phone number is on the blacklist, the call is rejected, and if it is on the whitelist, the call is allowed. If it does not match the list, an AI automatic response is initiated. The input is the list comparison result, and the output is instructions for the next process. In terms of specific operation, the internal logic performs conditional branching to select the appropriate process.

[1246] Step 4: Start the AI ​​Automated Response

[1247] Terminal: Upon receiving instructions from the server, the terminal switches to auto-answer mode. At this time, the call is redirected to the server and the generative AI begins to respond. The input is the auto-answer instruction from the server, and the output is the start of the auto-answer. Specifically, the terminal sets up the call redirection, and the AI ​​model begins to generate a response.

[1248] Step 5: Real-time analysis of call content

[1249] Server: Generative AI converts the contents of the call into text data in real time and detects keywords and phrases that may be fraudulent. The input is the audio data of the call, and the output is the text data and detection results. Specifically, speech recognition technology (for example, Google Speech-to-Text API) converts the audio into text, and uses that text to analyze keywords and phrases.

[1250] Step 6: Calculate the risk score

[1251] Server: Calculates fraud risk score based on the analysis results of the call content. The input is the analyzed text data and detected keywords, and the output is the risk score. Specific operations involve quantifying risk using a specific algorithm (e.g., a rule-based scoring system).

[1252] Step 7: End call instructions

[1253] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to end the call. It also adds phone numbers suspected of fraud to a blacklist. The input is the risk score, and the output is an instruction to end the call and an update to the blacklist. Specifically, it makes an API call to issue an end instruction and adds a new blacklist item to the database.

[1254] Step 8: Synchronizing the Blacklist

[1255] Terminal: Receives the call termination instruction from the server, terminates the call, and saves the updated blacklist in the local database. The input is the instruction from the server and the blacklist data, and the output is the end of the call and the updated blacklist. Specific operations include disconnecting the call and downloading and saving the blacklist information locally.

[1256] Step 9: Update and Sync Whitelist

[1257] Server: Adds call numbers that are determined to be free of fraud risk to a whitelist. The new whitelist is then synchronized to all communication devices. The input is the call number, and the output is updating and synchronizing the whitelist. Specific operations include adding the number to the whitelist and sending a notification to distribute it to all devices.

[1258] Step 10: Download the whitelist

[1259] Terminal: Upon receiving notification from the server, downloads the new whitelist and stores it locally. The input is the whitelist update notification from the server, and the output is the locally stored whitelist. The specific operation is to download the new whitelist and store it in the local database.

[1260] (Application example 1)

[1261] 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."

[1262] Currently, countermeasures against fraudulent calls in communication devices are extremely difficult because users must identify signs of fraud themselves. Furthermore, existing systems only check incoming numbers against blacklists or whitelists, and are unable to detect potential fraud in real time and respond immediately. As a result, users are at increased risk of becoming victims of fraud. The objective of this invention is to provide a system that automatically detects fraudulent calls and immediately terminates the call, thereby protecting users from fraud.

[1263] 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.

[1264] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords and phrases that are suspected of being fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a threshold, means for adding phone numbers that are suspected of being fraudulent to the blacklist, means for synchronizing the updated blacklist with all related communication devices, means for detecting and automatically notifying of suspected fraudulent calls in real time, and means for dynamically managing the blacklist and whitelist and synchronizing the latest list information with the communication device. This makes it possible to more quickly and accurately detect fraudulent calls through real-time analysis of call content and dynamic list management, thereby safely protecting users.

[1265] A "communication device" is a device for communicating, and includes mobile devices such as telephones, smartphones, and tablets, as well as landlines.

[1266] "Incoming call" refers to the reception of an external call or message to a communication device.

[1267] "Telephone number information" is data including the telephone number itself and its associated attribute information.

[1268] A "blacklist" is a list of phone numbers that are suspected of fraud or have been found to be involved in fraudulent activities, and when a number is added to this list, it is blocked when it receives a call.

[1269] A "whitelist" is a list of safe phone numbers, and by registering a number on this list, it is given priority when receiving calls.

[1270] "Generative artificial intelligence" is an AI technology that learns on its own based on large data sets and can perform highly accurate analysis and responses to new data, and is particularly good at natural language processing.

[1271] "Automatic response" is a function in which a communication device mechanically responds on behalf of a user.

[1272] "Real-time analysis" is a technology that instantly analyzes the contents of calls and quickly reflects the results.

[1273] "Suspected fraudulent keywords and phrases" are specific words or expressions in the content of a call that suggest fraudulent activity.

[1274] The "risk score" is an evaluation index that quantifies the likelihood of fraud based on the content of the call.

[1275] A "threshold" is a value that indicates a certain standard, and when this value is exceeded, a specific process is executed.

[1276] "Dynamic management" refers to a management method in which the contents of the list are updated in real time or as needed, and the latest information is always synchronized with the entire system.

[1277] To implement the present invention, it is necessary to build a system using a server, a communication device (hereinafter referred to as a "terminal"), and a generative artificial intelligence (hereinafter referred to as a "generative AI model"). Specific embodiments of each function are described below.

[1278] Overview of the entire system

[1279] The system stores blacklists and whitelists on a server, communicates with devices in real time, and processes incoming calls. It uses a generative AI model to analyze the content of calls and detect potential fraud.

[1280] server

[1281] 1. Acquiring and verifying phone numbers

[1282] The system receives incoming phone number information sent from the device and compares it with the blacklist and whitelist stored on the server, instantly identifying known fraud numbers and safe numbers.

[1283] 2. Instructions for starting the automatic response

[1284] Based on the result of the comparison, an instruction is sent to the terminal to initiate an automatic response for numbers that are not on the list.

[1285] 3. Real-time analysis and risk assessment

[1286] Using a generative AI model, the call is transcribed in real time to detect keywords and phrases that may be fraudulent, and a risk score is calculated based on this, instructing the device to immediately terminate the call if a threshold is exceeded.

[1287] 4. Update and sync your list

[1288] Add suspected fraud numbers to a blacklist that is synced in real time to all relevant devices, and add safe calling numbers to a whitelist.

[1289] Terminal

[1290] 1. Receiving a call and sending the number to the server

[1291] When the user's terminal receives an incoming call, the telephone number information is immediately sent to the server.

[1292] 2. Implementing an automatic response

[1293] Upon receiving instructions from the server, the system switches to automatic answering mode, which transmits the contents of the call to the server in real time.

[1294] 3. End the call

[1295] When it receives a call termination instruction from the server, it immediately terminates the call and adds the number to the blacklist, saving the updated list in a local database.

[1296] Generative AI Models

[1297] The generative AI model is used to transcribe the call and analyze it for specific keywords and phrases, such as "please prepare money" which are indicative of fraud, to generate a risk score.

[1298] Specific examples

[1299] Scenario 1: Detecting and terminating fraudulent calls

[1300] 1. The user receives a call from an unknown number.

[1301] 2. The device sends the phone number to the server.

[1302] 3. The server checks the blacklist and whitelist and determines that there is no match.

[1303] 4. The server sends an automatic response instruction to the terminal.

[1304] 5. The call content is analyzed by a generative AI model, and phrases such as "Please prepare some money" are detected.

[1305] 6. Because the risk score exceeds the threshold, the server sends an instruction to end the call to the terminal, and the call ends.

[1306] 7. The number will be added to the blacklist and synced to all your devices.

[1307] Scenario 2: Whitelisting Secure Calls

[1308] 1. You receive a call from a trusted person and the number is not on your list.

[1309] 2. The device sends the phone number to the server.

[1310] 3. The server checks the blacklist and whitelist and determines that there is no match.

[1311] 4. The server sends an automatic response instruction to the terminal.

[1312] 5. The call content is analyzed by a generative AI model and determined to be free of fraud risk.

[1313] 6. The call completes successfully and the number is added to the whitelist.

[1314] Prompt Sentence Examples

[1315] "Please analyze the following call and assess its potential for fraud.

[1316] Text: [insert text of call here]

[1317] Please provide a risk score based on relevant phrases and keywords."

[1318] In this way, the system uses real-time analysis and dynamic list management to quickly and accurately detect fraudulent calls and keep users safe.

[1319] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1320] Step 1:

[1321] The server receives an incoming call from the terminal and obtains the phone number information. The input is the phone number received by the terminal, which is sent to the server. The server receives this phone number information and proceeds to the next step.

[1322] Step 2:

[1323] The server compares the acquired phone number information with existing blacklists and whitelists. The server has blacklists and whitelists stored in a database. The inputs are the received phone number and the blacklist and whitelist, and the comparison results in the output being no match, blacklist, or whitelist.

[1324] Step 3:

[1325] The device switches to automatic response mode based on the server's matching results. In particular, automatic response using a generative AI model is initiated for phone numbers that are not on the list. The server provides response instructions as input, and the device starts automatic response.

[1326] Step 4:

[1327] The generative AI model analyzes the content of the call in real time. The device sends the captured voice data to the server, which then converts it into text data. The input is the voice data of the call, and the output is text data generated based on that data.

[1328] Step 5:

[1329] The server uses the generative AI model to analyze the call content and detect keywords and phrases that may be fraudulent. The input is the generated text data, which is analyzed to extract fraud risk factors. The output is a risk score based on the fraud risk.

