system

Abstract

To provide a system.SOLUTION: A system comprises the means of: performing natural communication through an artificial intelligence; generating a fictitious contact address through the artificial intelligence; detecting approach from a criminal of scam to the fictitious contact address; guiding the criminal of scam to continue a criminal act; collecting an evidence of the criminal act; reporting the collected evidence to police; receiving a message sent by the criminal of scam and storing it to a database; analyzing the stored message through a natural language processing tool; extracting characteristic of the message from the analysis result; evaluating possibility of scam through inputting the extracted characteristic into the generative AI; and notifying a user of the message which has a high possibility of scam.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] Fraud crimes are currently on the rise, and fraud in the digital space is a particular problem. However, it is difficult to identify fraudsters and collect evidence, as well as to understand their behavioral patterns and implement preventative measures. [Means for solving the problem]

[0005] The invention uses artificial intelligence to communicate naturally and generate fictitious contacts. It detects approaches from fraudsters and allows them to escape, thereby collecting evidence of criminal activity. The collected evidence is reported to the police, which helps identify fraudsters and develop preventative measures. [Brief explanation of the drawings]

[0006] [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. 2 is a sequence diagram showing a flow of processing in the data processing system according to the first embodiment of the first form example. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Embodiment 1. [Figure 13] FIG. 10 is a sequence diagram showing a processing flow of a data processing system in a second embodiment of the second form example. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Embodiment Example 2. [Figure 15] FIG. 10 is a sequence diagram showing the flow of processing in a data processing system according to a third embodiment of the third embodiment. [Figure 16]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Embodiment 3. [Figure 17] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the first embodiment of the first form example when an emotion engine is combined. [Figure 18] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Form Example 1 when an emotion engine is combined. [Figure 19] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the second embodiment of the second form example when an emotion engine is combined. [Figure 20] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Form Example 2 when an emotion engine is combined. [Figure 21] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the third embodiment of the third form example when an emotion engine is combined. [Figure 22] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Form Example 3 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0007] 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.

[0008] First, the terms used in the following description will be explained.

[0009] 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, the 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), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0010] 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.

[0011] 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.

[0012] 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.

[0013] 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."

[0014] [First embodiment]

[0015] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0016] 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.

[0017] 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).

[0018] 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.

[0019] 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.

[0020] 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.

[0021] 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.

[0022] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0023] 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.

[0024] 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.

[0025] 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.

[0026] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0027] "Example 1"

[0028] One embodiment of the present invention is a system in which artificial intelligence (AI) can communicate naturally. This AI can use natural language processing (NLP) technology to have conversations similar to those of humans. The AI ​​also has the ability to generate fictitious contact information. This allows the AI ​​to detect attempts by fraudsters to contact the AI ​​and allow them to escape.

[0029] "Example 2"

[0030] Another embodiment of the present invention is a system for detecting approaches from fraudsters. This system detects approaches from fraudsters by analyzing the content of messages and behavioral patterns from fraudsters. Specifically, AI analyzes the wording, topic, and sending time of messages from fraudsters, and from this information, it understands the behavioral patterns of fraudsters.

[0031] "Example 3"

[0032] Furthermore, one embodiment of the present invention is a system that collects evidence of criminal activity and reports it to the police. AI collects information obtained through communication with fraudsters as evidence and reports it to the police as necessary. Specifically, AI collects information such as messages from fraudsters, the sender's IP address, and the time of transmission, and provides this information to the police. This can be useful in identifying fraudsters and preventing crimes.

[0033] The processing flow of each embodiment will be described below.

[0034] "Example 1"

[0035] Step 1: AI uses natural language processing (NLP) techniques to engage in human-like dialogue.

[0036] Step 2: The AI ​​generates fictitious contacts, which are digital accounts such as email addresses and social media accounts.

[0037] Step 3: If a fraudster attempts to contact the AI, the AI ​​will detect the attempt.

[0038] Step 4: The AI ​​lets the fraudsters go free while collecting evidence of their criminal activity.

[0039] "Example 2"

[0040] Step 1: AI analyzes the content of messages and behavioral patterns from fraudsters.

[0041] Step 2: The AI ​​analyzes the scammers' messages for wording, topic, and time of sending.

[0042] Step 3: Use this information to understand the behavioral patterns of fraudsters and detect their approaches.

[0043] "Example 3"

[0044] Step 1: The AI ​​collects evidence from communications with fraudsters.

[0045] Step 2: The AI ​​collects information from the fraudsters, such as the message, the IP address from which it was sent, and the time of sending.

[0046] Step 3: The AI ​​will provide the collected information to law enforcement, which can help identify fraudsters and prevent crime.

[0047] Example 1

[0048] Next, a description will be given of Example 1 of Form 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."

[0049] Conventional AI systems have had difficulty effectively detecting approaches from fraudsters and responding appropriately. They also lacked the means to generate fictitious contact information to lure fraudsters and collect evidence of criminal activity. Furthermore, they needed to be able to detect signs of fraud and generate appropriate responses while engaging in natural communication.

[0050] 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.

[0051] In this invention, the server includes means for conducting natural communication using artificial intelligence, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing input data using natural language processing technology, means for generating responses using a generative AI model, means for executing an algorithm for detecting signs of fraud, and means for sending the generated responses or fictitious contact information to the terminal. This makes it possible to effectively detect approaches from fraudsters and deal with them appropriately.

[0052] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0053] "Natural communication" is communication that mimics the dialogue and conversation that humans have on a daily basis and that occurs without any sense of incongruity.

[0054] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[0055] A "fraudster" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0056] "Approach detection methods" are techniques and methods used to identify and detect contact or attempts by fraudsters.

[0057] "Means to keep fraudsters going" are methods used to encourage fraudsters to continue committing crimes and to monitor their behavior.

[0058] "Means for collecting evidence of criminal activity" refers to techniques or methods that record the actions or communications of fraudsters so that they can later be used as evidence.

[0059] A "police reporting method" is a method of providing collected evidence to law enforcement agencies to assist in the investigation or prosecution of a crime.

[0060] "Natural language processing technology" is a technology that allows computers to understand and analyze human language.

[0061] A "generative AI model" is an artificial intelligence model that generates new text or responses based on input data.

[0062] A "fraud indicator detection algorithm" is a computational procedure designed to identify patterns or characteristics that indicate possible fraudulent activity.

[0063] A "terminal" is a device used by a user, such as a computer or smartphone.

[0064] MODE FOR CARRYING OUT THE INVENTION

[0065] This invention is a system that uses artificial intelligence to communicate naturally, detect approaches from fraudsters, and respond appropriately. This system works in cooperation with three parties: a server, a terminal, and a user.

[0066] Hardware and software used

[0067] Hardware: High-performance servers (e.g., virtual servers from cloud service providers)

[0068] Software: Natural language processing engines (e.g., natural language processing APIs), generative AI models (e.g., generative AI models), fraud detection algorithms

[0069] System configuration

[0070] 1. Server:

[0071] The server executes a program for natural communication using artificial intelligence.

[0072] The server analyzes the input data from the user using natural language processing techniques.

[0073] The server uses a generative AI model to generate an appropriate response.

[0074] The server runs algorithms that detect signs of fraud and detect approaches from fraudsters.

[0075] The server generates fictitious contact details to allow fraudsters to operate.

[0076] The server collects evidence of criminal activity and reports the collected evidence to the police.

[0077] 2. Terminal:

[0078] The terminal provides an interface for the user to enter input.

[0079] The terminal transmits the user's input data to the server.

[0080] The terminal receives the response or the fictitious contact from the server and displays it to the user.

[0081] 3. User:

[0082] A user uses a terminal to input questions or requests to the system.

[0083] The user checks the response from the server displayed on the terminal.

[0084] Specific examples

[0085] For example, if a user types "Hello, how's the weather today?", the device sends this input to the server. The server uses a natural language processing engine to parse the input data and uses a generative AI model to generate the response "Hello, the weather is sunny today." After verifying that there are no signs of fraud, the server sends this response to the device, which displays it to the user.

[0086] On the other hand, if a fraudster types "I want to add a new contact, how do I do that?", the server will detect the signs of fraud and generate a fictitious contact and send it to the device, which will then display this fictitious contact to the fraudster.

[0087] Prompt Sentence Examples

[0088] "Hello, how's the weather today?"

[0089] "I want to add a new contact, how do I do that?"

[0090] "I've been getting a lot of scam calls lately, what should I do?"

[0091] In this way, the present invention makes it possible to effectively detect approaches from fraudsters and deal with them appropriately.

[0092] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0093] Step 1:

[0094] The user enters input from the terminal.

[0095] The user uses the terminal's interface to input a question or request, for example, "Hello, what's the weather like today?" The input data is sent to the terminal in text form.

[0096] Step 2:

[0097] The terminal sends the input data to the server.

[0098] The terminal sends the text data entered by the user to the server via the Internet. At this time, the data is encrypted before being sent. The input data reaches the server.

[0099] Step 3:

[0100] The server receives the input data and analyzes it using a natural language processing (NLP) engine.

[0101] The server receives input data sent from the device. The received data is analyzed using a natural language processing engine. Specifically, the intent and sentiment of the input data are extracted. For example, the input "Hello, how's the weather today?" is analyzed and determined to be a question about the weather.

[0102] Step 4:

[0103] The server generates a response using a generative AI model based on the analysis results.

[0104] The server uses the generative AI model to generate an appropriate response based on the analysis results of the NLP engine. For example, if the analysis result is a question about the weather, the generative AI model will generate the response "Hello, the weather is sunny today." The generated response is stored in text format on the server.

[0105] Step 5:

[0106] The server runs fraud detection algorithms to detect signs of fraud.

[0107] Before sending the generated response to the user, the server runs a fraud detection algorithm. This algorithm detects whether the input data contains any signs of fraud. For example, if the input "I want to add a new contact, how do I do this?" indicates signs of fraud, the server will detect this. The detection results are stored on the server.

[0108] Step 6:

[0109] Send a server-generated response or fictitious contact to the device.

[0110] If fraud signs are detected, the server generates a fictitious contact and sends it to the device. If no fraud signs are detected, the server sends the generated response to the device as is. The transmitted data is encrypted before reaching the device.

[0111] Step 7:

[0112] The terminal displays the response from the server to the user.

[0113] The terminal displays the response or fictitious contact information received from the server to the user. The user checks the displayed information and decides the next action. For example, the response "Hello, the weather is sunny today" is displayed.

[0114] (Application example 1)

[0115] Next, a description will be given of Application Example 1 of Embodiment 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."

[0116] In recent years, fraud crimes have been increasing, especially frauds using digital communication. It is difficult to respond to such fraud crimes quickly and effectively using conventional methods. In addition, there are limited means to identify fraudsters and collect evidence. This has led to an increase in the number of victims, which has become a social problem.

[0117] 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.

[0118] In this invention, the server includes means for using artificial intelligence to perform natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for performing sentiment analysis, and means for notifying users using the fictitious contact information. This makes it possible to quickly detect fraudsters, collect evidence, and take appropriate action.

[0119] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0120] "Natural communication" refers to the natural verbal exchanges that humans engage in on a daily basis, and in particular to dialogues that are realized using natural language processing technology.

[0121] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[0122] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0123] "Approach detection methods" refers to technologies and methods used to detect contact or messages from fraudsters.

[0124] "Making fraudsters vulnerable" refers to deliberately maintaining contact with fraudsters in order to monitor their behavior and gather evidence.

[0125] "Evidence of criminal activity" refers to data or information that proves the fraudulent activities of fraudsters.

[0126] "Sentiment analysis" is a technique for analyzing emotions and the intensity of emotions from text data.

[0127] "Means of notifying users" refers to the methods and technologies used to communicate information or warnings detected by the system to users.

[0128] As an embodiment of this invention, we will explain how to install a fraud detection AI assistant system on a smartphone. This system uses artificial intelligence to communicate naturally, detects approaches from fraudsters, and generates fictitious contact information to allow fraudsters to escape.

[0129] System configuration

[0130] The system consists of the following main components:

[0131] 1. Artificial intelligence module: Uses natural language processing technology to communicate naturally with users.

[0132] 2. Fictitious Contact Generation Module: Generates fictitious contacts to deceive fraudsters.

[0133] 3. Fraud detection module: Analyzes message content and behavioral patterns to detect approaches from fraudsters.

[0134] 4. Sentiment Analysis Module: Analyzes the sentiment of the message and assesses its likelihood of fraud.

[0135] 5. Notification module: Notifies users of potential fraud and prompts them to take appropriate action.