[1330] Step 6:

[1331] The server calculates a risk score based on the fraud detection results. The input is data on keywords and phrases that are suspected to be fraudulent, and the risk score is calculated based on this. The output is a specific numerical value for the risk score.

[1332] Step 7:

[1333] If the risk score exceeds the threshold, the server sends an instruction to the terminal to end the call. The inputs are the risk score and threshold information, and the output is an instruction to end the call.

[1334] Step 8:

[1335] The terminal receives a call termination instruction from the server and terminates the call. The input is a call termination instruction from the server, and the output is the end of the call.

[1336] Step 9:

[1337] The server adds suspected fraudulent phone numbers to a blacklist and synchronizes the updated blacklist to all relevant devices. The inputs are fraudulent phone numbers and blacklists, and a new blacklist is created. The output is the new blacklist synchronized to an online database.

[1338] Step 10:

[1339] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the updated whitelist to all devices. The inputs are the phone numbers deemed safe and the whitelist, and a new whitelist is created. The output is the new whitelist synchronized to an online database.

[1340] These are the specific processing steps of this system. This series of steps enables fraudulent calls to be detected in real time and the necessary defensive measures to be taken automatically.

[1341] 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.

[1342] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[1343] System configuration

[1344] 1. Server

[1345] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates a risk score, recognizes the user's emotions through an emotion engine, and updates and synchronizes the list.

[1346] 2. Terminal

[1347] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call content and the user's emotional information are sent to the server in real time.

[1348] 3. List

[1349] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[1350] Program processing

[1351] Response when receiving a call

[1352] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[1353] Server: The server checks the phone number against blacklists and whitelists and directs the appropriate action.

[1354] Launch of AI automated response

[1355] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[1356] Server: The server uses generative artificial intelligence to initiate responses and analyze call content in real time.

[1357] Real-time analysis and risk assessment

[1358] Server: Generative AI converts call content into text data and detects keywords and phrases that may indicate fraud.

[1359] Server: The emotion engine recognizes the user's emotions in real time and analyzes them along with the call content.

[1360] Calculating the risk score

[1361] Server: Calculates a risk score based on the fraud detection results and the emotional information recognized by the emotion engine.

[1362] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[1363] End call and update list

[1364] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[1365] Terminal: If the caller number of the terminated call is to be added to the blacklist, it sends the corresponding information to the server.

[1366] List Synchronization

[1367] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[1368] On your device: Download the new blacklist and update the local database.

[1369] Safe Call Rating and Whitelist Updates

[1370] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[1371] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[1372] Specific examples

[1373] Example 1: Automatically detect fraudulent calls and terminate them

[1374] User: I get a call from an unknown number.

[1375] Terminal: Sends the phone number to the server.

[1376] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1377] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1378] Server: The generative AI detects the phrase "please prepare money," and the emotion engine detects the user's tension and anxiety. It then assigns a high risk score.

[1379] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[1380] Terminal: The call is ended and the number is added to the blacklist.

[1381] Example 2: Secure Call Whitelisting Scenario

[1382] User: A call is coming from someone you trust.

[1383] Terminal: Sends the phone number to the server.

[1384] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1385] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1386] Server: Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. It determines that there is no risk of fraud.

[1387] Server: The call is completed successfully and the phone number is added to the whitelist.

[1388] Device: Download the new whitelist.

[1389] In this way, the system of the present invention combines advanced analytical capabilities with user emotion recognition to automate responses to fraud and provide users with a safe and secure communication environment.

[1390] The processing flow will be explained below.

[1391] Step 1:

[1392] User: A call comes in on the user's device.

[1393] Terminal: Detects an incoming call and obtains the caller's phone number information.

[1394] Terminal: Sends the acquired phone number information to the server.

[1395] Step 2:

[1396] Server: Checks the received phone number information against existing blacklists and whitelists.

[1397] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[1398] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[1399] Step 3:

[1400] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[1401] Terminal: Transmits voice data and user emotion information to the server in real time.

[1402] Step 4:

[1403] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[1404] Server: The emotion engine recognizes the user's emotions in real time and analyzes the emotion information along with the call content.

[1405] Step 5:

[1406] Server: Calculates a risk score based on the results of analyzing the call content and the user's emotion recognition.

[1407] Server: If a keyword or phrase is detected that is suspected to be fraudulent, set a higher risk score.

[1408] Step 6:

[1409] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to terminate the call.

[1410] Server: Adds the phone number to the blacklist and syncs the update to all devices.

[1411] Step 7:

[1412] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[1413] Terminal: Apply the received updated blacklist to the local database.

[1414] Step 8:

[1415] Server: If the call is assessed as safe, add the caller ID to a whitelist and manage it.

[1416] Server: Sends instructions to synchronize whitelist updates to all devices.

[1417] On the device: Download the new whitelist and apply it to the local database.

[1418] Step 9:

[1419] Device: Prepare for the next incoming call by quickly assessing its trustworthiness using the latest blacklists and whitelists.

[1420] Example 2

[1421] 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."

[1422] In modern society, fraudulent activities are becoming increasingly sophisticated, making it difficult to protect personal information and property. It is difficult for users to respond immediately to fraudulent calls made through communication devices, and the mental burden is also great. Therefore, there is a need to provide an environment in which users can use communication devices with peace of mind.

[1423] 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.

[1424] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for recognizing the user's emotions and calculating a risk score along with the fraud detection result, means for terminating the call if the risk score exceeds a threshold, and means for adding phone numbers suspected of being fraudulent to the blacklist and synchronizing the updated blacklist with all related communication devices. This makes it possible to automatically detect calls suspected of being fraudulent and prevent users from becoming victims of fraudulent calls.

[1425] A "communication device" is a device through which a user makes voice calls, and generally includes a mobile phone or a landline phone.

[1426] "Telephone number information" is data indicating the telephone number of the caller, and is used by the communication device to identify the caller.

[1427] A "blacklist" is a list of telephone numbers that have previously been the subject of fraudulent or illicit activity, and is used to prevent fraudulent calls.

[1428] A "whitelist" is a list of safe and trusted phone numbers that is used to accurately classify important calls.

[1429] "Generative artificial intelligence" refers to an algorithm or system that uses AI technology to generate natural, human-like dialogue and has the ability to analyze call content in real time.

[1430] An "emotion engine" is a system that uses voice recognition technology to analyze a user's emotional state and acquires that emotion as data.

[1431] A "risk score" is a quantified indicator of risk calculated based on the content of the call and the user's emotional state, and is used to assess the degree of suspected fraud.

[1432] "Synchronizing" refers to matching data between multiple communication devices, and refers to the process of reflecting updates to the blacklist or whitelist in all communication devices.

[1433] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[1434] System configuration

[1435] server

[1436] The server is the main processing unit that performs real-time analysis of call content, calculates risk scores, recognizes user emotions using an emotion engine, and manages blacklists and whitelists. It performs these processes using generative artificial intelligence (AI) and an emotion engine.

[1437] Terminal

[1438] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The content of the call and the user's emotional information are sent to the server in real time.

[1439] list

[1440] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[1441] Program processing

[1442] Response when receiving a call

[1443] When a user receives a call, the number is sent to the server. The server checks the incoming number against a blacklist and a whitelist and instructs the appropriate action. If the number is not on either list, the device switches to auto-answer mode and the server begins responding using generative artificial intelligence. The content of the call is sent to the server in real time.

[1444] Real-time analysis and risk assessment

[1445] The server uses generative artificial intelligence to convert the call content into text data and detect keywords and phrases that may indicate fraud. At the same time, an emotion engine recognizes the user's emotions in real time and analyzes them along with the call content. A risk score is calculated based on the fraud detection results and the emotional information recognized by the emotion engine. If the risk score exceeds a threshold, the server sends an instruction to end the call to the terminal.

[1446] Ending a call and updating the list

[1447] Upon receiving the call termination instruction, the terminal immediately terminates the call and, if the caller number is to be added to the blacklist, sends corresponding information to the server. The server updates the blacklist and sends a synchronization instruction to all related communication devices. The terminal downloads the new blacklist and updates its local database.

[1448] Safe Call Rating and Whitelist Updates

[1449] If the call is deemed safe, the server adds the caller ID to the whitelist, and the device downloads the new whitelist from the server and prepares for the next incoming call.

[1450] Specific examples

[1451] Example 1: Automatically detect fraudulent calls and terminate them

[1452] The user receives a call from an unknown number. The device sends the phone number to the server. Since the server does not match the list, it instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. The generative AI detects the phrase "please prepare some money" and the emotion engine detects the user's tension or anxiety, resulting in a high risk score. If the score exceeds the threshold, the server sends an instruction to end the call to the device. The device ends the call and adds the number to the blacklist.

[1453] Example 2: Secure Call Whitelisting Scenario

[1454] The user receives a call from a trusted party. The device sends the phone number to the server. Since the call does not match the list, the server instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. If it is determined that there is no risk of fraud, the call is terminated normally and the phone number is added to the whitelist. The device downloads the new whitelist and prepares for the next call.

[1455] Prompt Sentence Examples

[1456] "How does this system detect and block suspected fraudulent calls?"

[1457] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1458] Program processing flow

[1459] Step 1: Incoming call detection and number matching

[1460] Terminal: The user receives a call on their terminal. The phone number is sent to the server.