[0136] Hardware and software used

[0137] Hardware: Smartphone

[0138] Software: Python, NLTK (Natural Language Toolkit)

[0139] Data processing and calculation

[0140] 1. Receiving a message: The system retrieves the message received by the user.

[0141] 2. Sentiment Analysis: Calculate the sentiment score of the message using NLTK's SentimentIntensityAnalyzer.

[0142] 3. Fraud detection: If the sentiment score exceeds a certain threshold, it is determined to be a possible fraud.

[0143] 4. Fictitious Contact Generation: Randomly generate fictitious contacts if fraud is suspected.

[0144] 5. User Notification: Notify users of potential scams and fictitious contacts.

[0145] Specific examples

[0146] For example, if a user receives a message saying "You've won the lottery! Please send us your bank details," the system can analyze the message and determine that it is likely a scam. It can then generate a fake contact and notify the user, saying "Suspicious activity has been detected. Please contact fake_contact_1@example.com for more information."

[0147] Prompt Sentence Examples

[0148] Examples of prompts to input to a generative AI model include:

[0149] "Analyze the following message for potential fraud: 'You have won a lottery! Please send your bank details.'"

[0150] Using this prompt, a generative AI model is asked to analyze the message and determine whether it is likely to be fraudulent.

[0151] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0152] Step 1:

[0153] The user receives a message. The terminal acquires the message received by the user. This message becomes the input data for the system.

[0154] Step 2:

[0155] The device sends the received message to the sentiment analysis module, which calculates the sentiment score of the message using NLTK's SentimentIntensityAnalyzer. This process outputs the message's sentiment score as positive, negative, or neutral.

[0156] Step 3:

[0157] The server analyzes the emotion scores and determines that fraud is likely if the negative emotion score exceeds a certain threshold. This determination result becomes the input data for the next step.

[0158] Step 4:

[0159] If the server determines that fraud is likely, it invokes a fictitious contact generation module, which randomly generates a fictitious contact and outputs the contact information.

[0160] Step 5:

[0161] The server notifies the user using the generated fictitious contact information. The notification module sends a message to the user informing them of the potential fraud and including the fictitious contact information. The notification is displayed on the user's device.

[0162] Step 6:

[0163] The user is notified and made aware of the potential scam, and can either continue to contact the scammer using the fictitious contact details provided, or take appropriate action.

[0164] Step 7:

[0165] The server monitors interactions between users and fraudsters and collects evidence of criminal activity, which is then stored for later reporting to law enforcement.

[0166] Step 8:

[0167] The server reports the collected evidence to the police. The reporting module organizes the evidence data and sends it to the police in the appropriate format. This process helps identify and apprehend fraudsters.

[0168] Example 2

[0169] Next, a description will be given of Example 2 of Form 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."

[0170] There is a need to quickly and accurately detect approaches from fraudsters and protect users. However, conventional systems have low accuracy in detecting fraudulent messages, putting users at high risk of becoming victims of fraud. In addition, analyzing and evaluating fraudulent messages takes time, making it difficult to respond in real time.

[0171] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving messages sent by fraudsters and storing them in a database, means for analyzing the stored messages using a natural language processing tool, means for extracting message characteristics from the analysis results, means for inputting the extracted characteristics into a generative AI model to evaluate the likelihood of fraud, and means for notifying users of messages that are likely to be fraudulent. This makes it possible to quickly and accurately detect approaches from fraudsters and protect users in real time.

[0172] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0173] "Natural communication" refers to exchanging information using natural language and non-verbal means, as humans do in their daily lives.

[0174] "Fictitious contacts" are non-existent contact methods created to attract fraudsters.

[0175] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0176] "Approach detection methods" are technologies and methods used to identify and detect contact or messages from fraudsters.

[0177] "Methods to let fraudsters go free" refers to the practice of monitoring fraudsters' activities and intentionally allowing them to continue their actions in order to gather evidence.

[0178] "Means of collecting evidence of criminal activity" are methods of recording the actions and messages of fraudsters to gather evidence for later use in legal proceedings.

[0179] A "police reporting method" is a method for providing collected evidence to law enforcement and reporting criminal activity.

[0180] A "database" is a system for storing and managing data in an organized manner.

[0181] "Natural language processing tools" are technologies and software that enable computers to understand and analyze human language.

[0182] A "generative AI model" is a model of artificial intelligence trained to perform a specific task, especially a generative task.

[0183] A "feature extraction method" is a method for identifying and extracting important patterns or attributes from data.

[0184] A "means for assessing the likelihood of fraud" is a technique or method for assessing and scoring the risk of fraud based on the extracted features.

[0185] A "means for notifying a user" is a method for informing a user of a detected risk of fraud.

[0186] This invention is a system for detecting approaches from fraudsters and protecting users. This system operates in cooperation with a server, terminals, and users.

[0187] The server receives messages sent by fraudsters and stores them in a database. Database management systems such as MySQL (registered trademark) and PostgreSQL are used for the database. The stored messages are analyzed using natural language processing tools. Specifically, natural language processing libraries such as "spaCy" and "NLTK" are used to tokenize the messages and tag them by part of speech.

[0188] From the analysis results, the server extracts message features. These features include specific keywords in the message (e.g., "bank," "account," "link," etc.) and the fact that the message was sent late at night. These features are input into a generative AI model. For example, OpenAI's GPT-4 (registered trademark) is used as the generative AI model.

[0189] The generative AI model evaluates the likelihood of fraud based on the input features and assigns a score. For example, the following prompt sentence is input into the generative AI model:

[0190] "Please rate the following message for possible fraud. It reads: 'Hello, there's a problem with your bank account. Click this link to check it now.'"

[0191] The generative AI model analyzes the message based on this prompt and scores the likelihood of it being fraudulent. If the score is high, the server sends a notification to the user. Notifications can be sent via email, in-app notifications, or other methods. For example, a push notification could be sent to the user's smartphone, warning them, "You have received a message that is likely to be fraudulent. Please be careful."

[0192] This system makes it possible to quickly and accurately detect approaches from fraudsters and protect users in real time.

[0193] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0194] Step 1: Receiving a message

[0195] The server retrieves messages received by the user. Inputs include messages from the user's email or chat app. The server receives these in real time and passes them on to the next processing step. Specifically, it retrieves messages from the email server or chat server via an API.

[0196] Step 2: Save the message

[0197] The server stores the received message in a database. The input is the message received in step 1. The output is the message stored in the database. Specifically, a database management system such as MySQL or PostgreSQL is used to store the message content and metadata (sender information, receipt time, etc.).

[0198] Step 3: Parse the message

[0199] The server analyzes the stored messages using natural language processing tools. The input is the message stored in step 2. The output is the analysis results. Specifically, it uses "spaCy" and "NLTK" to tokenize the messages and tag them as parts of speech.

[0200] Step 4: Feature extraction

[0201] The server extracts message features from the analysis results. The input is the analysis result obtained in step 3. The output is the extracted features. Specifically, features are extracted such as specific keywords in the message (such as "bank," "account," or "link"), or the fact that the message was sent late at night.

[0202] Step 5: Assess the likelihood of fraud

[0203] The server inputs the extracted features into a generative AI model to assess the likelihood of fraud. The input is the features extracted in step 4. The output is a score indicating the likelihood of fraud. Specifically, the following prompt sentence is input into the generative AI model:

[0204] "Please rate the following message for possible fraud. It reads: 'Hello, there's a problem with your bank account. Click this link to check it now.'"

[0205] The generative AI model analyzes the message based on this prompt and scores it for its likelihood of fraud.

[0206] Step 6: Notify users

[0207] The server notifies the user of messages that are likely to be fraudulent based on the score returned by the generative AI model. The input is the score obtained in step 5. The output is a notification to the user. Specifically, it sends a push notification to the user's smartphone, warning them, "You have received a message that is likely to be fraudulent. Please be careful."

[0208] In this way, a system is realized in which the server, terminal, and user work together to detect approaches from fraudsters and protect users.

[0209] (Application example 2)

[0210] Next, a description will be given of Application Example 2 of Form 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."

[0211] In recent years, fraudulent activities via messages by fraudsters have been increasing, resulting in many cases of users becoming victims. Conventional fraud prevention systems have had problems with low accuracy in detecting fraudulent messages and difficulty in displaying warnings in real time. In addition, there are insufficient means for users to report fraudulent messages, making it difficult to detect and prevent fraud early. To solve these problems, there is a need for a system that can detect fraudulent messages with greater accuracy in real time and display warnings to users.

[0212] 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.

[0213] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for monitoring a user's message application and detecting potentially fraudulent messages in real time, means for displaying a warning for detected potentially fraudulent messages, and means for the user to report fraudulent messages. This enables highly accurate detection of fraudulent messages and display of warnings in real time.

[0214] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0215] "Natural communication" refers to communication using natural language and in a conversational format, as humans do on a daily basis.

[0216] "Fictitious contacts" refer to digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[0217] A "fraud criminal" is someone who attempts to fraudulently obtain money or property by deceiving others.

[0218] "Means of approach detection" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to identify potential fraudulent contacts.

[0219] "Fraudster tactics" refers to techniques used to induce fraudsters to continue their activities and to gather evidence.

[0220] "Means of collecting evidence of criminal activity" refers to technology that records the actions and messages of fraudsters and gathers evidence that can later be used in legal proceedings.

[0221] "Law reporting" refers to techniques that provide collected evidence to law enforcement agencies to assist in the arrest and prosecution of fraudsters.

[0222] "Messaging Application" means software that allows a User to send and receive text messages.

[0223] "Real-time detection measures" refers to technology that instantly analyzes potentially fraudulent messages and displays a warning to users.

[0224] "Warning measures" refers to technologies that notify users of potentially fraudulent messages and warn them.

[0225] "Means for reporting fraudulent messages" refers to technology that allows users to report potentially fraudulent messages they receive to the system, contributing to improving the accuracy of AI models.

[0226] As an embodiment of the present invention, a method for installing a fraud prevention assistant system on a smartphone will be described. The system monitors the user's messaging application, detects potentially fraudulent messages in real time, and displays a warning.

[0227] System configuration

[0228] The system consists of the following main components:

[0229] 1. Artificial Intelligence (AI) module: Contains machine learning models for detecting fraudulent messages.

[0230] 2. Message monitoring module: Monitor users' message applications in real time.

[0231] 3. Warning display module: Displays a warning to the user when a potentially fraudulent message is detected.

[0232] 4. Reporting module: Provides an interface for users to report scam messages.

[0233] Hardware and software used

[0234] Hardware: Smartphone

[0235] Software: Python, scikit-learn, joblib

[0236] Data processing and calculation

[0237] 1. Message preprocessing: Lowercase the message and remove special characters.

[0238] 2. Vectorization: Use TfidfVectorizer to convert the message into a numeric vector.

[0239] 3. Predict: Use a pre-trained Logistic Regression model to predict whether the message is fraudulent or not.

[0240] 4. Generate analysis results: Generate analysis results including message sending times and prediction results.

[0241] 5. Display warning: If a potentially fraudulent message is detected, a warning will be displayed to the user.

[0242] Specific examples

[0243] Message received by users: "Your account has been compromised. Click this link now."

[0244] The application detects this and displays a warning: "Warning: A potentially fraudulent message has been detected. Details: {'message': 'Your account has been compromised. Click this link now.', 'send_time': '2023-10-01 12:34:56', 'is_fraud': True}"

[0245] Prompt Sentence Examples

[0246] Generate Python code to develop a fraud prevention assistant app. This app monitors users' messaging apps and detects potentially fraudulent messages in real time. It uses a pre-trained Logistic Regression model and TfidfVectorizer to detect fraudulent messages.

[0247] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0248] Step 1:

[0249] The user receives the message.

[0250] Input: A new message that arrives in the user's messaging application.

[0251] Specific operation: A new message is received by the user's smartphone, and the message monitoring module detects the message.

[0252] Step 2:

[0253] The terminal preprocesses the message.

[0254] Input: The received message.

[0255] Data manipulation: Lowercase the message and remove special characters.

[0256] Output: The preprocessed message.

[0257] What happens: The device converts the message text to lowercase and removes special characters and unnecessary spaces.

[0258] Step 3:

[0259] The terminal vectorizes the preprocessed messages.

[0260] Input: The preprocessed message.

[0261] Data operation: Convert the message into a numeric vector using TfidfVectorizer.

[0262] Output: Vectorized messages.

[0263] Specific operation: The terminal uses TfidfVectorizer to convert the preprocessed message into a numeric vector.

[0264] Step 4:

[0265] The device uses an AI model to predict the vectorized message.

[0266] Input: Vectorized message.