[1461] Input: Incoming call phone number

[1462] Output: Phone number data sent to the server

[1463] Specific operation: The moment a call comes in, the "+Notification" API is called and the phone number is sent to the server.

[1464] Step 2: List Matching

[1465] Server: Immediately checks the received phone number against blacklists and whitelists.

[1466] Input: Phone number data sent to the server

[1467] Output: List match result (existing in blacklist, existing in whitelist, not in either)

[1468] Specific operation: Rapidly search the incoming number against the blacklist and whitelist in the database.

[1469] Step 3: Start the AI ​​Automated Response

[1470] Server: If the number is not on either list, instruct the device to start an AI automatic response.

[1471] Input: List match result (not in either list)

[1472] Output: Instructions to start AI automatic response sent to the device

[1473] Specific operation: Launches the specified generative artificial intelligence (Generative AI) model and generates a response script.

[1474] Step 4: Start Auto-Answer Mode

[1475] Terminal: Switches to AI automatic response mode upon receiving instructions from the server.

[1476] Input: Instruction to start AI automatic response

[1477] Output: Start auto attendant

[1478] What it does: Loads an audio file for the auto attendant and keeps the call active.

[1479] Step 5: Real-time analysis of call content

[1480] Terminal: Sends voice data during a call to the server in real time.

[1481] Server: Generative AI converts the received speech into text data and detects keywords and phrases that may be fraudulent.

[1482] Input: Voice data during a call

[1483] Output: Real-time text data, detection of potentially fraudulent keywords and phrases

[1484] Specific operation:

[1485] Device: Uses WebRTC to send audio data to the server in streaming format.

[1486] Server: Uses the Speech-to-Text API to convert speech to text in real time and leverages natural language processing (NLP) techniques to perform text analysis.

[1487] Step 6: Real-time sentiment analysis

[1488] Server: Analyzes user emotions in real time using an emotion engine.

[1489] Input: Voice data during a call

[1490] Output: Analysis of the user's emotional state

[1491] Specific operation: The emotion engine analyzes the user's emotions from the voice in real time and generates emotion data.

[1492] Step 7: Risk Score and Call Evaluation

[1493] Server: Calculates risk scores based on the results of text analysis and the emotion engine's recognition data.

[1494] Input: Text analysis results, sentiment analysis results

[1495] Output: Risk score

[1496] Specific operation: Calculates a risk score based on an algorithm that takes into account fraud detection results and emotional information.

[1497] Step 8: End call instructions

[1498] Server: If the risk score exceeds the threshold, sends an instruction to end the call to the terminal.

[1499] Input: Risk score

[1500] Output: End call instruction

[1501] Specific operation: If the risk score exceeds a preset threshold, a call termination instruction is sent to the terminal using a RESTful API.

[1502] Step 9: End the call and update the blacklist

[1503] Terminal: Upon receiving the instruction to end the call, it immediately ends the call and reports the caller ID to the server.

[1504] Server: Add the received caller number to the blacklist and store the relevant information in the database.

[1505] Input: End call instruction, ended caller number

[1506] Output: Updated blacklist

[1507] Specific operation:

[1508] Terminal: Upon receiving a call end instruction, immediately execute the Hangup API.

[1509] Server: Updates the database to add the caller number to the blacklist and immediately block the next incoming call.

[1510] Step 10: Synchronizing the Blacklist

[1511] Server: Sends a synchronization instruction to all associated communication devices to update the blacklist.

[1512] On your device: Download the new blacklist and update the local database.

[1513] Input: Updated blacklist

[1514] Output: Synchronization instructions, updated local database

[1515] Specific operation:

[1516] Server: Notifies the communication device of blacklist updates using WebSocket.

[1517] Terminal: The new blacklist is retrieved via an API call and reflected in the local database.

[1518] Step 11: Update the whitelist

[1519] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[1520] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[1521] Input: Caller ID after call rated safe

[1522] Output: Updated whitelist, updated local database on device

[1523] Specific operation:

[1524] Server: If the call is assessed as safe, add the number to the whitelist and update the database.

[1525] Device: Download the updated whitelist and apply it the next time you receive a call.

[1526] (Application example 2)

[1527] 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."

[1528] Telephone customer service on online shopping sites is an important means of communication for users. However, the number of calls suspected to be fraudulent is increasing, threatening user safety. To solve this problem, a system that automatically detects suspicious calls and allows only safe calls is needed. In addition, more accurate fraud detection is expected by performing a comprehensive risk assessment that takes user emotions into account.

[1529] 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.

[1530] In this invention, the server includes means for receiving calls from communication devices and acquiring phone number information, means for comparing the calls with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results and the user's emotional information recognized by the emotion engine, means for terminating the call if the risk score exceeds a threshold, means for adding the phone number suspected of fraud to the blacklist, means for synchronizing the updated blacklist with all related communication devices, and means for converting the content of the call into text data in real time, thereby ensuring the security of customer service calls and reducing the risk of fraud.

[1531] "Communication device" is a general term for electronic devices used to make calls and send and receive messages.

[1532] An "incoming call" refers to an incoming call or message from an external device.

[1533] "Telephone number information" is a combination of numbers and codes for identifying a specific communication device.

[1534] A "blacklist" is a list of phone numbers suspected of being fraudulent or of unreliable callers.

[1535] A "whitelist" is a list of safe and trusted sources.

[1536] "Generative AI" is AI that has the ability to generate new information and answers based on input data.

[1537] "Auto-answer" is the ability of a communication device to automatically reply without human intervention.

[1538] "Real-time analysis" is the process of analyzing data as soon as it is received.

[1539] "Suspected fraudulent keywords and phrases" refer to specific words or expressions that suggest fraudulent activity.

[1540] A "risk score" is an evaluation value that quantifies the likelihood of fraud.

[1541] An "emotion engine" is software or an algorithm for analyzing and recognizing the emotions of a user of a communication device.

[1542] Converting to "text data" means converting formal data such as audio and video into text information.

[1543] "Updating" means replacing existing information with new information.

[1544] "Synchronization" is the process of matching data between multiple devices.

[1545] An "online shopping site" is a website for buying and selling products and services over the Internet.

[1546] "Customer service" refers to the support activities provided when customers inquire about products or services via communication devices.

[1547] To realize a fraud prevention system for customer service at a mail-order site, the present invention uses a communication device, a server, and necessary programs.

[1548] 1. System Configuration

[1549] server

[1550] The server performs the following main tasks:

[1551] When a call comes in, the phone number information is received and compared against the blacklist and whitelist.

[1552] If the phone number is not on the list, an automated response is initiated using generative artificial intelligence.

[1553] Calls are converted into text data in real time to detect keywords and phrases that may be fraudulent.

[1554] The emotion engine analyzes the user's emotions and calculates a risk score along with the fraud detection results.

[1555] If the risk score exceeds a threshold, the call is terminated and the suspected fraudulent phone number is added to a blacklist.

[1556] The updated blacklist is synchronized to all associated communication devices.

[1557] Communication equipment (user terminal)

[1558] The user's communication device has the following roles:

[1559] An incoming call is received and the telephone number information is sent to the server.

[1560] Receives instructions from the server and switches to auto-response mode.

[1561] The call contents and user emotional information are sent to the server in real time.

[1562] When an instruction to end the call is received, the call is immediately ended.

[1563] Download new blacklists and whitelists from the server and update the local database.

[1564] 2. Program processing explanation

[1565] The programs in the server and communication device operate using the following hardware and software:

[1566] Hardware:

[1567] Server: A computer system equipped with a high-performance processor

[1568] Communication devices: Smartphones and internet-enabled phones

[1569] software:

[1570] Generative AI: Hugging Face Transformer model (e.g. 'gpt-3')

[1571] Emotion Engine: Emotion recognition algorithm (e.g., EmotionRecognizer)

[1572] Speech Recognition: Speech to text conversion using the speech_recognition package

[1573] HTTP communication: Use the requests library to send and receive data

[1574] The server receives the phone number sent from the user's device and compares it with the blacklist and whitelist. If the phone number is not on the list, it uses generative artificial intelligence to initiate an automatic response and converts the call content into text data in real time. It then detects keywords and phrases that may be fraudulent, while an emotion engine simultaneously analyzes the user's emotional information. Based on this information, it calculates a risk score, and if the score exceeds a threshold, it sends an instruction to the user's device to terminate the call.

[1575] 3. Examples of concrete examples and prompts

[1576] Specific examples

[1577] When an unknown number calls the customer service of an online shopping site, the user's device sends the phone number to the server. The server checks the number and, if it is not in the list, it issues an automatic response. During the automatic response, the server analyzes the content of the call and

[1578] The emotion engine reads the user's emotions and calculates a risk score. If there is a high suspicion of fraud, the call is terminated and the number is added to a blacklist.

[1579] Prompt Sentence Examples

[1580] Prompts that pass call content through a generative AI model to detect fraudulent phrases

[1581] User: "Hello, is this customer service from an online shopping site?"

[1582] AI: "Hi, this is customer service. How can I help you?"

[1583] By inputting this prompt into a generative AI model, the call content begins to be analyzed and fraudulent phrases are detected.

[1584] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1585] Step 1:

[1586] A user receives an incoming call and the communication device acquires the phone number information. The communication device (user terminal) sends the received phone number information to the server. The server receives the phone number information as input and prepares to start processing.