[0267] Data Computation: Use a pre-trained Logistic Regression model to predict whether a message is fraudulent.

[0268] Output: Prediction results of fraudulent messages.

[0269] Specific operation: The device uses a logistic regression model to predict whether the vectorized message is a fraudulent message.

[0270] Step 5:

[0271] The device analyzes the prediction results and displays a warning.

[0272] Input: Predicted results of scam messages.

[0273] Data processing: Generate analysis results, including message sending times and prediction results.

[0274] Output: The warning message that is displayed to the user.

[0275] Specific operation: The device analyzes the prediction results and displays a warning message to the user if there is a possibility of fraud.

[0276] Step 6:

[0277] A user reports a fraudulent message.

[0278] Input: The scam message reported by the user.

[0279] Data processing: Reported messages are recorded and used to improve the accuracy of AI models.

[0280] Output: The updated AI model.

[0281] How it works: A user reports a fraudulent message, and the device records the message and adds it to the training data for the AI ​​model.

[0282] Example 3

[0283] Next, a description will be given of a third embodiment of the third embodiment. 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."

[0284] Fraud crimes are becoming more sophisticated every year, making it difficult to identify fraudsters and collect evidence using conventional methods. It is also necessary to quickly detect approaches from fraudsters and take appropriate action. Furthermore, a system is needed to efficiently report collected evidence to the police.

[0285] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0286] In this invention, the server includes means for using artificial intelligence to conduct natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for storing the collected evidence in a database, means for analyzing messages using a generative AI model, and means for reporting suspected fraud to the police. This makes it possible to quickly detect approaches from fraudsters, efficiently collect and store evidence, and report to the police.

[0287] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0288] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to attract fraudsters.

[0289] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to determine the possibility of fraud.

[0290] "Fraudster tactics" are techniques used to induce fraudsters to continue their activities and gather more evidence.

[0291] "Means of collecting evidence of criminal activity" refers to technology that collects information such as the content of messages from fraudsters, the IP address of the sender, and the time of sending.

[0292] "Means for storing collected evidence in a database" refers to a technique for storing collected evidence information in a database so that it can be referenced later.

[0293] A "generative AI model" is an artificial intelligence model trained to perform natural language processing and data analysis.

[0294] "Means for analyzing messages" refers to technology that uses a generative AI model to analyze the content of received messages and determine whether they are fraudulent.

[0295] "Means of reporting to the police" refers to technology that provides collected evidence information to the police to help identify fraudsters and prevent crime.

[0296] MODE FOR CARRYING OUT THE INVENTION

[0297] The present invention relates to a system for detecting approaches from fraudsters, collecting evidence, and reporting to the police. A specific embodiment of this system will be described below.

[0298] System configuration

[0299] This system consists of three main components: a server, a terminal, and a user. The server receives messages from fraudsters, analyzes them, collects evidence, and reports them to the police. The terminal is a device that users use to access and operate the system. Users use the system to monitor approaches from fraudsters and take necessary action.

[0300] Hardware and software used

[0301] The server is a computer system equipped with a high-performance processor and a large amount of memory. A generative AI model (e.g., GPT-4) and a database (e.g., MySQL) are installed on the server. The generative AI model is used to analyze messages from fraudsters. The database is used to store collected evidence information.

[0302] Data processing and calculation

[0303] When the server receives a message from a fraudster, it first inputs the message into a generative AI model. The generative AI model analyzes the message's content and determines whether it is likely to be fraudulent. The server then collects metadata, such as the message's content, the sender's IP address, and the time of sending, and stores this information in a database. Finally, if the server suspects fraud based on the analysis results, it reports the collected evidence to the police.

[0304] Specific examples

[0305] For example, consider the case where a fraudster sends a phishing email to a user. The server collects the content of the phishing email, the sender's IP address, and the time of sending. The generative AI model analyzes this information and determines that it is suspected to be phishing. The server then reports the analysis results and the collected evidence to the police.

[0306] Prompt Sentence Examples

[0307] An example of a prompt sentence to input to the generative AI model is as follows:

[0308] Please analyze this message for possible fraud. The message reads as follows:

[0309] [Message content]

[0310] In this way, a system can be constructed in which the server can efficiently collect information from fraudsters and report it to the police. The flow of the identification process in the third embodiment will be explained with reference to FIG.

[0311] Step 1:

[0312] The server receives messages from fraudsters. The input is the message sent by the fraudster. The server periodically checks for new emails from the mail server and retrieves emails that are suspected to be fraudulent. The output is the received message data.

[0313] Step 2:

[0314] The server inputs the received message into the generative AI model for analysis. The input is the received message data. The generative AI model (e.g., GPT-4) analyzes the content of the message and determines whether it is fraudulent. The output is the analysis result. Specifically, the following prompt sentence is input into the generative AI model:

[0315] Please analyze this message for possible fraud. The message reads as follows:

[0316] [Message content]

[0317] Step 3:

[0318] The server collects not only the message content but also metadata such as the sender's IP address and the time of sending. The input is the received message data. The server extracts the sender's IP address and the time of sending from the email header information and collects this information as evidence. The output is the collected evidence information.

[0319] Step 4:

[0320] The server stores the collected evidence information in a database. The input is the collected evidence information. The server connects to a database (e.g., MySQL) and executes SQL queries to store the evidence information. The output is the evidence information stored in the database.

[0321] Step 5:

[0322] The server determines whether fraud is suspected based on the analysis results of the generative AI model. The input is the analysis results of the generative AI model. The server checks the analysis results and flags any suspected fraud. The output is the determination result of whether fraud is suspected.

[0323] Step 6:

[0324] If the server determines that fraud is suspected, it reports the collected evidence to the police. The input is the collected evidence and the judgment result. The server creates and sends a request to the police API to send the evidence. The output is the evidence reported to the police.

[0325] In this way, the server can efficiently collect, analyze, and report information from fraudsters to law enforcement.

[0326] (Application example 3)

[0327] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[0328] In modern society, fraud crimes are becoming increasingly sophisticated, requiring victims to respond quickly and effectively when they receive fraudulent messages. However, traditional methods require time and effort to analyze fraudulent messages and gather evidence, and reporting to the police is often delayed. This poses a challenge in identifying fraudsters and preventing crimes.

[0329] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0330] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing fraudulent messages and collecting evidence if there is a possibility of fraud, and means for transmitting the collected evidence to the police's API. This makes it possible to automatically analyze fraudulent messages when they are received, quickly collect evidence, and report it to the police.

[0331] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0332] "Natural communication" refers to interfaces and processes that allow humans and machines to interact seamlessly.

[0333] "Fictitious contacts" refer to digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[0334] A "fraud criminal" is someone who attempts to fraudulently obtain money or property by deceiving others.

[0335] "Means of approach detection" refers to technologies and methods that analyze messages and behavioral patterns from fraudsters to determine the likelihood of fraud.

[0336] "Fraudster tactics" refers to techniques and methods used to induce fraudsters to provide more information and to gather evidence.

[0337] "Evidence of criminal activity" refers to data such as message content, source IP address, and time of transmission that can be used to prove fraudulent activities committed by fraudsters.

[0338] "Means of reporting to police" refers to the techniques and methods used to provide collected evidence to police.

[0339] "Means for analyzing fraudulent messages" refers to techniques or methods for analyzing received messages and determining whether they are fraudulent.

[0340] "Means of sending to police API" refers to the technology or method for automatically sending collected evidence to police systems.

[0341] As an embodiment of the present invention, the following system is constructed.

[0342] First, the server has a means for natural communication using artificial intelligence. This artificial intelligence uses a generative AI model to collect information through dialogue with fraudsters. The server also has a means for generating fictitious contact information, which allows it to create email addresses and social networking service accounts to lure fraudsters.

[0343] Next, the server is equipped with a means for detecting approaches from fraudsters. This means determines the possibility of fraud by analyzing the content of messages and behavioral patterns from fraudsters. The server also includes a means for letting fraudsters go, encouraging them to provide more information.

[0344] Furthermore, the server has a means for collecting evidence of criminal activity. This evidence includes the fraudulent message, the source IP address, the time of transmission, etc. The collected evidence is provided to the police through a means for reporting. Specifically, the server has a means for analyzing the fraudulent message and collecting evidence if there is a possibility of fraud. The collected evidence is automatically reported to the police through a means for sending it to the police's API.

[0345] To implement this system, programs are written using programming languages ​​such as Python. The server automatically analyzes fraudulent messages when they are received, quickly collects evidence, and reports them to the police. The hardware used is a smartphone, and the software uses Python. The AI ​​model uses a pre-trained generative AI model.

[0346] For example, if a user receives a message saying, "Your bank account has been frozen. Click here for more information," the system automatically analyzes the message and determines it may be fraudulent. It then collects the message content, the sending IP address, and the time of receipt as evidence and reports it to the police.

[0347] An example of a prompt is as follows:

[0348] "When a fraudulent message is received, please create an application that analyzes the message and, if there is a possibility of fraud, collects evidence and reports it to the police. The message content, source IP address, and time of receipt will be collected as evidence and sent to the police API."

[0349] The above is an embodiment of the present invention.

[0350] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0351] Step 1:

[0352] A user receives a fraudulent message.

[0353] Input: Scam message

[0354] Output: Content of the scam message

[0355] Specific operation: The user's smartphone receives the fraudulent message and sends its contents to the server.

[0356] Step 2:

[0357] The server analyzes the fraudulent message.

[0358] Input: The content of the scam message

[0359] Output: Possibility of fraud judgement result

[0360] How it works: The server uses a generative AI model to analyze the message content and determine whether it is potentially fraudulent.

[0361] Step 3:

[0362] If the server determines that fraud is possible, it will collect evidence.

[0363] Input: Fraudulent message content, source IP address, and time of receipt

[0364] Output: Collected evidence (message content, source IP address, received time)

[0365] Specific operation: The server records the content of the fraudulent message, the sending IP address, and the time of receipt, and collects these as evidence.

[0366] Step 4:

[0367] The server sends the collected evidence to the police API.

[0368] Input: Collected evidence (message content, source IP address, time of receipt)

[0369] Output: Police report result (success or failure)

[0370] What it does: The server sends the collected evidence to the police API and checks whether the report was successful.

[0371] Step 5:

[0372] The server notifies the user of the report results.

[0373] Input: Police report result (success or failure)

[0374] Output: Notification to the user

[0375] Specific operation: The server notifies the user of the results of the police report. If the report is successful, it notifies the user by saying "The report has been submitted to the police." If the report is unsuccessful, it notifies the user by saying "The report failed."

[0376] 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.

[0377] "Example 1"

[0378] One embodiment of the present invention is an artificial intelligence system incorporating an emotion engine. This system recognizes emotions from a user's text messages and voice and adjusts the AI's response based on the emotions. For example, if a user sends a message expressing anger or frustration, the emotion engine recognizes this and the AI ​​generates a response that calms the user. Alternatively, if a user sends a message expressing joy or satisfaction, the emotion engine recognizes this and the AI ​​generates a response that shares the user's joy.

[0379] "Example 2"

[0380] The emotion engine can also track changes in a user's emotions over time, allowing the AI ​​to generate more appropriate responses that reflect the user's changing emotions. For example, if a user initially expresses joy but becomes frustrated over time, the emotion engine will recognize this change and the AI ​​will generate a response that reflects the user's frustration.

[0381] "Example 3"

[0382] Furthermore, the emotion engine has a learning function for recognizing the user's emotions. This learning function allows the emotion engine to learn how the user expresses their emotions and improve the accuracy of emotion recognition over time. For example, if a particular user uses specific words and expressions to express joy, the emotion engine can learn this and become able to more accurately recognize the joy of that user.

[0383] The processing flow of each embodiment will be described below.

[0384] "Example 1"

[0385] Step 1: A text message or voice message from the user is entered into the system.

[0386] Step 2: The emotion engine recognizes the user's emotion from the input.

[0387] Step 3: Based on the recognized emotion, the AI ​​generates a response.

[0388] "Example 2"

[0389] Step 1: A text message or voice message from the user is entered into the system.

[0390] Step 2: The emotion engine recognizes the user's emotions from the input and tracks changes in those emotions over time.

[0391] Step 3: Generate an AI response that corresponds to the change in emotion.

[0392] "Example 3"

[0393] Step 1: A text message or voice message from the user is entered into the system.

[0394] Step 2: The emotion engine recognizes the user's emotions from the input and learns how to express those emotions.

[0395] Step 3: Based on the learned emotional expression methods, the emotion engine improves the accuracy of emotion recognition.

[0396] Example 1

[0397] Next, a description will be given of Example 1 of Form 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."

[0398] Conventional AI systems struggled to communicate naturally and were unable to properly recognize user emotions and generate responses. Furthermore, they lacked the means to detect approaches from fraudsters and respond appropriately, making it difficult to prevent fraud and collect evidence. This made it difficult to ensure user safety and led to a risk of increasing fraud victims.

[0399] 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.

[0400] In this invention, the server includes means for conducting natural communication using artificial intelligence, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing messages using natural language processing technology, means for recognizing user emotions using an emotion engine, and means for generating responses based on the recognized emotions. This enables natural communication with users, effectively detecting approaches from fraudsters, and responding appropriately. Furthermore, recognizing user emotions and generating responses enables more human-like interactions and improves user satisfaction.

[0401] "Artificial intelligence" is the technology that enables computer systems to learn, reason, and self-correct by imitating human intelligence.

[0402] "Natural communication" refers to a dialogue between a human and a computer system using natural language, exchanging information in a manner similar to a dialogue between humans.

[0403] "Fictitious contacts" are non-existent contact information and digital accounts, such as email addresses or social media accounts, created to deceive fraudsters.

[0404] A "fraudster" is a person or organization that attempts to fraudulently obtain money or information by deceiving others.

[0405] "Approach detection" refers to technologies and algorithms that analyze contacts and messages from fraudsters to identify their intent.

[0406] "Letting fraudsters run wild" refers to intentionally maintaining contact with fraudsters in order to monitor their behavior and gather evidence.

[0407] "Means of collecting evidence of criminal activity" refers to technology that records the actions and messages of fraudsters and collects data for later use in legal proceedings.

[0408] A "police reporting method" is a method for providing collected evidence to law enforcement agencies and requesting appropriate action.

[0409] "Natural language processing technology" is a technology that enables computers to understand and generate human language, and involves analyzing and generating text.

[0410] An "emotion engine" is a technology that recognizes emotions from a user's text or voice and adjusts responses based on those emotions.

[0411] A "response generator" is a technique or algorithm that creates an appropriate reply based on the user's input and sentiment.

[0412] The present invention is a system for natural communication using artificial intelligence, which can detect approaches from fraudsters and respond appropriately. It can also recognize the user's emotions and generate responses based on those emotions. This system is implemented using the following hardware and software.

[0413] Hardware and software used

[0414] 1. Server:

[0415] The server hosts a generative AI model (e.g., OpenAI's GPT-4) and parses messages using natural language processing techniques.

[0416] The server uses an emotion engine to recognize the user's emotions and generate an appropriate response.

[0417] The server runs a fraud detection algorithm to detect approaches from fraudsters.

[0418] 2. Terminal:

[0419] A terminal is a device that allows a user to input and send messages, such as a smartphone or a computer.

[0420] The terminal receives the response from the server and displays it to the user.

[0421] 3. User:

[0422] Users interact with the system using terminals: they send text and voice messages and receive responses from the system.

[0423] Program processing

[0424] The server receives messages sent by users and analyzes the content of the messages using natural language processing techniques. Specifically, it performs tokenization, grammatical analysis, and semantic analysis of the messages. The server then uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the content and tone of the messages to identify the user's emotional state.

[0425] Based on the recognized emotion, the server uses a generative AI model to generate an appropriate response. The generated response is then tailored to create a natural dialogue according to the user's emotion. If an approach from a fraudster is detected, the server generates a fictitious contact and allows the fraudster to escape. This allows for the collection of evidence of criminal activity and the reporting of the collected evidence to the police.

[0426] Specific examples

[0427] For example, if a user sends a message saying "I'm very tired today," the following processing occurs:

[0428] 1. A user types "I'm so tired today" in a chat app and sends it.

[0429] 2. The server receives the message and performs tokenization, grammar analysis, and semantic analysis.

[0430] 3. The server uses an emotion engine to recognize "fatigue."

[0431] 4. The server generates a response such as "Take it easy today."

[0432] 5. The server sends the response to the user's device.

[0433] 6. The user sees the response "Take it easy today" on their device.

[0434] Prompt Sentence Examples

[0435] Examples of prompts to be input to a generative AI model include:

[0436] User: I'm very tired today.

[0437] AI: Please take it easy today.

[0438] In this way, AI can communicate naturally according to the user's emotions.

[0439] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0440] Step 1:

[0441] The server launches the AI ​​model.

[0442] The server loads and initializes a generative AI model (e.g., OpenAI's GPT-4). The server reads the model's parameters and settings and prepares it for interaction.

[0443] Input: AI model library and configuration file

[0444] Output: Initialized AI model

[0445] Specific operation: The server imports the AI ​​model library and initializes the model. The server loads the model settings.

[0446] Step 2:

[0447] A user sends a message.

[0448] Users use devices such as smartphones and personal computers to send text messages or voice messages to the system.

[0449] Input: User's text or voice message

[0450] Output: Message sent from terminal to server

[0451] Specific behavior: A user opens a chat app, types a message, and presses the send button.

[0452] Step 3:

[0453] The server receives the message and performs NLP processing.

[0454] The server receives messages sent by users and analyzes the content of the messages using natural language processing (NLP) techniques, specifically tokenizing, grammatical analysis, and semantic analysis of the messages.

[0455] Input: Message from the user

[0456] Output: Structure and meaning of the parsed message

[0457] Specific operation: The server calls the API to receive the message, and performs tokenization, grammatical analysis, and semantic analysis on the message.

[0458] Step 4:

[0459] The server uses an emotion engine to recognize emotions.

[0460] The server recognizes emotions from the user's messages using an emotion engine, which analyzes the content and tone of the messages to identify the user's emotional state.

[0461] Input: Parsed message structure and meaning

[0462] Output: Perceived emotional state

[0463] Specific operation: The server invokes the emotion engine to perform emotion analysis of the message and identify the emotional state.

[0464] Step 5:

[0465] The server generates an appropriate response.

[0466] The server generates appropriate responses based on the recognized emotions, using a generative AI model to engage in natural dialogue based on the user's emotions.

[0467] Input: Perceived emotional state

[0468] Output: The generated response

[0469] Specific operation: The server inputs a prompt sentence into the generative AI model, obtains the generated response, and formats it into an appropriate format.

[0470] Step 6:

[0471] The server generates fictitious contacts (in case of fraud detection).

[0472] If a fraudster attempts to make contact, the server uses fraud detection algorithms to detect the attempt and generate fictitious contact information.

[0473] Input: Results of the fraud detection algorithm

[0474] Output: Generated fictitious contacts

[0475] What it does: The server runs a fraud detection algorithm to detect fraud attempts and generate fictitious contacts.

[0476] Step 7:

[0477] The server sends the response to the user.

[0478] The server sends the generated response to the user's device, where the user can check the response from the AI.

[0479] Input: Generated response

[0480] Output: The response sent to the user's terminal

[0481] Specific operation: The server calls the API to send a response, and sends the response to the user's device. The user checks the response on the device.

[0482] (Application example 1)

[0483] Next, a description will be given of Application Example 1 of Embodiment 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."

[0484] Conventional AI systems have had difficulty effectively detecting approaches from fraudsters and protecting users. Furthermore, they lacked the ability to recognize users' emotions and generate appropriate responses based on those emotions, making it difficult to reduce user stress. This has led to a need to improve user safety and satisfaction.

[0485] 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.

[0486] In this invention, the server includes means for using artificial intelligence to perform natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing a user's emotions and adjusting a response based on the emotions, means for detecting possible fraud and displaying a warning to the user, and means for generating an appropriate response based on the emotions. This makes it possible to effectively detect approaches from fraudsters, protect the user, and generate an appropriate response according to the user's emotions.

[0487] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0488] "Natural communication" refers to using language and interacting in a similar way to humans.

[0489] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[0490] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0491] "Approach detection methods" are technologies used to identify and detect contact and activity from fraudsters.

[0492] "Fraudsters' lethal measures" are methods used by fraudsters to gather evidence while making it appear as if they are continuing their criminal activities.

[0493] "Evidence of criminal activity" is data or information that shows fraudulent activity committed by fraudsters.

[0494] A "police reporting method" is a method for providing collected evidence to law enforcement.

[0495] "Means for recognizing emotions" refers to technology that analyzes and identifies emotions from a user's text messages and voice.

[0496] A "means for tailoring responses" is a technique for generating appropriate responses based on recognized emotions.

[0497] "Potential fraud detection methods" are technologies that analyze messages and behavioral patterns to identify fraud risks.

[0498] A "means for displaying a warning" is a method for informing users of the risk of fraud.

[0499] "Means for generating appropriate responses" refers to technology that creates responses that correspond to the user's emotions and circumstances.

[0500] A system for implementing the present invention includes means for using artificial intelligence to communicate naturally, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing a user's emotions and adjusting responses based on the emotions, means for detecting possible fraud and displaying a warning to the user, and means for generating an appropriate response based on the emotions.

[0501] The server uses OpenAI's API to detect possible fraud and TextBlob to analyze the user's emotions. If there is a possibility of fraud, the server displays a warning to the user and generates an appropriate response based on the user's emotions. The hardware used is a smartphone, and the software used is OpenAI API and TextBlob.

[0502] For example, if a user receives the message "Your bank account has been fraudulently accessed. Please take action now," the server will detect the possibility of fraud and warn the user. Alternatively, if a user sends the message "I'm so happy today!", the server will use an emotion engine to recognize joy and generate an appropriate response.

[0503] Examples of prompts to input to a generative AI model include:

[0504] Fraud Detection: "Is this message potentially fraudulent?: Your bank account has been compromised. Take action now."

[0505] Emotional response: "The user is expressing the emotion of happiness. Generate an appropriate response."

[0506] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0507] Step 1:

[0508] The user receives the message.

[0509] The user's terminal sends the received message to the server.

[0510] Input: Message received by the user

[0511] Output: Message sent to the server

[0512] Step 2:

[0513] The server analyzes the received messages to detect possible fraud.

[0514] The server uses the OpenAI API to prompt the message content and determine the likelihood of fraud.

[0515] Input: The message sent to the server

[0516] Output: Judgment result on likelihood of fraud

[0517] Step 3:

[0518] The server will warn the user if there is a possibility of fraud.

[0519] The server generates a warning message and sends it to the user's terminal.

[0520] Input: Determination result regarding the possibility of fraud

[0521] Output: A warning message that is displayed to the user.

[0522] Step 4:

[0523] The server analyzes the sentiment of the received messages.

[0524] The server uses the TextBlob to parse the message for emotion and determine the type and intensity of the emotion.

[0525] Input: The message sent to the server

[0526] Output: Analysis results on the type and intensity of emotions

[0527] Step 5:

[0528] The server generates an appropriate response based on the emotion.

[0529] The server uses a generative AI model to input emotional prompts and generate appropriate responses.

[0530] Input: Analysis results on type and intensity of emotions

[0531] Output: The appropriate response generated

[0532] Step 6:

[0533] The server sends the generated response to the user's terminal.

[0534] The user's terminal displays the received response to the user.

[0535] Input: The appropriate response generated

[0536] Output: The response that is displayed to the user

[0537] Example 2

[0538] Next, a description will be given of Example 2 of Form 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."

[0539] Conventional fraud detection systems only analyze the content of messages and behavioral patterns from fraudsters, and are unable to respond to changes in users' emotions. This makes it difficult to understand how users feel about fraud and take appropriate action. In addition, there are limited means to improve the accuracy of fraud detection. This makes it difficult to fully ensure user safety.

[0540] 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.

[0541] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go free, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for tracking changes in the user's emotions, and means for generating responses corresponding to the user's emotions. This makes it possible to detect approaches from fraudsters with high accuracy and to respond appropriately according to the user's emotions.

[0542] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and perform tasks such as learning, reasoning, and problem-solving.

[0543] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[0544] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to assess the likelihood of fraud.

[0545] "Methods of letting fraudsters go free" refers to a technique for collecting more evidence by not taking immediate action against fraudsters but by allowing them to continue certain behaviors.

[0546] "Means for collecting evidence of criminal activity" refers to technology that records the actions and message content of fraudsters and stores them in a form that can later be used in legal proceedings.

[0547] "Means for reporting collected evidence to the police" refers to techniques for providing collected evidence in an appropriate format to the police or other law enforcement agencies.

[0548] The "means for tracking changes in user emotions" is a technology that collects user input and behavioral data and analyzes changes in user emotions using an emotion analysis model.

[0549] "Means for generating responses that correspond to the user's emotions" refers to technology that uses a generative AI model to create messages that match the user's emotions and provide an appropriate response.