[1587] Step 2:

[1588] The server compares the received phone number information with the existing blacklist and whitelist. The input is the phone number information, blacklist, and whitelist, and the output is the comparison result. Based on the comparison result, it checks whether the phone number is in the list.

[1589] Step 3:

[1590] If the matching result shows that the phone number is not on either the blacklist or the whitelist, the server instructs the communication device to switch to automatic response mode. The input is the matching result, and the output is the start of automatic response mode. The communication device receives this instruction and starts an automatic response using generative artificial intelligence.

[1591] Step 4:

[1592] When the automatic answering starts, the communication device sends the contents of the call to the server in real time. The input of the communication device is the voice data of the call contents, and the output is the voice data sent to the server. The server analyzes the received voice data and converts it from voice to text data.

[1593] Step 5:

[1594] The server then analyzes the converted text data using generative artificial intelligence to detect keywords and phrases that may be fraudulent. The input is text data, and the output is the detection results of keywords and phrases that may be fraudulent. This allows the server to identify potential fraudulent activity.

[1595] Step 6:

[1596] Meanwhile, the server uses an emotion engine to analyze the user's emotional information and determine the user's emotional state. The input is voice data and the output is emotional information. The emotional information serves to detect signs of anxiety or tension in the user.

[1597] Step 7:

[1598] The server combines the analysis results of the generative AI with the emotional information from the emotion engine to calculate a risk score. The input is the keyword and phrase detection results and emotional information, and the output is a risk score. This score quantifies the likelihood of fraud.

[1599] Step 8:

[1600] If the risk score exceeds the threshold, the server sends an instruction to the communication device to end the call. The input is the risk score, and the output is an instruction to end the call. The communication device receives this instruction and immediately ends the call.

[1601] Step 9:

[1602] After the call is completed, the server processes the suspected fraudulent phone number by adding it to the blacklist. The input is the phone number information, and the output is the updated blacklist. The server updates the blacklist and synchronizes it to all relevant communication devices.

[1603] Step 10:

[1604] The communication device downloads a new blacklist from the server and updates the local database. The input is the updated blacklist, and the output is the updated local database. This allows for a more robust response to future incoming calls.

[1605] Prompt Sentence Examples

[1606] Here's a prompt that feeds the call content to a generative AI model to detect scam phrases:

[1607] User: "Hello, is this customer service from an online shopping site?"

[1608] AI: "Hi, this is customer service. How can I help you?"

[1609] By inputting this prompt into a generative AI model, analysis of the call content begins and fraudulent phrases are detected.

[1610] 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.

[1611] 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.

[1612] 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.

[1613] [Fourth embodiment]

[1614] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1615] 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.

[1616] 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).

[1617] 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.

[1618] 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.

[1619] 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).

[1620] 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.

[1621] 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.

[1622] 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.

[1623] 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.

[1624] 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.

[1625] 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.

[1626] 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."

[1627] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking the telephone carrier's server with the user's communication device.

[1628] System configuration

[1629] 1. Server

[1630] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates the risk score, and updates and synchronizes the list.

[1631] 2. Terminal

[1632] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call contents are sent to the server in real time.

[1633] 3. List

[1634] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[1635] Program processing

[1636] Response when receiving a call

[1637] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[1638] Server: The server checks the phone number against a blacklist and whitelist and issues the appropriate action.

[1639] Launch of AI automated response

[1640] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[1641] Server: The server uses generative artificial intelligence to initiate the call and analyze the call content in real time.

[1642] Real-time analysis and risk assessment

[1643] Server: Generative AI converts the call into text data, detects keywords and phrases that may be fraudulent, and calculates an overall risk score for the call.

[1644] Actions based on risk scores

[1645] Server: If the score exceeds the threshold, the server sends an instruction to the device to terminate the call. It also adds the number to the blacklist and synchronizes the updated information to all devices.

[1646] Terminal: Upon receiving the call termination instruction, the terminal immediately terminates the call and saves the updated blacklist in its local database.

[1647] Update the list after the call ends

[1648] Server: Call numbers that are deemed safe are added to a whitelist.

[1649] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[1650] Specific examples

[1651] Example 1: Automatically detect fraudulent calls and terminate them

[1652] User: I get a call from an unknown number.

[1653] Terminal: Sends the phone number to the server.

[1654] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1655] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1656] Server: Generative AI detects the phrase "please prepare money" and assigns a high risk score.

[1657] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[1658] Terminal: The call is ended and the number is added to the blacklist.

[1659] Example 2: Secure Call Whitelisting Scenario

[1660] User: A call is coming from someone you trust.

[1661] Terminal: Sends the phone number to the server.

[1662] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1663] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1664] Server: Generative AI analyzes the content of the call and determines there is no risk of fraud.

[1665] Server: The call is completed successfully and the phone number is added to the whitelist.

[1666] Device: Download the new whitelist.

[1667] In this way, the system utilizes its advanced analytical capabilities to automate responses to fraud and provide users with a safe and secure communications environment.

[1668] The processing flow will be explained below.

[1669] Step 1:

[1670] User: A call comes in on the user's device.

[1671] Terminal: Detects an incoming call and obtains the caller's phone number information.

[1672] Terminal: Sends the acquired phone number information to the server.

[1673] Step 2:

[1674] Server: Checks the received phone number information against existing blacklists and whitelists.

[1675] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[1676] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[1677] Step 3:

[1678] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[1679] Terminal: Initiates an automatic response and transmits the call contents to the server in real time.

[1680] Step 4:

[1681] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[1682] Server: Detects potentially fraudulent keywords and phrases.

[1683] Step 5:

[1684] Server: Calculates the risk score based on the fraud detection results.

[1685] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[1686] Step 6:

[1687] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[1688] Device: Add the caller number of the ended call to the blacklist.

[1689] Step 7:

[1690] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[1691] On your device: Download the new blacklist and update the local database.

[1692] Step 8:

[1693] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[1694] Server: Updates the whitelist and sends instructions to synchronize to all relevant communication devices.

[1695] Device: Download the new whitelist and update the local database.

[1696] Example 1

[1697] 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."

[1698] In recent years, fraudulent activities via communication devices have been on the rise, increasing the risk that users will inadvertently provide personal information or assets to fraudsters. To prevent such fraudulent activities, a system that can detect suspected fraudulent calls in real time and respond automatically is needed. Conventional systems mainly use blacklists and whitelists, which often result in delayed responses to new fraudulent calls. Additionally, manual analysis of call content has the problem of being unable to respond quickly.

[1699] 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.

[1700] In this invention, the server includes means for detecting an incoming call from a communication terminal and acquiring telephone number information, means for comparing the acquired telephone number information with a blacklist and a whitelist in an internal database, means for initiating an automatic response using generative artificial intelligence for telephone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a set threshold, means for adding the suspected fraudulent telephone number to the blacklist, means for synchronizing the updated blacklist with all related communication terminals, and means for converting the content of the call into text data using voice recognition technology, thereby enabling new and existing fraudulent calls to be effectively detected and dealt with in real time.

[1701] 1. "Communications terminal" means a device used by a user to communicate.

[1702] 2. "Incoming call" refers to a call coming into a communication terminal.

[1703] 3. "Telephone number information" refers to data relating to a telephone number obtained when a call is received.

[1704] 4. "Internal database" refers to the system that manages data stored within the server.

[1705] 5. A "blacklist" is a list in which malicious communications are registered in advance.

[1706] 6. A "whitelist" is a list in which safe communications are registered in advance.

[1707] 7. "Generative AI" means AI that has the ability to analyze natural language and generate responses.

[1708] 8. "Automatic response" refers to the function in which the system automatically responds on behalf of the user in response to certain conditions.

[1709] 9. "Real-time analysis" refers to the process of instantly analyzing the contents of a call.

[1710] 10. "Fraud" refers to malicious conduct intended to deceive others.

[1711] 11. "Keywords and phrases" are words or short phrases that have a specific meaning.

[1712] 12. "Risk Score" is a numerical assessment of the likelihood of fraud.

[1713] 13. "Set threshold" means a numerical value that serves as a benchmark for risk assessment.

[1714] 14. "End a call" means to stop making or receiving calls.

[1715] 15. "Updated Blacklist" means the most recent blacklist with new fraud numbers added.

[1716] 16. "Sync to Device" means to share data with all associated devices.

[1717] 17. "Speech recognition technology" refers to technology that converts speech into text data.

[1718] 18. "Text data" means data converted into character information.

[1719] The present invention is a system that uses generative artificial intelligence to automatically detect and block suspicious calls from incoming calls made to a user's communication device. This system is realized by linking a server and the user's communication device.

[1720] System Hardware and Software

[1721] The server acts as the main processing unit and has the following responsibilities:

[1722] 1. Phone number verification when receiving a call

[1723] 2. Start the automatic response

[1724] 3. Real-time analysis of call content

[1725] 4. Calculating the risk score

[1726] 5. Updating and Syncing Lists

[1727] Specific software used includes database management systems (e.g., MySQL or PostgreSQL) and generative artificial intelligence (e.g., GPT-3), as well as voice recognition technologies such as the Google Speech-to-Text API.