[0550] MODE FOR CARRYING OUT THE INVENTION

[0551] The present invention is a system for detecting approaches from fraudsters and responding to changes in a user's emotions. A specific embodiment of this system will be described below.

[0552] A system to detect approaches from fraudsters

[0553] The server analyzes the content and behavioral patterns of messages from fraudsters. Specifically, the server uses an AI model to analyze the wording, topic, and sending time of messages from fraudsters. This analysis uses natural language processing (NLP) technology. For example, Python libraries such as NLTK and spaCy can be used. This allows the server to understand the behavioral patterns of fraudsters and detect messages that are likely to be fraudulent.

[0554] Examples:

[0555] An example of a scam message: "Your account has been compromised. Click this link to check now."

[0556] An example prompt you might use is: "Please rate the likelihood that this message is a scam."

[0557] Tracking user emotional changes with an emotion engine

[0558] The device tracks changes in the user's emotions over time. The emotion engine uses a sentiment analysis model to analyze emotions from the user's input and behavior. For example, it can use Microsoft® Azure®'s Text Analytics API or Google® Cloud Natural Language API. This allows the device to recognize changes in the user's emotions in real time and generate appropriate responses.

[0559] Examples:

[0560] An example of a change in user emotions: At first, the user expresses joy, saying, "This service is great!", but later expresses dissatisfaction, saying, "The recent update has made it difficult to use."

[0561] An example prompt you might use is: "Generate an appropriate response if the user's emotion changes from happy to frustrated."

[0562] System configuration

[0563] This system is implemented using the following hardware and software.

[0564] Server: A server with a powerful processor and large memory capacity is used. The server analyzes fraudulent messages and detects fraudulent patterns.

[0565] Device: A device is used to collect user input and behavioral data and track emotional changes. The device has sufficient processing power to run the emotion engine.

[0566] Software: Use software libraries and APIs such as Python, NLTK, spaCy, Microsoft Azure Text Analytics API, Google Cloud Natural Language API, etc.

[0567] Example

[0568] A specific example of this system will be described below.

[0569] 1. Analysis of the fraudulent message:

[0570] The server retrieves messages received by the user.

[0571] The server uses NLTK to tokenize messages and spaCy to classify topics.

[0572] The server uses a generative AI model to assess the likelihood that a message is fraudulent.

[0573] 2. Tracking user emotional changes:

[0574] The device collects user input and behavioral data in real time.

[0575] The device uses the Microsoft Azure Text Analytics API to analyze user sentiment and the Google Cloud Natural Language API to classify sentiment.

[0576] The device uses a generative AI model to generate appropriate responses that correspond to the user's emotions.

[0577] In this way, the system can detect approaches from fraudsters with high accuracy and respond appropriately based on the user's emotions.

[0578] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0579] Processing steps of a system to detect approaches from fraudsters

[0580] Step 1: Receiving a message

[0581] The server retrieves messages received by users, which can be collected from different platforms such as email, SMS, and chat apps.

[0582] Input: Message received by the user

[0583] Output: Message data stored on the server

[0584] Specific behavior:

[0585] The server retrieves new mail from the email server using the IMAP protocol.

[0586] Use the chat app's API to get new messages.

[0587] Step 2: Parse the message

[0588] The server analyzes the received messages using natural language processing (NLP) techniques, specifically extracting information such as the message's wording, topic, and time of sending.

[0589] Input: Message data stored on the server

[0590] Output: Parsed message feature data

[0591] Specific behavior:

[0592] The server uses the NLTK library in Python to tokenize the messages.

[0593] Use spaCy to classify the topic of messages.

[0594] Step 3: Detect fraud patterns

[0595] The server uses the analysis results to detect fraud patterns, and the AI ​​model is trained using a dataset of past fraudulent messages.

[0596] Input: Parsed message feature data

[0597] Output: Message data assessed for fraudulent potential

[0598] Specific behavior:

[0599] The server uses a trained generative AI model (e.g., BERT or GPT-3®) to assess the likelihood that the message is fraudulent.

[0600] If there is a high possibility of fraud, a warning will be sent to the user.

[0601] Processing step of tracking user emotion changes by emotion engine

[0602] Step 1: Collecting User Input

[0603] The device collects user input and behavior data, including interactions such as text entry, clicking, and scrolling.

[0604] Input: User input and behavioral data

[0605] Output: User behavior data stored on the device

[0606] Specific behavior:

[0607] The terminal captures text entered by the user in real time.

[0608] Log user click and scroll data.

[0609] Step 2: Sentiment Analysis

[0610] The device analyzes the user's emotions based on the collected data, and the emotion engine uses an emotion analysis model to identify the user's emotions.

[0611] Input: User behavior data stored on the device

[0612] Output: Analyzed user emotion data

[0613] Specific behavior:

[0614] The device uses Microsoft Azure's Text Analytics API to extract sentiment from the user's text input.

[0615] Classify user sentiment using the Google Cloud Natural Language API.

[0616] Step 3: Track your emotions

[0617] The device tracks changes in the user's emotions over time, allowing it to understand how the user's emotions are changing.

[0618] Input: Parsed user emotion data

[0619] Output: User emotion data stored in time series

[0620] Specific behavior:

[0621] The terminal stores the user's emotion data in a time-series database.

[0622] It monitors emotional changes in real time and generates alerts if abnormal changes are detected.

[0623] Step 4: Generate an appropriate response

[0624] The device generates appropriate responses based on the user's emotional changes, using a generative AI model to create messages that match the user's emotions.

[0625] Input: User emotion data stored in time series

[0626] Output: The generated response message

[0627] Specific behavior:

[0628] The device uses a trained generative AI model (e.g., GPT-3) to generate responses that correspond to the user's emotions.

[0629] The generated response is displayed to the user.

[0630] (Application example 2)

[0631] Next, a description will be given of Application Example 2 of Form 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."

[0632] In recent years, the sophisticated methods used by fraudsters have become more prevalent, increasing the risk of users receiving fraudulent messages. Furthermore, there have been many cases where users have been unable to respond appropriately to fraudulent messages and have become victims. Furthermore, there is a lack of systems that provide appropriate advice and warnings in response to changes in users' emotions, making it difficult to ensure user safety.

[0633] 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.

[0634] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing messages received by users in real time, means for detecting potentially fraudulent messages and issuing warnings, means for tracking changes in the user's emotions, and means for providing appropriate advice and warnings. This allows users to respond quickly and appropriately when they receive fraudulent messages, making it possible to prevent fraud damage before it occurs.

[0635] "Artificial intelligence" is the technology that enables computer systems to learn, reason, and self-correct by imitating human intelligence.

[0636] "Natural communication" refers to interfaces and means that allow humans and machines to converse seamlessly.

[0637] "Fictitious contacts" refer to digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[0638] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0639] "Means of approach detection" refers to technologies and methods that analyze the content of messages and behavioral patterns from fraudsters to identify potential fraud.

[0640] "Fraudster tactics" refers to techniques and methods used to intentionally target fraudsters and gather further evidence.

[0641] "Means of collecting evidence of criminal activity" refers to technologies and methods that record the actions and messages of fraudsters and store them in a form that can later be used in legal proceedings.

[0642] "Means of reporting to law enforcement" refers to the techniques and methods for providing collected evidence to police or law enforcement agencies in an appropriate format.

[0643] "Real-time analysis means" refers to technologies and methods that instantly analyze messages received by users and assess their likelihood of fraud.

[0644] "Warning measures" refers to techniques or methods for alerting users when a potentially fraudulent message is detected.

[0645] "Means for tracking emotional changes" refers to techniques or methods for monitoring a user's emotional state over time and recording those changes.

[0646] "Means for providing appropriate advice and warnings" refers to technologies and methods for providing optimal countermeasures or warnings based on the user's emotional state and the content of the message.

[0647] As an embodiment of the present invention, the fraud detection system is implemented as an application installed on a smartphone, as will be described in detail below.

[0648] System Program

[0649] The system includes a program with the following main functions:

[0650] 1. Natural communication using artificial intelligence:

[0651] The server uses a generative AI model to interact naturally with the user, generating appropriate responses to user input.

[0652] 2. Generate fictitious contacts:

[0653] The server generates fictitious email addresses and social media accounts to attract fraudsters, and these contacts are used to monitor the fraudsters' activities.

[0654] 3. Detecting fraudster approaches:

[0655] The server analyzes messages received by users in real time to detect potentially fraudulent messages, including analyzing message content, time of sending, and behavioral patterns.

[0656] 4. Letting fraudsters run wild:

[0657] The server will intentionally react to fraudsters and collect further evidence, a process that will provide a detailed record of fraudsters' actions.

[0658] 5. Collecting evidence of criminal activity:

[0659] The server records the actions and messages of fraudsters and stores them in a form that can later be used in legal proceedings.

[0660] 6. Police Report:

[0661] The server provides the collected evidence to police and law enforcement agencies in an appropriate format.

[0662] 7. Tracking user emotional changes:

[0663] The server monitors the user's emotional state over time and records changes, allowing it to provide appropriate advice and warnings in response to changes in the user's emotions.

[0664] Hardware and software used

[0665] Hardware: Smartphone

[0666] Software: Python, TextBlob library, generative AI model

[0667] Data processing and calculation

[0668] The server performs the following data processing and calculations to analyze messages received by users in real time.

[0669] 1. Message content analysis:

[0670] Analyzes the text of messages to detect potentially fraudulent keywords and patterns.

[0671] 2. Sentiment analysis:

[0672] Use the TextBlob library to analyze the sentiment of the message and assess the emotional state of the user.

[0673] 3. Behavioral Pattern Analysis:

[0674] Analyze the time and frequency of messages sent to identify fraudsters' behavioral patterns.

[0675] Specific examples

[0676] Received message: "Congratulations! You've won a prize."

[0677] The app responds: "Warning: This message may be a scam."

[0678] Prompt Sentence Examples

[0679] Develop an application that analyzes messages received by users in real time, detects potentially fraudulent messages, and issues a warning. Also, analyze the sentiment of the message and add a warning feature if the message is likely to evoke negative sentiment.

[0680] The above is an embodiment of the present invention.

[0681] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0682] Step 1:

[0683] The server retrieves messages received by the user. The input is the message received by the user, and the output is the message text to be analyzed. Specifically, the server retrieves messages from the user's mailbox or SNS account.

[0684] Step 2:

[0685] The server analyzes the retrieved message text to detect keywords and patterns that may indicate fraud. The input is the message text, and the output is a flag indicating whether it is likely to be fraudulent. Specifically, the server uses regular expressions to search for fraudulent keywords in the message.

[0686] Step 3:

[0687] The server analyzes the sentiment of a message using the TextBlob library. The input is the message text and the output is a sentiment score. Specifically, the server passes the message text to TextBlob and obtains a positive or negative sentiment score.

[0688] Step 4:

[0689] The server analyzes the time and frequency of message sending to identify fraudsters' behavioral patterns. The input is message metadata (sent time, sender information, etc.), and the output is the evaluation result of the behavioral pattern. Specifically, the server analyzes message timestamps to detect anomalous sending patterns.

[0690] Step 5:

[0691] If the server detects a potentially fraudulent message, it issues a warning to the user. The input is a flag indicating whether the message is potentially fraudulent and an emotion score, and the output is a warning message. Specifically, the server sends a notification to the user's smartphone to warn of the possibility of fraud.

[0692] Step 6:

[0693] The server monitors the user's emotional state over time and records the changes. The input is the history of emotional scores, and the output is the pattern of emotional changes. Specifically, the server periodically records the emotional scores and graphs the changes.

[0694] Step 7:

[0695] The server then provides optimal countermeasures and warnings based on the user's emotional state and the content of the message. The input is the pattern of emotional changes and the message content, and the output is advice or a warning message. Specifically, the server uses a generative AI model to generate an appropriate response for the user.

[0696] The above are the specific processing steps of the fraud detection system.

[0697] Example 3

[0698] Next, a description will be given of a third embodiment of the third embodiment. 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."

[0699] Fraud crimes are becoming more sophisticated every year, making it difficult to identify fraudsters and collect evidence using conventional methods. Furthermore, there is a lack of systems to effectively analyze information obtained through communication with fraudsters and report it to the police. Furthermore, there is a need to recognize users' emotions and respond appropriately.