[1728] The user's communication terminal has the following roles:

[1729] 1. Receiving a call

[1730] 2. Start of AI automatic response based on instructions from the server

[1731] 3. Sending call content to the server

[1732] 4. Blacklist and Whitelist Updates

[1733] System operation explanation

[1734] Incoming call detection:

[1735] When the user's communication device detects an incoming call, the device sends the phone number information to the server. For example, the device performs a process to call an API that sends the phone number to the server via a network connection.

[1736] Matching the list:

[1737] The server checks the incoming phone number against an internal database of blacklists and whitelists, using efficient database queries, such as SQL select queries:

[1738] Start Auto-Reply:

[1739] If the number is not on either list, the communication terminal switches to automatic response mode and starts an automatic response upon receiving instructions from the server. At this time, the server uses generative artificial intelligence to generate the response content and start the call.

[1740] Real-time analytics:

[1741] Generative AI converts call content into text data in real time, detects keywords and phrases that may be fraudulent, and uses voice recognition technology (e.g., Google Speech-to-Text API) to convert speech to text.

[1742] Risk score calculation:

[1743] Generative AI calculates a risk score for a call, using specific algorithms based on keywords and phrases that are suspected to be fraudulent.

[1744] End the call and refresh the list:

[1745] If the risk score exceeds a set threshold, the server sends an instruction to the communication device to terminate the call and adds the phone number to a blacklist, and the updated blacklist is synchronized with all related communication devices.

[1746] Adding Secure Calls to the whitelist:

[1747] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the new whitelist to all associated communication terminals.

[1748] Specific examples

[1749] Example 1: Automatic detection and termination of fraudulent calls

[1750] A user receives a call from an unknown number on their communication terminal.

[1751] The terminal sends the phone number to the server.

[1752] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[1753] The terminal starts an automatic response and transmits the contents of the call to the server.

[1754] The server uses generative artificial intelligence to detect the phrase "I need money" and calculates a high risk score.

[1755] Because the risk score exceeds the threshold, the server instructs the call to end.

[1756] The device will end the call and add the phone number to the blacklist.

[1757] Example 2: Adding Secure Calls to the Whitelist

[1758] A user receives a call from a trusted party on their communication terminal.

[1759] The terminal sends the phone number to the server.

[1760] Since the server does not fall under the blacklist or whitelist, an AI automatic response is instructed.

[1761] The terminal starts an automatic response and transmits the contents of the call to the server.

[1762] The server uses generative artificial intelligence to analyze the content of the call and determine that there is no risk of fraud.

[1763] After the call is successfully completed, the server adds the phone number to the whitelist.

[1764] The device will download the new whitelist.

[1765] Example prompt sentence:

[1766] "I need some money," "This is an urgent call," "There's an investment opportunity."

[1767] In this way, the system leverages powerful generative artificial intelligence and speech recognition technology to provide users with a safe and secure communication experience.

[1768] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1769] Step 1: Detecting an incoming call and obtaining phone number information

[1770] Terminal: The user's communication terminal detects an incoming call. At this point, the terminal obtains the incoming call's phone number information and sends it to the server. The input is the incoming call's phone number, and the output is sending the phone number data to the server. Specifically, the terminal's communication module receives the incoming call signal and sends the phone number as data to the server.

[1771] Step 2: Match list of phone numbers

[1772] Server: Upon receiving phone number information, the server checks the phone number against the blacklist and whitelist in its internal database. The input is the submitted phone number, and the output is the list check result. Specifically, the server searches the blacklist and whitelist using a database query such as SQL to obtain the check result.

[1773] Step 3: Processing the list match results

[1774] Server: Based on the comparison results, if the phone number is on the blacklist, the call is rejected, and if it is on the whitelist, the call is allowed. If it does not match the list, an AI automatic response is initiated. The input is the list comparison result, and the output is instructions for the next process. In terms of specific operation, the internal logic performs conditional branching to select the appropriate process.

[1775] Step 4: Start the AI ​​Automated Response

[1776] Terminal: Upon receiving instructions from the server, the terminal switches to auto-answer mode. At this time, the call is redirected to the server and the generative AI begins to respond. The input is the auto-answer instruction from the server, and the output is the start of the auto-answer. Specifically, the terminal sets up the call redirection, and the AI ​​model begins to generate a response.

[1777] Step 5: Real-time analysis of call content

[1778] Server: Generative AI converts the contents of the call into text data in real time and detects keywords and phrases that may be fraudulent. The input is the audio data of the call, and the output is the text data and detection results. Specifically, speech recognition technology (for example, Google Speech-to-Text API) converts the audio into text, and uses that text to analyze keywords and phrases.

[1779] Step 6: Calculate the risk score

[1780] Server: Calculates fraud risk score based on the analysis results of the call content. The input is the analyzed text data and detected keywords, and the output is the risk score. Specific operations involve quantifying risk using a specific algorithm (e.g., a rule-based scoring system).

[1781] Step 7: End call instructions

[1782] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to end the call. It also adds phone numbers suspected of fraud to a blacklist. The input is the risk score, and the output is an instruction to end the call and an update to the blacklist. Specifically, it makes an API call to issue an end instruction and adds a new blacklist item to the database.

[1783] Step 8: Synchronizing the Blacklist

[1784] Terminal: Receives the call termination instruction from the server, terminates the call, and saves the updated blacklist in the local database. The input is the instruction from the server and the blacklist data, and the output is the end of the call and the updated blacklist. Specific operations include disconnecting the call and downloading and saving the blacklist information locally.

[1785] Step 9: Update and Sync Whitelist

[1786] Server: Adds call numbers that are determined to be free of fraud risk to a whitelist. The new whitelist is then synchronized to all communication devices. The input is the call number, and the output is updating and synchronizing the whitelist. Specific operations include adding the number to the whitelist and sending a notification to distribute it to all devices.

[1787] Step 10: Download the whitelist

[1788] Terminal: Upon receiving notification from the server, downloads the new whitelist and stores it locally. The input is the whitelist update notification from the server, and the output is the locally stored whitelist. The specific operation is to download the new whitelist and store it in the local database.

[1789] (Application example 1)

[1790] 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."

[1791] Currently, countermeasures against fraudulent calls in communication devices are extremely difficult because users must identify signs of fraud themselves. Furthermore, existing systems only check incoming numbers against blacklists or whitelists, and are unable to detect potential fraud in real time and respond immediately. As a result, users are at increased risk of becoming victims of fraud. The objective of this invention is to provide a system that automatically detects fraudulent calls and immediately terminates the call, thereby protecting users from fraud.

[1792] 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.

[1793] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords and phrases that are suspected of being fraudulent, means for calculating a risk score based on the fraud detection results, means for terminating the call if the risk score exceeds a threshold, means for adding phone numbers that are suspected of being fraudulent to the blacklist, means for synchronizing the updated blacklist with all related communication devices, means for detecting and automatically notifying of suspected fraudulent calls in real time, and means for dynamically managing the blacklist and whitelist and synchronizing the latest list information with the communication device. This makes it possible to more quickly and accurately detect fraudulent calls through real-time analysis of call content and dynamic list management, thereby safely protecting users.

[1794] A "communication device" is a device for communicating, and includes mobile devices such as telephones, smartphones, and tablets, as well as landlines.

[1795] "Incoming call" refers to the reception of an external call or message to a communication device.

[1796] "Telephone number information" is data including the telephone number itself and its associated attribute information.

[1797] A "blacklist" is a list of phone numbers that are suspected of fraud or have been found to be involved in fraudulent activities, and when a number is added to this list, it is blocked when it receives a call.

[1798] A "whitelist" is a list of safe phone numbers, and by registering a number on this list, it is given priority when receiving calls.

[1799] "Generative artificial intelligence" is an AI technology that learns on its own based on large data sets and can perform highly accurate analysis and responses to new data, and is particularly good at natural language processing.

[1800] "Automatic response" is a function in which a communication device mechanically responds on behalf of a user.

[1801] "Real-time analysis" is a technology that instantly analyzes the contents of calls and quickly reflects the results.

[1802] "Suspected fraudulent keywords and phrases" are specific words or expressions in the content of a call that suggest fraudulent activity.

[1803] The "risk score" is an evaluation index that quantifies the likelihood of fraud based on the content of the call.

[1804] A "threshold" is a value that indicates a certain standard, and when this value is exceeded, a specific process is executed.

[1805] "Dynamic management" refers to a management method in which the contents of the list are updated in real time or as needed, and the latest information is always synchronized with the entire system.

[1806] To implement the present invention, it is necessary to build a system using a server, a communication device (hereinafter referred to as a "terminal"), and a generative artificial intelligence (hereinafter referred to as a "generative AI model"). Specific embodiments of each function are described below.

[1807] Overview of the entire system

[1808] The system stores blacklists and whitelists on a server, communicates with devices in real time, and processes incoming calls. It uses a generative AI model to analyze the content of calls and detect potential fraud.

[1809] server

[1810] 1. Acquiring and verifying phone numbers

[1811] The system receives incoming phone number information sent from the device and compares it with the blacklist and whitelist stored on the server, instantly identifying known fraud numbers and safe numbers.

[1812] 2. Instructions for starting the automatic response

[1813] Based on the result of the comparison, an instruction is sent to the terminal to initiate an automatic response for numbers that are not on the list.

[1814] 3. Real-time analysis and risk assessment

[1815] Using a generative AI model, the call is transcribed in real time to detect keywords and phrases that may be fraudulent, and a risk score is calculated based on this, instructing the device to immediately terminate the call if a threshold is exceeded.

[1816] 4. Update and sync your list

[1817] Add suspected fraud numbers to a blacklist that is synced in real time to all relevant devices, and add safe calling numbers to a whitelist.

[1818] Terminal

[1819] 1. Receiving a call and sending the number to the server

[1820] When the user's terminal receives an incoming call, the telephone number information is immediately sent to the server.

[1821] 2. Implementing an automatic response

[1822] Upon receiving instructions from the server, the system switches to automatic answering mode, which transmits the contents of the call to the server in real time.

[1823] 3. End the call

[1824] When it receives a call termination instruction from the server, it immediately terminates the call and adds the number to the blacklist, saving the updated list in a local database.

[1825] Generative AI Models

[1826] The generative AI model is used to transcribe the call and analyze it for specific keywords and phrases, such as "please prepare money" which are indicative of fraud, to generate a risk score.

[1827] Specific examples

[1828] Scenario 1: Detecting and terminating fraudulent calls

[1829] 1. The user receives a call from an unknown number.

[1830] 2. The device sends the phone number to the server.

[1831] 3. The server checks the blacklist and whitelist and determines that there is no match.

[1832] 4. The server sends an automatic response instruction to the terminal.

[1833] 5. The call content is analyzed by a generative AI model, and phrases such as "Please prepare some money" are detected.

[1834] 6. Because the risk score exceeds the threshold, the server sends an instruction to end the call to the terminal, and the call ends.

[1835] 7. The number will be added to the blacklist and synced to all your devices.

[1836] Scenario 2: Whitelisting Secure Calls

[1837] 1. You receive a call from a trusted person and the number is not on your list.

[1838] 2. The device sends the phone number to the server.

[1839] 3. The server checks the blacklist and whitelist and determines that there is no match.

[1840] 4. The server sends an automatic response instruction to the terminal.

[1841] 5. The call content is analyzed by a generative AI model and determined to be free of fraud risk.

[1842] 6. The call completes successfully and the number is added to the whitelist.

[1843] Prompt Sentence Examples

[1844] "Please analyze the following call and assess its potential for fraud.

[1845] Text: [insert text of call here]

[1846] Please provide a risk score based on relevant phrases and keywords."

[1847] In this way, the system uses real-time analysis and dynamic list management to quickly and accurately detect fraudulent calls and keep users safe.

[1848] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1849] Step 1:

[1850] The server receives an incoming call from the terminal and obtains the phone number information. The input is the phone number received by the terminal, which is sent to the server. The server receives this phone number information and proceeds to the next step.

[1851] Step 2:

[1852] The server compares the acquired phone number information with existing blacklists and whitelists. The server has blacklists and whitelists stored in a database. The inputs are the received phone number and the blacklist and whitelist, and the comparison results in the output being no match, blacklist, or whitelist.

[1853] Step 3:

[1854] The device switches to automatic response mode based on the server's matching results. In particular, automatic response using a generative AI model is initiated for phone numbers that are not on the list. The server provides response instructions as input, and the device starts automatic response.

[1855] Step 4:

[1856] The generative AI model analyzes the content of the call in real time. The device sends the captured voice data to the server, which then converts it into text data. The input is the voice data of the call, and the output is text data generated based on that data.

[1857] Step 5:

[1858] The server uses the generative AI model to analyze the call content and detect keywords and phrases that may be fraudulent. The input is the generated text data, which is analyzed to extract fraud risk factors. The output is a risk score based on the fraud risk.

[1859] Step 6:

[1860] The server calculates a risk score based on the fraud detection results. The input is data on keywords and phrases that are suspected to be fraudulent, and the risk score is calculated based on this. The output is a specific numerical value for the risk score.

[1861] Step 7:

[1862] If the risk score exceeds the threshold, the server sends an instruction to the terminal to end the call. The inputs are the risk score and threshold information, and the output is an instruction to end the call.

[1863] Step 8:

[1864] The terminal receives a call termination instruction from the server and terminates the call. The input is a call termination instruction from the server, and the output is the end of the call.

[1865] Step 9:

[1866] The server adds suspected fraudulent phone numbers to a blacklist and synchronizes the updated blacklist to all relevant devices. The inputs are fraudulent phone numbers and blacklists, and a new blacklist is created. The output is the new blacklist synchronized to an online database.

[1867] Step 10:

[1868] If the call is deemed safe, the server adds the phone number to a whitelist and synchronizes the updated whitelist to all devices. The inputs are the phone numbers deemed safe and the whitelist, and a new whitelist is created. The output is the new whitelist synchronized to an online database.

[1869] These are the specific processing steps of this system. This series of steps enables fraudulent calls to be detected in real time and the necessary defensive measures to be taken automatically.

[1870] 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.

[1871] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[1872] System configuration

[1873] 1. Server

[1874] The server is the main processing unit that performs number verification when an incoming call is received, initiates an automated response, analyzes the call content in real time, calculates a risk score, recognizes the user's emotions through an emotion engine, and updates and synchronizes the list.

[1875] 2. Terminal

[1876] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The call content and the user's emotional information are sent to the server in real time.

[1877] 3. List

[1878] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[1879] Program processing

[1880] Response when receiving a call

[1881] Terminal: When a call comes in to the user's terminal, the number is sent to the server.

[1882] Server: The server checks the phone number against blacklists and whitelists and directs the appropriate action.

[1883] Launch of AI automated response

[1884] Terminal: If the number is not on either list, the terminal switches to auto-answer mode.

[1885] Server: The server uses generative artificial intelligence to initiate responses and analyze call content in real time.

[1886] Real-time analysis and risk assessment

[1887] Server: Generative AI converts call content into text data and detects keywords and phrases that may indicate fraud.

[1888] Server: The emotion engine recognizes the user's emotions in real time and analyzes them along with the call content.

[1889] Calculating the risk score

[1890] Server: Calculates a risk score based on the fraud detection results and the emotional information recognized by the emotion engine.

[1891] Server: Sends instructions to the terminal to terminate the call if the risk score exceeds a threshold.

[1892] End call and update list

[1893] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[1894] Terminal: If the caller number of the terminated call is to be added to the blacklist, it sends the corresponding information to the server.

[1895] List Synchronization

[1896] Server: Updates the blacklist and sends instructions to synchronize to all relevant communication devices.

[1897] On your device: Download the new blacklist and update the local database.

[1898] Safe Call Rating and Whitelist Updates

[1899] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[1900] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[1901] Specific examples

[1902] Example 1: Automatically detect fraudulent calls and terminate them

[1903] User: I get a call from an unknown number.

[1904] Terminal: Sends the phone number to the server.

[1905] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1906] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1907] Server: The generative AI detects the phrase "please prepare money," and the emotion engine detects the user's tension and anxiety. It then assigns a high risk score.

[1908] Server: Because the score exceeds the threshold, it sends an instruction to end the call to the terminal.

[1909] Terminal: The call is ended and the number is added to the blacklist.

[1910] Example 2: Secure Call Whitelisting Scenario

[1911] User: A call is coming from someone you trust.

[1912] Terminal: Sends the phone number to the server.

[1913] Server: Since it does not match the list, it instructs the AI ​​to automatically respond.

[1914] Terminal: An automatic response is initiated and the call contents are sent to the server.

[1915] Server: Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. It determines that there is no risk of fraud.

[1916] Server: The call is completed successfully and the phone number is added to the whitelist.

[1917] Device: Download the new whitelist.

[1918] In this way, the system of the present invention combines advanced analytical capabilities with user emotion recognition to automate responses to fraud and provide users with a safe and secure communication environment.

[1919] The processing flow will be explained below.

[1920] Step 1:

[1921] User: A call comes in on the user's device.

[1922] Terminal: Detects an incoming call and obtains the caller's phone number information.

[1923] Terminal: Sends the acquired phone number information to the server.

[1924] Step 2:

[1925] Server: Checks the received phone number information against existing blacklists and whitelists.

[1926] Server: If the call is on the blacklist, it sends an instruction to the device to reject the call.

[1927] Server: If the call is on the whitelist, it sends an instruction to the device to allow the call.

[1928] Step 3:

[1929] Terminal: If the phone number is not on either the blacklist or the whitelist, it switches to auto-answer mode.

[1930] Terminal: Transmits voice data and user emotion information to the server in real time.

[1931] Step 4:

[1932] Server: Generative AI analyzes the contents of the call in real time and converts it into text data.

[1933] Server: The emotion engine recognizes the user's emotions in real time and analyzes the emotion information along with the call content.

[1934] Step 5:

[1935] Server: Calculates a risk score based on the results of analyzing the call content and the user's emotion recognition.

[1936] Server: If a keyword or phrase is detected that is suspected to be fraudulent, set a higher risk score.

[1937] Step 6:

[1938] Server: If the risk score exceeds the set threshold, it sends an instruction to the terminal to terminate the call.

[1939] Server: Adds the phone number to the blacklist and syncs the update to all devices.

[1940] Step 7:

[1941] Terminal: Upon receiving an instruction to end the call, the terminal immediately ends the call.

[1942] Terminal: Apply the received updated blacklist to the local database.

[1943] Step 8:

[1944] Server: If the call is assessed as safe, add the caller ID to a whitelist and manage it.

[1945] Server: Sends instructions to synchronize whitelist updates to all devices.