[0700] The identification process by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes a means for natural communication using artificial intelligence, a means for generating fictitious contact information, a means for detecting approaches from fraudsters, a means for letting fraudsters go unpunished, a means for collecting evidence of criminal activity, a means for reporting the collected evidence to the police, a means for receiving messages from fraudsters, a means for storing the content of the received messages, the sender's IP address, and the sending time in a database, a means for analyzing the messages using a natural language processing engine, a means for recognizing the user's emotions using an emotion engine, and a means for formatting the collected data and sending a report to the police. This makes it possible to identify fraudsters, effectively collect evidence, and report to the police. It also recognizes the user's emotions and takes appropriate action.

[0701] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0702] "Natural communication" refers to technology that allows humans and computers to exchange information smoothly and without any sense of awkwardness when interacting with each other.

[0703] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[0704] A "fraudster" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0705] "Approach detection methods" are technologies that analyze the content of messages and behavioral patterns from fraudsters to identify fraud attempts.

[0706] "Methods of letting fraudsters go free" refers to a technique of not taking immediate action against fraudsters, but observing them for a certain period of time to gather more information.

[0707] "Evidence of criminal activity" is information or data that proves the fraudulent activities of a fraudster.

[0708] "Means of reporting to the police" are the methods and techniques used to provide the collected evidence to the police.

[0709] "Means for receiving messages" refers to the technology that allows the server to receive messages sent by fraudsters.

[0710] A "database" is a system that systematically stores information and allows it to be searched and retrieved as needed.

[0711] A "natural language processing engine" is a technology for understanding and analyzing human language.

[0712] An "emotion engine" is a technology for recognizing and analyzing a user's emotions.

[0713] The "means for transmitting the report" is the technology used to format and transmit the collected data electronically to the police.

[0714] This invention is a system for collecting information obtained through communication with fraudsters and reporting it to the police. A specific embodiment of this system will be described below.

[0715] The server uses email and chat applications to receive messages from fraudsters. The SMTP protocol is used to receive emails, and WebSocket is used to receive messages from chat applications. Received messages are temporarily stored in memory.

[0716] The server then stores information such as the content of the received message, the IP address of the sender, and the time of sending in a database. This database uses a relational database management system (RDBMS) such as MySQL. The stored data is stored in an appropriate format for later analysis.

[0717] The server analyzes the received messages using a natural language processing (NLP) engine, which uses generative AI models such as Google's BERT and OpenAI's GPT-3. The results are used to identify fraudulent patterns and messages containing specific keywords, such as "bank account," "fraudulent use," and "click on link."

[0718] Furthermore, the server uses an emotion engine to recognize the user's emotions. This emotion engine extracts emotions from the user's messages and improves the accuracy of emotion recognition through learning functions. For example, if a particular user uses specific words and expressions to express joy, the emotion engine can learn this and more accurately recognize the user's joy.

[0719] Finally, the server reports the collected evidence to the police via email or a dedicated reporting system. The server formats the collected data and creates a report, which is sent to the police via email or uploaded to a dedicated reporting system.

[0720] As a concrete example, consider a case where a fraudster sends a message saying, "Your bank account has been fraudulently accessed. Please click this link now to check." The server receives this message and stores the sender's IP address and the time of sending in a database. The server then uses an NLP engine to analyze the message and identify keywords such as "bank account," "fraudulent access," and "click link." The emotion engine recognizes the emotion the user felt when receiving this message and determines that they are likely feeling fear or anxiety. The collected information is reported to the police and used to identify fraudsters and prevent crime.

[0721] Example prompt sentence:

[0722] "Analyze messages from fraudsters to identify the IP addresses from which they were sent and the time of sending. Also, recognize user sentiment and gather information to report to law enforcement."

[0723] In this way, the server effectively collects information obtained through communication with fraudsters and reports it to the police. The flow of the identification process in the third embodiment will be described with reference to FIG.

[0724] Step 1: Receiving a message

[0725] The server receives messages from fraudsters. The input is email or a message from a chat application. The server receives email using the SMTP protocol and messages from chat applications using WebSocket. The received messages are temporarily stored in memory. The output is the content of the received messages.

[0726] Step 2: Save your data

[0727] The server stores information such as the content of the received message, the sender's IP address, and the time of sending in a database. The input is the content of the message received in step 1, the sender's IP address, and the time of sending. The server uses a relational database management system (RDBMS) such as MySQL to store this information in the database in an appropriate format. The output is the information stored in the database.

[0728] Step 3: Analyze the data

[0729] The server uses a natural language processing (NLP) engine to analyze the collected data. The input is the content of messages stored in a database. The server uses generative AI models such as Google's BERT or OpenAI's GPT-3 to analyze the content of the messages. The analysis results in identifying fraud patterns and specific keywords. The output is the analysis results.

[0730] Step 4: Emotion Recognition

[0731] The server uses an emotion engine to recognize the user's emotions. The input is the user's message. The server inputs the user's message into the emotion engine and extracts the emotion. The emotion engine improves the accuracy of emotion recognition through a learning function. The output is the user's emotion recognition result.

[0732] Step 5: File a police report

[0733] The server reports the collected evidence to the police. The inputs are the analysis results and emotion recognition results. The server formats the collected data and creates a report. The report is sent to the police via email or uploaded to a dedicated reporting system. The output is the report sent to the police.

[0734] As a concrete example, consider the situation where you receive a message from a fraudster saying, "Your bank account has been compromised. Click this link now to check."

[0735] Step 1:

[0736] The server receives a message from the fraudster using the SMTP protocol, which reads: "Your bank account has been compromised. Click this link now to check."

[0737] Step 2:

[0738] The server stores the received message content ("Your bank account has been fraudulently used. Click this link now to check"), the sender's IP address (192.168.1.1), and the sending time (2023-10-01 12:00:00) in a database.

[0739] Step 3:

[0740] The server inputs the message content into an NLP engine and identifies keywords such as "bank account," "fraudulent use," and "click on link." The analysis results indicate that the message is "highly likely to be fraudulent."

[0741] Step 4:

[0742] The server inputs the user's message into the emotion engine and determines that the user is likely feeling fear or anxiety. The emotion recognition result is "fear" or "anxiety."

[0743] Step 5:

[0744] The server formats the collected data and creates a report, which is then sent to the police via email. The report contains information about the likely fraudulent message, the sender's IP address, the time of sending, and the user's emotion recognition results.

[0745] (Application example 3)

[0746] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[0747] Fraud crimes are becoming more sophisticated every year, and there are an increasing number of cases where victims fall victim to fraud without realizing the scam methods. Furthermore, the difficulty of quickly and accurately collecting evidence of fraud and reporting it to the police often delays the identification and arrest of fraudsters. Furthermore, the lack of means to properly recognize the victim's emotions and assess the possibility of fraud makes it difficult to prevent damage before it occurs. To solve these issues, a system is needed that can analyze suspicious messages and calls in real time and issue a warning to users.

[0748] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0749] In this invention, the server includes means for using artificial intelligence to conduct natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing the user's emotions and assessing the possibility of fraud, and means for displaying a warning to the user if fraud is suspected. This makes it possible to analyze messages or calls suspected of fraud in real time and issue a warning to the user. Furthermore, by quickly and accurately collecting evidence of fraud and reporting it to the police, fraudsters can be quickly identified and arrested. Furthermore, by appropriately recognizing the user's emotions and assessing the possibility of fraud, damage can be prevented before it occurs.

[0750] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0751] "Natural communication" refers to interfaces and methods designed to allow humans and machines to interact seamlessly.

[0752] "Fictitious contacts" are digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[0753] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to determine whether or not there is suspicion of fraud.

[0754] "Methods of letting fraudsters get away" are techniques for extracting information and collecting evidence from fraudsters without raising suspicion.

[0755] "Means for collecting evidence of criminal activity" refers to technology that records information obtained from fraudsters, such as messages, the IP address of the sender, and the time of sending.

[0756] "Means of reporting collected evidence to the police" refers to techniques for providing collected evidence of fraud to the police in an appropriate format.

[0757] The "means for recognizing a user's emotions and assessing the possibility of fraud" is a technology that analyzes a user's emotions and assesses the possibility of fraud based on those emotions.

[0758] The "means for displaying a warning to the user when fraud is suspected" is a technique for displaying a warning to the user when it is determined that fraud is suspected.

[0759] A system for implementing this invention has the following configuration: a server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact points, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing a user's emotions and evaluating the possibility of fraud, and means for displaying a warning to the user if fraud is suspected.

[0760] Hardware and Software Configuration

[0761] 1. Hardware:

[0762] Smartphone (iOS or ANDROID (registered trademark))

[0763] Smart glasses (e.g., Google Glass (registered trademark))

[0764] 2. Software:

[0765] AI models (e.g. GPT-4)

[0766] Emotion engine (e.g. Affectiva SDK)

[0767] Message analysis engine (e.g. IBM Watson(R))

[0768] Data processing and calculation

[0769] The server analyzes messages and call content received by the user in real time. It uses an AI model (GPT-4) to analyze the message content and determine whether it is suspected of fraud. If fraud is suspected, it collects the sender's IP address, the time of sending, and the message content. It uses an emotion engine (Affectiva SDK) to analyze the user's emotions and further evaluate the possibility of fraud. If fraud is suspected, it displays a warning to the user. If necessary, it reports the collected evidence to the police.

[0770] Specific examples

[0771] When a user receives a message on their smartphone, the fraud prevention AI assistant analyzes the message content and, if it detects a suspected fraud, displays a warning to the user, while also collecting evidence of the fraud and reporting it to the police.

[0772] Prompt Sentence Examples

[0773] Analyze messages received by users to determine whether they are suspected to be fraudulent. If so, collect the sending IP address, time of sending, and message content to generate data for reporting to law enforcement. Also, analyze user sentiment to assess the likelihood of fraud.

[0774] In this way, the fraud prevention AI assistant protects users from fraud and provides a system that helps prevent crime.

[0775] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0776] Step 1:

[0777] The user receives the message.

[0778] Input: Message content, sender information

[0779] Output: Message content, sender information

[0780] What it does: The user's smartphone or smart glasses receives a new message and sends the message's content and sender information to the server.

[0781] Step 2:

[0782] The server analyzes the message content.

[0783] Input: Message content

[0784] Output: Suspected fraud?

[0785] How it works: The server uses an AI model (GPT-4) to analyze the message content and determine whether it is suspected of fraud. If it is, it passes the information to the next step.

[0786] Step 3:

[0787] If the server suspects fraud, it will collect evidence.

[0788] Input: Message content, sender information, sending time

[0789] Output: Collected evidence data

[0790] Specific operation: If the server suspects fraud, it records the sender's IP address, the time of sending, and the message content, and saves them as evidence.

[0791] Step 4:

[0792] The server analyzes the user's emotions.

[0793] Input: User reaction data (voice, facial expressions, etc.)

[0794] Output: User's sentiment rating

[0795] How it works: The server uses an emotion engine (Affectiva SDK) to analyze the user's reaction data and evaluate the user's emotions, which can then be used to further assess the likelihood of fraud.

[0796] Step 5:

[0797] The server displays a warning to the user if fraud is suspected.

[0798] Input: Suspected fraud, user sentiment rating

[0799] Output: Warning message

[0800] Specific behavior: If the server determines that fraud is suspected and the user's emotion rating indicates the possibility of fraud, a warning message will be displayed on the user's smartphone or smart glasses.

[0801] Step 6:

[0802] Report any evidence collected by the server to the police.

[0803] Input: Collected evidence data

[0804] Output: Report data sent to police

[0805] Specific operations: The server converts the collected evidence data into an appropriate format, generates data for reporting to the police, and sends it.

[0806] In this way, the fraud prevention AI assistant protects users from fraud and provides a system that helps prevent crime.

[0807] 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.

[0808] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ) 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.

[0809] Another example of generative AI is Gemini (registered trademark) (Internet search engine). <url: https: gemini.google.com ?hl="ja">) are listed.

[0810] 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.

[0811] [Second embodiment]

[0812] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0813] 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.

[0814] 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).

[0815] 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.

[0816] 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.

[0817] 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).

[0818] 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.

[0819] 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.

[0820] 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.

[0821] 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.

[0822] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0823] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0824] "Example 1"

[0825] One embodiment of the present invention is a system in which artificial intelligence (AI) can communicate naturally. This AI can use natural language processing (NLP) technology to have conversations similar to those of humans. The AI ​​also has the ability to generate fictitious contact information. This allows the AI ​​to detect attempts by fraudsters to contact the AI ​​and allow them to escape.

[0826] "Example 2"

[0827] Another embodiment of the present invention is a system for detecting approaches from fraudsters. This system detects approaches from fraudsters by analyzing the content of messages and behavioral patterns from fraudsters. Specifically, AI analyzes the wording, topic, and sending time of messages from fraudsters, and from this information, it understands the behavioral patterns of fraudsters.