[1946] On the device: Download the new whitelist and apply it to the local database.

[1947] Step 9:

[1948] Device: Prepare for the next incoming call by quickly assessing its trustworthiness using the latest blacklists and whitelists.

[1949] Example 2

[1950] 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."

[1951] In modern society, fraudulent activities are becoming increasingly sophisticated, making it difficult to protect personal information and property. It is difficult for users to respond immediately to fraudulent calls made through communication devices, and the mental burden is also great. Therefore, there is a need to provide an environment in which users can use communication devices with peace of mind.

[1952] 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.

[1953] In this invention, the server includes means for receiving calls from a communication device and acquiring phone number information, means for comparing the acquired phone number information with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for recognizing the user's emotions and calculating a risk score along with the fraud detection result, means for terminating the call if the risk score exceeds a threshold, and means for adding phone numbers suspected of being fraudulent to the blacklist and synchronizing the updated blacklist with all related communication devices. This makes it possible to automatically detect calls suspected of being fraudulent and prevent users from becoming victims of fraudulent calls.

[1954] A "communication device" is a device through which a user makes voice calls, and generally includes a mobile phone or a landline phone.

[1955] "Telephone number information" is data indicating the telephone number of the caller, and is used by the communication device to identify the caller.

[1956] A "blacklist" is a list of telephone numbers that have previously been the subject of fraudulent or illicit activity, and is used to prevent fraudulent calls.

[1957] A "whitelist" is a list of safe and trusted phone numbers that is used to accurately classify important calls.

[1958] "Generative artificial intelligence" refers to an algorithm or system that uses AI technology to generate natural, human-like dialogue and has the ability to analyze call content in real time.

[1959] An "emotion engine" is a system that uses voice recognition technology to analyze a user's emotional state and acquires that emotion as data.

[1960] A "risk score" is a quantified indicator of risk calculated based on the content of the call and the user's emotional state, and is used to assess the degree of suspected fraud.

[1961] "Synchronizing" refers to matching data between multiple communication devices, and refers to the process of reflecting updates to the blacklist or whitelist in all communication devices.

[1962] The present invention is a system that automatically detects and blocks suspicious fraudulent calls using generative artificial intelligence and an emotion engine for incoming calls made to a user's communication device. This system is realized by linking a telephone carrier's server with the user's communication device.

[1963] System configuration

[1964] server

[1965] The server is the main processing unit that performs real-time analysis of call content, calculates risk scores, recognizes user emotions using an emotion engine, and manages blacklists and whitelists. It performs these processes using generative artificial intelligence (AI) and an emotion engine.

[1966] Terminal

[1967] The user's communication device receives the incoming call and initiates an AI-powered automatic response in response to instructions from the server. The content of the call and the user's emotional information are sent to the server in real time.

[1968] list

[1969] The server stores blacklists and whitelists, and checks the phone number against these lists for each incoming call.

[1970] Program processing

[1971] Response when receiving a call

[1972] When a user receives a call, the number is sent to the server. The server checks the incoming number against a blacklist and a whitelist and instructs the appropriate action. If the number is not on either list, the device switches to auto-answer mode and the server begins responding using generative artificial intelligence. The content of the call is sent to the server in real time.

[1973] Real-time analysis and risk assessment

[1974] The server uses generative artificial intelligence to convert the call content into text data and detect keywords and phrases that may indicate fraud. At the same time, an emotion engine recognizes the user's emotions in real time and analyzes them along with the call content. A risk score is calculated based on the fraud detection results and the emotional information recognized by the emotion engine. If the risk score exceeds a threshold, the server sends an instruction to end the call to the terminal.

[1975] Ending a call and updating the list

[1976] Upon receiving the call termination instruction, the terminal immediately terminates the call and, if the caller number is to be added to the blacklist, sends corresponding information to the server. The server updates the blacklist and sends a synchronization instruction to all related communication devices. The terminal downloads the new blacklist and updates its local database.

[1977] Safe Call Rating and Whitelist Updates

[1978] If the call is deemed safe, the server adds the caller ID to the whitelist, and the device downloads the new whitelist from the server and prepares for the next incoming call.

[1979] Specific examples

[1980] Example 1: Automatically detect fraudulent calls and terminate them

[1981] The user receives a call from an unknown number. The device sends the phone number to the server. Since the server does not match the list, it instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. The generative AI detects the phrase "please prepare some money" and the emotion engine detects the user's tension or anxiety, resulting in a high risk score. If the score exceeds the threshold, the server sends an instruction to end the call to the device. The device ends the call and adds the number to the blacklist.

[1982] Example 2: Secure Call Whitelisting Scenario

[1983] The user receives a call from a trusted party. The device sends the phone number to the server. Since the call does not match the list, the server instructs an AI automatic answering system. An automatic answering system is initiated on the device, and the call content is sent to the server. Generative AI analyzes the call content, and the emotion engine recognizes that the user is relaxed. If it is determined that there is no risk of fraud, the call is terminated normally and the phone number is added to the whitelist. The device downloads the new whitelist and prepares for the next call.

[1984] Prompt Sentence Examples

[1985] "How does this system detect and block suspected fraudulent calls?"

[1986] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1987] Program processing flow

[1988] Step 1: Incoming call detection and number matching

[1989] Terminal: The user receives a call on their terminal. The phone number is sent to the server.

[1990] Input: Incoming call phone number

[1991] Output: Phone number data sent to the server

[1992] Specific operation: The moment a call comes in, the "+Notification" API is called and the phone number is sent to the server.

[1993] Step 2: List Matching

[1994] Server: Immediately checks the received phone number against blacklists and whitelists.

[1995] Input: Phone number data sent to the server

[1996] Output: List match result (existing in blacklist, existing in whitelist, not in either)

[1997] Specific operation: Rapidly search the incoming number against the blacklist and whitelist in the database.

[1998] Step 3: Start the AI ​​Automated Response

[1999] Server: If the number is not on either list, instruct the device to start an AI automatic response.

[2000] Input: List match result (not in either list)

[2001] Output: Instructions to start AI automatic response sent to the device

[2002] Specific operation: Launches the specified generative artificial intelligence (Generative AI) model and generates a response script.

[2003] Step 4: Start Auto-Answer Mode

[2004] Terminal: Switches to AI automatic response mode upon receiving instructions from the server.

[2005] Input: Instruction to start AI automatic response

[2006] Output: Start auto attendant

[2007] What it does: Loads an audio file for the auto attendant and keeps the call active.

[2008] Step 5: Real-time analysis of call content

[2009] Terminal: Sends voice data during a call to the server in real time.

[2010] Server: Generative AI converts the received speech into text data and detects keywords and phrases that may be fraudulent.

[2011] Input: Voice data during a call

[2012] Output: Real-time text data, detection of potentially fraudulent keywords and phrases

[2013] Specific operation:

[2014] Device: Uses WebRTC to send audio data to the server in streaming format.

[2015] Server: Uses the Speech-to-Text API to convert speech to text in real time and leverages natural language processing (NLP) techniques to perform text analysis.

[2016] Step 6: Real-time sentiment analysis

[2017] Server: Analyzes user emotions in real time using an emotion engine.

[2018] Input: Voice data during a call

[2019] Output: Analysis of the user's emotional state

[2020] Specific operation: The emotion engine analyzes the user's emotions from the voice in real time and generates emotion data.

[2021] Step 7: Risk Score and Call Evaluation

[2022] Server: Calculates risk scores based on the results of text analysis and the emotion engine's recognition data.

[2023] Input: Text analysis results, sentiment analysis results

[2024] Output: Risk score

[2025] Specific operation: Calculates a risk score based on an algorithm that takes into account fraud detection results and emotional information.

[2026] Step 8: End call instructions

[2027] Server: If the risk score exceeds the threshold, sends an instruction to end the call to the terminal.

[2028] Input: Risk score

[2029] Output: End call instruction

[2030] Specific operation: If the risk score exceeds a preset threshold, a call termination instruction is sent to the terminal using a RESTful API.

[2031] Step 9: End the call and update the blacklist

[2032] Terminal: Upon receiving the instruction to end the call, it immediately ends the call and reports the caller ID to the server.

[2033] Server: Add the received caller number to the blacklist and store the relevant information in the database.

[2034] Input: End call instruction, ended caller number

[2035] Output: Updated blacklist

[2036] Specific operation:

[2037] Terminal: Upon receiving a call end instruction, immediately execute the Hangup API.

[2038] Server: Updates the database to add the caller number to the blacklist and immediately block the next incoming call.

[2039] Step 10: Synchronizing the Blacklist

[2040] Server: Sends a synchronization instruction to all associated communication devices to update the blacklist.

[2041] On your device: Download the new blacklist and update the local database.

[2042] Input: Updated blacklist

[2043] Output: Synchronization instructions, updated local database

[2044] Specific operation:

[2045] Server: Notifies the communication device of blacklist updates using WebSocket.

[2046] Terminal: The new blacklist is retrieved via an API call and reflected in the local database.

[2047] Step 11: Update the whitelist

[2048] Server: If the call is assessed as safe, add the caller ID to the whitelist.

[2049] Device: Downloads the new whitelist from the server and prepares for the next incoming call.

[2050] Input: Caller ID after call rated safe

[2051] Output: Updated whitelist, updated local database on device

[2052] Specific operation:

[2053] Server: If the call is assessed as safe, add the number to the whitelist and update the database.