[0828] "Example 3"

[0829] Furthermore, one embodiment of the present invention is a system that collects evidence of criminal activity and reports it to the police. AI collects information obtained through communication with fraudsters as evidence and reports it to the police as necessary. Specifically, AI collects information such as messages from fraudsters, the sender's IP address, and the time of transmission, and provides this information to the police. This can be useful in identifying fraudsters and preventing crimes.

[0830] The processing flow of each embodiment will be described below.

[0831] "Example 1"

[0832] Step 1: AI uses natural language processing (NLP) techniques to engage in human-like dialogue.

[0833] Step 2: The AI ​​generates fictitious contacts, which are digital accounts such as email addresses and social media accounts.

[0834] Step 3: If a fraudster attempts to contact the AI, the AI ​​will detect the attempt.

[0835] Step 4: The AI ​​lets the fraudsters go free while collecting evidence of their criminal activity.

[0836] "Example 2"

[0837] Step 1: AI analyzes the content of messages and behavioral patterns from fraudsters.

[0838] Step 2: The AI ​​analyzes the scammers' messages for wording, topic, and time of sending.

[0839] Step 3: Use this information to understand the behavioral patterns of fraudsters and detect their approaches.

[0840] "Example 3"

[0841] Step 1: The AI ​​collects evidence from communications with fraudsters.

[0842] Step 2: The AI ​​collects information from the fraudsters, such as the message, the IP address from which it was sent, and the time of sending.

[0843] Step 3: The AI ​​will provide the collected information to law enforcement, which can help identify fraudsters and prevent crime.

[0844] Example 1

[0845] Next, a description will be given of Example 1 of Form 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."

[0846] Conventional AI systems have had difficulty effectively detecting approaches from fraudsters and responding appropriately. They also lacked the means to generate fictitious contact information to lure fraudsters and collect evidence of criminal activity. Furthermore, they needed to be able to detect signs of fraud and generate appropriate responses while engaging in natural communication.

[0847] 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.

[0848] In this invention, the server includes means for conducting natural communication using artificial intelligence, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing input data using natural language processing technology, means for generating responses using a generative AI model, means for executing an algorithm for detecting signs of fraud, and means for sending the generated responses or fictitious contact information to the terminal. This makes it possible to effectively detect approaches from fraudsters and deal with them appropriately.

[0849] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0850] "Natural communication" is communication that mimics the dialogue and conversation that humans have on a daily basis and that occurs without any sense of incongruity.

[0851] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[0852] A "fraudster" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0853] "Approach detection methods" are techniques and methods used to identify and detect contact or attempts by fraudsters.

[0854] "Means to keep fraudsters going" are methods used to encourage fraudsters to continue committing crimes and to monitor their behavior.

[0855] "Means for collecting evidence of criminal activity" refers to techniques or methods that record the actions or communications of fraudsters so that they can later be used as evidence.

[0856] A "police reporting method" is a method of providing collected evidence to law enforcement agencies to assist in the investigation or prosecution of a crime.

[0857] "Natural language processing technology" is a technology that allows computers to understand and analyze human language.

[0858] A "generative AI model" is an artificial intelligence model that generates new text or responses based on input data.

[0859] A "fraud indicator detection algorithm" is a computational procedure designed to identify patterns or characteristics that indicate possible fraudulent activity.

[0860] A "terminal" is a device used by a user, such as a computer or smartphone.

[0861] MODE FOR CARRYING OUT THE INVENTION

[0862] This invention is a system that uses artificial intelligence to communicate naturally, detect approaches from fraudsters, and respond appropriately. This system works in cooperation with three parties: a server, a terminal, and a user.

[0863] Hardware and software used

[0864] Hardware: High-performance servers (e.g., virtual servers from cloud service providers)

[0865] Software: Natural language processing engines (e.g., natural language processing APIs), generative AI models (e.g., generative AI models), fraud detection algorithms

[0866] System configuration

[0867] 1. Server:

[0868] The server executes a program for natural communication using artificial intelligence.

[0869] The server analyzes the input data from the user using natural language processing techniques.

[0870] The server uses a generative AI model to generate an appropriate response.

[0871] The server runs algorithms that detect signs of fraud and detect approaches from fraudsters.

[0872] The server generates fictitious contact details to allow fraudsters to operate.

[0873] The server collects evidence of criminal activity and reports the collected evidence to the police.

[0874] 2. Terminal:

[0875] The terminal provides an interface for the user to enter input.

[0876] The terminal transmits the user's input data to the server.

[0877] The terminal receives the response or the fictitious contact from the server and displays it to the user.

[0878] 3. User:

[0879] A user uses a terminal to input questions or requests to the system.

[0880] The user checks the response from the server displayed on the terminal.

[0881] Specific examples

[0882] For example, if a user types "Hello, how's the weather today?", the device sends this input to the server. The server uses a natural language processing engine to parse the input data and uses a generative AI model to generate the response "Hello, the weather is sunny today." After verifying that there are no signs of fraud, the server sends this response to the device, which displays it to the user.

[0883] On the other hand, if a fraudster types "I want to add a new contact, how do I do that?", the server will detect the signs of fraud and generate a fictitious contact and send it to the device, which will then display this fictitious contact to the fraudster.

[0884] Prompt Sentence Examples

[0885] "Hello, how's the weather today?"

[0886] "I want to add a new contact, how do I do that?"

[0887] "I've been getting a lot of scam calls lately, what should I do?"

[0888] In this way, the present invention makes it possible to effectively detect approaches from fraudsters and deal with them appropriately.

[0889] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0890] Step 1:

[0891] The user enters input from the terminal.

[0892] The user uses the terminal's interface to input a question or request, for example, "Hello, what's the weather like today?" The input data is sent to the terminal in text form.

[0893] Step 2:

[0894] The terminal sends the input data to the server.

[0895] The terminal sends the text data entered by the user to the server via the Internet. At this time, the data is encrypted before being sent. The input data reaches the server.

[0896] Step 3:

[0897] The server receives the input data and analyzes it using a natural language processing (NLP) engine.

[0898] The server receives input data sent from the device. The received data is analyzed using a natural language processing engine. Specifically, the intent and sentiment of the input data are extracted. For example, the input "Hello, how's the weather today?" is analyzed and determined to be a question about the weather.

[0899] Step 4:

[0900] The server generates a response using a generative AI model based on the analysis results.

[0901] The server uses the generative AI model to generate an appropriate response based on the analysis results of the NLP engine. For example, if the analysis result is a question about the weather, the generative AI model will generate the response "Hello, the weather is sunny today." The generated response is stored in text format on the server.

[0902] Step 5:

[0903] The server runs fraud detection algorithms to detect signs of fraud.

[0904] Before sending the generated response to the user, the server runs a fraud detection algorithm. This algorithm detects whether the input data contains any signs of fraud. For example, if the input "I want to add a new contact, how do I do this?" indicates signs of fraud, the server will detect this. The detection results are stored on the server.

[0905] Step 6:

[0906] Send a server-generated response or fictitious contact to the device.

[0907] If fraud signs are detected, the server generates a fictitious contact and sends it to the device. If no fraud signs are detected, the server sends the generated response as is. The transmitted data is encrypted before reaching the device.

[0908] Step 7:

[0909] The terminal displays the response from the server to the user.

[0910] The terminal displays the response or fictitious contact information received from the server to the user. The user checks the displayed information and decides the next action. For example, the response "Hello, the weather is sunny today" is displayed.

[0911] (Application example 1)

[0912] Next, a description will be given of Application Example 1 of Form 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."

[0913] In recent years, fraud crimes have been increasing, especially frauds using digital communication. It is difficult to respond to such fraud crimes quickly and effectively using conventional methods. In addition, there are limited means to identify fraudsters and collect evidence. This has led to an increase in the number of victims, which has become a social problem.

[0914] 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.

[0915] In this invention, the server includes means for using artificial intelligence to perform natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for performing sentiment analysis, and means for notifying users using the fictitious contact information. This makes it possible to quickly detect fraudsters, collect evidence, and take appropriate action.

[0916] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0917] "Natural communication" refers to the natural verbal exchanges that humans engage in on a daily basis, and in particular to dialogues that are realized using natural language processing technology.

[0918] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[0919] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0920] "Approach detection methods" refers to technologies and methods used to detect contact or messages from fraudsters.

[0921] "Making fraudsters vulnerable" refers to deliberately maintaining contact with fraudsters in order to monitor their behavior and gather evidence.

[0922] "Evidence of criminal activity" refers to data or information that proves the fraudulent activities of fraudsters.

[0923] "Sentiment analysis" is a technique for analyzing emotions and the intensity of emotions from text data.

[0924] "Means of notifying users" refers to methods and technologies used to communicate information or warnings detected by the system to users.

[0925] As an embodiment of this invention, we will explain how to install a fraud detection AI assistant system on a smartphone. This system uses artificial intelligence to communicate naturally, detects approaches from fraudsters, and generates fictitious contact information to allow fraudsters to escape.

[0926] System configuration

[0927] The system consists of the following main components:

[0928] 1. Artificial intelligence module: Uses natural language processing technology to communicate naturally with users.

[0929] 2. Fictitious Contact Generation Module: Generates fictitious contacts to deceive fraudsters.

[0930] 3. Fraud detection module: Analyzes message content and behavioral patterns to detect approaches from fraudsters.

[0931] 4. Sentiment Analysis Module: Analyzes the sentiment of the message and assesses its likelihood of fraud.

[0932] 5. Notification module: Notifies users of potential fraud and prompts them to take appropriate action.

[0933] Hardware and software used

[0934] Hardware: Smartphone

[0935] Software: Python, NLTK (Natural Language Toolkit)

[0936] Data processing and calculation

[0937] 1. Receiving a message: The system retrieves the message received by the user.

[0938] 2. Sentiment Analysis: Calculate the sentiment score of the message using NLTK's SentimentIntensityAnalyzer.

[0939] 3. Fraud detection: If the sentiment score exceeds a certain threshold, it is determined to be a possible fraud.

[0940] 4. Fictitious Contact Generation: Randomly generate fictitious contacts if fraud is suspected.

[0941] 5. User Notification: Notify users of potential scams and fictitious contacts.

[0942] Specific examples

[0943] For example, if a user receives a message saying "You've won the lottery! Please send us your bank details," the system can analyze the message and determine that it is likely a scam. It can then generate a fake contact and notify the user, saying "Suspicious activity has been detected. Please contact fake_contact_1@example.com for more information."

[0944] Prompt Sentence Examples

[0945] Examples of prompts to input to a generative AI model include:

[0946] "Analyze the following message for potential fraud: 'You have won a lottery! Please send your bank details.'"

[0947] Using this prompt, a generative AI model is asked to analyze the message and determine whether it is likely to be fraudulent.

[0948] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0949] Step 1:

[0950] The user receives a message. The terminal acquires the message received by the user. This message becomes the input data for the system.

[0951] Step 2:

[0952] The device sends the received message to the sentiment analysis module, which calculates the sentiment score of the message using NLTK's SentimentIntensityAnalyzer. This process outputs the message's sentiment score as positive, negative, or neutral.

[0953] Step 3:

[0954] The server analyzes the emotion scores and determines that fraud is likely if the negative emotion score exceeds a certain threshold. This determination result becomes the input data for the next step.

[0955] Step 4:

[0956] If the server determines that fraud is likely, it invokes a fictitious contact generation module, which randomly generates a fictitious contact and outputs the contact information.

[0957] Step 5:

[0958] The server notifies the user using the generated fictitious contact information. The notification module sends a message to the user informing them of the potential fraud and including the fictitious contact information. The notification is displayed on the user's device.

[0959] Step 6:

[0960] The user is notified and made aware of the potential scam, and can either continue to contact the scammer using the fictitious contact details provided, or take appropriate action.

[0961] Step 7:

[0962] The server monitors interactions between users and fraudsters and collects evidence of criminal activity, which is then stored for later reporting to law enforcement.

[0963] Step 8:

[0964] The server reports the collected evidence to the police. The reporting module organizes the evidence data and sends it to the police in the appropriate format. This process helps identify and apprehend fraudsters.

[0965] Example 2

[0966] Next, a description will be given of Example 2 of Form 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."

[0967] There is a need to quickly and accurately detect approaches from fraudsters and protect users. However, conventional systems have low accuracy in detecting fraudulent messages, putting users at high risk of becoming victims of fraud. In addition, analyzing and evaluating fraudulent messages takes time, making it difficult to respond in real time.