[2054] Device: Download the updated whitelist and apply it the next time you receive a call.

[2055] (Application example 2)

[2056] 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."

[2057] Telephone customer service on online shopping sites is an important means of communication for users. However, the number of calls suspected to be fraudulent is increasing, threatening user safety. To solve this problem, a system that automatically detects suspicious calls and allows only safe calls is needed. In addition, more accurate fraud detection is expected by performing a comprehensive risk assessment that takes user emotions into account.

[2058] 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.

[2059] In this invention, the server includes means for receiving calls from communication devices and acquiring phone number information, means for comparing the calls with existing blacklists and whitelists, means for initiating an automatic response using generative artificial intelligence for phone numbers not on the lists, means for the generative artificial intelligence to analyze the content of the call in real time and detect keywords or phrases that may be fraudulent, means for calculating a risk score based on the fraud detection results and the user's emotional information recognized by the emotion engine, means for terminating the call if the risk score exceeds a threshold, means for adding the phone number suspected of fraud to the blacklist, means for synchronizing the updated blacklist with all related communication devices, and means for converting the content of the call into text data in real time, thereby ensuring the security of customer service calls and reducing the risk of fraud.

[2060] "Communication device" is a general term for electronic devices used to make calls and send and receive messages.

[2061] An "incoming call" refers to an incoming call or message from an external device.

[2062] "Telephone number information" is a combination of numbers and codes for identifying a specific communication device.

[2063] A "blacklist" is a list of phone numbers suspected of being fraudulent or of unreliable callers.

[2064] A "whitelist" is a list of safe and trusted sources.

[2065] "Generative AI" is AI that has the ability to generate new information and answers based on input data.

[2066] "Auto-answer" is the ability of a communication device to automatically reply without human intervention.

[2067] "Real-time analysis" is the process of analyzing data as soon as it is received.

[2068] "Suspected fraudulent keywords and phrases" refer to specific words or expressions that suggest fraudulent activity.

[2069] A "risk score" is an evaluation value that quantifies the likelihood of fraud.

[2070] An "emotion engine" is software or an algorithm for analyzing and recognizing the emotions of a user of a communication device.

[2071] Converting to "text data" means converting formal data such as audio and video into text information.

[2072] "Updating" means replacing existing information with new information.

[2073] "Synchronization" is the process of matching data between multiple devices.

[2074] An "online shopping site" is a website for buying and selling products and services over the Internet.

[2075] "Customer service" refers to the support activities provided when customers inquire about products or services via communication devices.

[2076] To realize a fraud prevention system for customer service at a mail-order site, the present invention uses a communication device, a server, and necessary programs.

[2077] 1. System Configuration

[2078] server

[2079] The server performs the following main tasks:

[2080] When a call comes in, the phone number information is received and compared against the blacklist and whitelist.

[2081] If the phone number is not on the list, an automated response is initiated using generative artificial intelligence.

[2082] Calls are converted into text data in real time to detect keywords and phrases that may be fraudulent.

[2083] The emotion engine analyzes the user's emotions and calculates a risk score along with the fraud detection results.

[2084] If the risk score exceeds a threshold, the call is terminated and the suspected fraudulent phone number is added to a blacklist.

[2085] The updated blacklist is synchronized to all associated communication devices.

[2086] Communication equipment (user terminal)

[2087] The user's communication device has the following roles:

[2088] An incoming call is received and the telephone number information is sent to the server.

[2089] Receives instructions from the server and switches to auto-response mode.

[2090] The call contents and user emotional information are sent to the server in real time.

[2091] When an instruction to end the call is received, the call is immediately ended.

[2092] Download new blacklists and whitelists from the server and update the local database.

[2093] 2. Program processing explanation

[2094] The programs in the server and communication device operate using the following hardware and software:

[2095] Hardware:

[2096] Server: A computer system equipped with a high-performance processor

[2097] Communication devices: Smartphones and internet-enabled phones

[2098] software:

[2099] Generative AI: Hugging Face Transformer model (e.g. 'gpt-3')

[2100] Emotion Engine: Emotion recognition algorithm (e.g., EmotionRecognizer)

[2101] Speech Recognition: Speech to text conversion using the speech_recognition package

[2102] HTTP communication: Use the requests library to send and receive data

[2103] The server receives the phone number sent from the user's device and compares it with the blacklist and whitelist. If the phone number is not on the list, it uses generative artificial intelligence to initiate an automatic response and converts the call content into text data in real time. It then detects keywords and phrases that may be fraudulent, while an emotion engine simultaneously analyzes the user's emotional information. Based on this information, it calculates a risk score, and if the score exceeds a threshold, it sends an instruction to the user's device to terminate the call.

[2104] 3. Examples of concrete examples and prompts

[2105] Specific examples

[2106] When an unknown number calls the customer service of an online shopping site, the user's device sends the phone number to the server. The server checks the number and, if it is not in the list, it issues an automatic response. During the automatic response, the server analyzes the content of the call and

[2107] The emotion engine reads the user's emotions and calculates a risk score. If there is a high suspicion of fraud, the call is terminated and the number is added to a blacklist.

[2108] Prompt Sentence Examples

[2109] Prompts that pass call content through a generative AI model to detect fraudulent phrases

[2110] User: "Hello, is this customer service from an online shopping site?"

[2111] AI: "Hi, this is customer service. How can I help you?"

[2112] By inputting this prompt into a generative AI model, the call content begins to be analyzed and fraudulent phrases are detected.

[2113] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2114] Step 1:

[2115] A user receives an incoming call and the communication device acquires the phone number information. The communication device (user terminal) sends the received phone number information to the server. The server receives the phone number information as input and prepares to start processing.

[2116] Step 2:

[2117] The server compares the received phone number information with the existing blacklist and whitelist. The input is the phone number information, blacklist, and whitelist, and the output is the comparison result. Based on the comparison result, it checks whether the phone number is in the list.

[2118] Step 3:

[2119] If the matching result shows that the phone number is not on either the blacklist or the whitelist, the server instructs the communication device to switch to automatic response mode. The input is the matching result, and the output is the start of automatic response mode. The communication device receives this instruction and starts an automatic response using generative artificial intelligence.

[2120] Step 4:

[2121] When the automatic answering starts, the communication device sends the contents of the call to the server in real time. The input of the communication device is the voice data of the call contents, and the output is the voice data sent to the server. The server analyzes the received voice data and converts it from voice to text data.

[2122] Step 5:

[2123] The server then analyzes the converted text data using generative artificial intelligence to detect keywords and phrases that may be fraudulent. The input is text data, and the output is the detection results of keywords and phrases that may be fraudulent. This allows the server to identify potential fraudulent activity.

[2124] Step 6:

[2125] Meanwhile, the server uses an emotion engine to analyze the user's emotional information and determine the user's emotional state. The input is voice data and the output is emotional information. The emotional information serves to detect signs of anxiety or tension in the user.

[2126] Step 7:

[2127] The server combines the analysis results of the generative AI with the emotional information from the emotion engine to calculate a risk score. The input is the keyword and phrase detection results and emotional information, and the output is a risk score. This score quantifies the likelihood of fraud.

[2128] Step 8:

[2129] If the risk score exceeds the threshold, the server sends an instruction to the communication device to end the call. The input is the risk score, and the output is an instruction to end the call. The communication device receives this instruction and immediately ends the call.

[2130] Step 9:

[2131] After the call is completed, the server processes the suspected fraudulent phone number by adding it to the blacklist. The input is the phone number information, and the output is the updated blacklist. The server updates the blacklist and synchronizes it to all relevant communication devices.

[2132] Step 10:

[2133] The communication device downloads a new blacklist from the server and updates the local database. The input is the updated blacklist, and the output is the updated local database. This allows for a more robust response to future incoming calls.

[2134] Prompt Sentence Examples

[2135] Here's a prompt that feeds the call content to a generative AI model to detect scam phrases:

[2136] User: "Hello, is this customer service from an online shopping site?"

[2137] AI: "Hi, this is customer service. How can I help you?"

[2138] By inputting this prompt into a generative AI model, analysis of the call content begins and fraudulent phrases are detected.

[2139] 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.

[2140] 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.

[2141] 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.

[2142] 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.

[2143] 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.

[2144] 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.

[2145] 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).

[2146] 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.

[2147] 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."

[2148] 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.

[2149] 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).

[2150] 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.

[2151] 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.

[2152] 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.

[2153] 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.

[2154] 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.

[2155] 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.

[2156] 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.

[2157] 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.

[2158] 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 explana...

Claims

1. means for receiving an incoming call from a communication device and acquiring telephone number information; means for checking the obtained telephone number information against existing blacklists and whitelists; means for initiating an automatic response using generative artificial intelligence to a telephone number not present on the list; Generative artificial intelligence analyzes call content in real time to detect keywords and phrases that may be fraudulent, and means for calculating a risk score based on the fraud detection results; means for terminating the call if the risk score exceeds a threshold; A way to blacklist phone numbers suspected of fraud, and means for synchronizing the updated blacklist with all associated communication devices; A system including:

2. 10. The system of claim 1, further comprising means for adding the phone number to a whitelist if the call is assessed as safe.

3. 2. The system according to claim 1, further comprising means for converting the analyzed call content into text data in real time.

Images