[0968] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving messages sent by fraudsters and storing them in a database, means for analyzing the stored messages using a natural language processing tool, means for extracting message characteristics from the analysis results, means for inputting the extracted characteristics into a generative AI model to evaluate the likelihood of fraud, and means for notifying users of messages that are likely to be fraudulent. This makes it possible to quickly and accurately detect approaches from fraudsters and protect users in real time.

[0969] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[0970] "Natural communication" refers to exchanging information using natural language and non-verbal means, as humans do in their daily lives.

[0971] "Fictitious contacts" are non-existent contact methods created to attract fraudsters.

[0972] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[0973] "Approach detection methods" are technologies and methods used to identify and detect contact or messages from fraudsters.

[0974] "Methods to let fraudsters go free" refers to the practice of monitoring fraudsters' activities and intentionally allowing them to continue their actions in order to gather evidence.

[0975] "Means of collecting evidence of criminal activity" are methods of recording the actions and messages of fraudsters to gather evidence for later use in legal proceedings.

[0976] A "police reporting method" is a method for providing collected evidence to law enforcement and reporting criminal activity.

[0977] A "database" is a system for storing and managing data in an organized manner.

[0978] "Natural language processing tools" are technologies and software that enable computers to understand and analyze human language.

[0979] A "generative AI model" is a model of artificial intelligence trained to perform a specific task, especially a generative task.

[0980] A "feature extraction method" is a method for identifying and extracting important patterns or attributes from data.

[0981] A "means for assessing the likelihood of fraud" is a technique or method for assessing and scoring the risk of fraud based on the extracted features.

[0982] A "means for notifying a user" is a method for informing a user of a detected risk of fraud.

[0983] This invention is a system for detecting approaches from fraudsters and protecting users. This system operates in cooperation with a server, terminals, and users.

[0984] The server receives messages sent by fraudsters and stores them in a database. Database management systems such as MySQL or PostgreSQL are used for the database. The stored messages are then analyzed using natural language processing tools. Specifically, natural language processing libraries such as "spaCy" and "NLTK" are used to tokenize the messages and tag them by part of speech.

[0985] From the analysis results, the server extracts message features. These features include specific keywords in the message (e.g., "bank," "account," "link," etc.) and the fact that the message was sent late at night. These features are input into a generative AI model. For example, OpenAI's GPT-4 is used as the generative AI model.

[0986] The generative AI model evaluates the likelihood of fraud based on the input features and assigns a score. For example, the following prompt sentence is input into the generative AI model:

[0987] "Please rate the following message for possible fraud. It reads: 'Hello, there's a problem with your bank account. Click this link to check it now.'"

[0988] The generative AI model analyzes the message based on this prompt and scores the likelihood of it being fraudulent. If the score is high, the server sends a notification to the user. Notifications can be sent via email, in-app notifications, or other methods. For example, a push notification could be sent to the user's smartphone, warning them, "You have received a message that is likely to be fraudulent. Please be careful."

[0989] This system makes it possible to quickly and accurately detect approaches from fraudsters and protect users in real time.

[0990] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0991] Step 1: Receiving a message

[0992] The server retrieves messages received by the user. Inputs include messages from the user's email or chat app. The server receives these in real time and passes them on to the next processing step. Specifically, it retrieves messages from the email server or chat server via an API.

[0993] Step 2: Save the message

[0994] The server stores the received message in a database. The input is the message received in step 1. The output is the message stored in the database. Specifically, a database management system such as MySQL or PostgreSQL is used to store the message content and metadata (sender information, receipt time, etc.).

[0995] Step 3: Parse the message

[0996] The server analyzes the stored messages using natural language processing tools. The input is the message stored in step 2. The output is the analysis results. Specifically, it uses "spaCy" and "NLTK" to tokenize the messages and tag them as parts of speech.

[0997] Step 4: Feature extraction

[0998] The server extracts message features from the analysis results. The input is the analysis result obtained in step 3. The output is the extracted features. Specifically, features are extracted such as specific keywords in the message (such as "bank," "account," or "link"), or the fact that the message was sent late at night.

[0999] Step 5: Assess the likelihood of fraud

[1000] The server inputs the extracted features into a generative AI model to assess the likelihood of fraud. The input is the features extracted in step 4. The output is a score indicating the likelihood of fraud. Specifically, the following prompt sentence is input into the generative AI model:

[1001] "Please rate the following message for possible fraud. It reads: 'Hello, there's a problem with your bank account. Click this link to check it now.'"

[1002] The generative AI model analyzes the message based on this prompt and scores it for its likelihood of fraud.

[1003] Step 6: Notify users

[1004] The server notifies the user of messages that are likely to be fraudulent based on the score returned by the generative AI model. The input is the score obtained in step 5. The output is a notification to the user. Specifically, it sends a push notification to the user's smartphone, warning them, "You have received a message that is likely to be fraudulent. Please be careful."

[1005] In this way, a system is realized in which the server, terminal, and user work together to detect approaches from fraudsters and protect users.

[1006] (Application example 2)

[1007] Next, a description will be given of Application Example 2 of Form 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."

[1008] In recent years, fraudulent activities via messages by fraudsters have been increasing, resulting in many cases of users becoming victims. Conventional fraud prevention systems have had problems with low accuracy in detecting fraudulent messages and difficulty in displaying warnings in real time. In addition, there are insufficient means for users to report fraudulent messages, making it difficult to detect and prevent fraud early. To solve these problems, there is a need for a system that can detect fraudulent messages with greater accuracy in real time and display warnings to users.

[1009] 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.

[1010] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for monitoring a user's message application and detecting potentially fraudulent messages in real time, means for displaying a warning for detected potentially fraudulent messages, and means for the user to report fraudulent messages. This enables highly accurate detection of fraudulent messages and display of warnings in real time.

[1011] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1012] "Natural communication" refers to communication using natural language and in a conversational format, as humans do on a daily basis.

[1013] "Fictitious contacts" refer to digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[1014] A "fraud criminal" is someone who attempts to fraudulently obtain money or property by deceiving others.

[1015] "Means of approach detection" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to identify potential fraudulent contacts.

[1016] "Fraudster tactics" refers to techniques used to induce fraudsters to continue their activities and to gather evidence.

[1017] "Means of collecting evidence of criminal activity" refers to technology that records the actions and messages of fraudsters and gathers evidence that can later be used in legal proceedings.

[1018] "Law reporting" refers to techniques that provide collected evidence to law enforcement agencies to assist in the arrest and prosecution of fraudsters.

[1019] "Messaging Application" means software that allows a User to send and receive text messages.

[1020] "Real-time detection measures" refers to technology that instantly analyzes potentially fraudulent messages and displays a warning to users.

[1021] "Warning measures" refers to technologies that notify users of potentially fraudulent messages and warn them.

[1022] "Means for reporting fraudulent messages" refers to technology that allows users to report potentially fraudulent messages they receive to the system, contributing to improving the accuracy of AI models.

[1023] As an embodiment of the present invention, a method for installing a fraud prevention assistant system on a smartphone will be described. The system monitors the user's messaging application, detects potentially fraudulent messages in real time, and displays a warning.

[1024] System configuration

[1025] The system consists of the following main components:

[1026] 1. Artificial Intelligence (AI) module: Contains machine learning models for detecting fraudulent messages.

[1027] 2. Message monitoring module: Monitor users' message applications in real time.

[1028] 3. Warning display module: Displays a warning to the user when a potentially fraudulent message is detected.

[1029] 4. Reporting module: Provides an interface for users to report scam messages.

[1030] Hardware and software used

[1031] Hardware: Smartphone

[1032] Software: Python, scikit-learn, joblib

[1033] Data processing and calculation

[1034] 1. Message preprocessing: Lowercase the message and remove special characters.

[1035] 2. Vectorization: Use TfidfVectorizer to convert the message into a numeric vector.

[1036] 3. Predict: Use a pre-trained Logistic Regression model to predict whether the message is fraudulent or not.

[1037] 4. Generate analysis results: Generate analysis results including message sending times and prediction results.

[1038] 5. Display warning: If a potentially fraudulent message is detected, a warning will be displayed to the user.

[1039] Specific examples

[1040] Message received by users: "Your account has been compromised. Click this link now."

[1041] The application detects this and displays a warning: "Warning: A potentially fraudulent message has been detected. Details: {'message': 'Your account has been compromised. Click this link now.', 'send_time': '2023-10-01 12:34:56', 'is_fraud': True}"

[1042] Prompt Sentence Examples

[1043] Generate Python code to develop a fraud prevention assistant app. This app monitors users' messaging apps and detects potentially fraudulent messages in real time. It uses a pre-trained Logistic Regression model and TfidfVectorizer to detect fraudulent messages.

[1044] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1045] Step 1:

[1046] The user receives the message.

[1047] Input: A new message that arrives in the user's messaging application.

[1048] Specific operation: A new message is received by the user's smartphone, and the message monitoring module detects the message.

[1049] Step 2:

[1050] The terminal preprocesses the message.

[1051] Input: The received message.

[1052] Data manipulation: Lowercase the message and remove special characters.

[1053] Output: The preprocessed message.

[1054] What happens: The device converts the message text to lowercase and removes special characters and unnecessary spaces.

[1055] Step 3:

[1056] The terminal vectorizes the preprocessed messages.

[1057] Input: The preprocessed message.

[1058] Data operation: Convert the message into a numeric vector using TfidfVectorizer.

[1059] Output: Vectorized messages.

[1060] Specific operation: The terminal uses TfidfVectorizer to convert the preprocessed message into a numeric vector.

[1061] Step 4:

[1062] The device uses an AI model to predict the vectorized message.

[1063] Input: Vectorized message.

[1064] Data Computation: Use a pre-trained Logistic Regression model to predict whether a message is fraudulent.

[1065] Output: Prediction results of fraudulent messages.

[1066] Specific operation: The device uses a logistic regression model to predict whether the vectorized message is a fraudulent message.

[1067] Step 5:

[1068] The device analyzes the prediction results and displays a warning.

[1069] Input: Predicted results of scam messages.

[1070] Data processing: Generate analysis results, including message sending times and prediction results.

[1071] Output: The warning message that is displayed to the user.

[1072] Specific operation: The device analyzes the prediction results and displays a warning message to the user if there is a possibility of fraud.

[1073] Step 6:

[1074] A user reports a fraudulent message.

[1075] Input: The scam message reported by the user.

[1076] Data processing: Reported messages are recorded and used to improve the accuracy of AI models.

[1077] Output: The updated AI model.

[1078] How it works: A user reports a fraudulent message, and the device records the message and adds it to the training data for the AI ​​model.

[1079] Example 3

[1080] Next, a description will be given of Example 3 of Form Example 3. 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."

[1081] Fraud crimes are becoming more sophisticated every year, making it difficult to identify fraudsters and collect evidence using conventional methods. It is also necessary to quickly detect approaches from fraudsters and take appropriate action. Furthermore, a system is needed to efficiently report collected evidence to the police.

[1082] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1083] In this invention, the server includes means for using artificial intelligence to conduct natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for storing the collected evidence in a database, means for analyzing messages using a generative AI model, and means for reporting suspected fraud to the police. This makes it possible to quickly detect approaches from fraudsters, efficiently collect and store evidence, and report to the police.

[1084] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1085] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to attract fraudsters.

[1086] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to determine the possibility of fraud.

[1087] "Fraudster tactics" are techniques used to induce fraudsters to continue their activities and gather more evidence.

[1088] "Means of collecting evidence of criminal activity" refers to technology that collects information such as the content of messages from fraudsters, the IP address of the sender, and the time of sending.

[1089] "Means for storing collected evidence in a database" refers to a technique for storing collected evidence information in a database so that it can be referenced later.

[1090] A "generative AI model" is an artificial intelligence model trained to perform natural language processing and data analysis.

[1091] "Means for analyzing messages" refers to technology that uses a generative AI model to analyze the content of received messages and determine whether they are fraudulent.

[1092] "Means of reporting to the police" refers to technology that provides collected evidence information to the police to help identify fraudsters and prevent crime.

[1093] MODE FOR CARRYING OUT THE INVENTION

[1094] The present invention relates to a system for detecting approaches from fraudsters, collecting evidence, and reporting to the police. A specific embodiment of this system will be described below.

[1095] System configuration

[1096] This system consists of three main components: a server, a terminal, and a user. The server receives messages from fraudsters, analyzes them, collects evidence, and reports them to the police. The terminal is a device that users use to access and operate the system. Users use the system to monitor approaches from fraudsters and take necessary action.

[1097] Hardware and software used

[1098] The server is a computer system equipped with a high-performance processor and a large amount of memory. A generative AI model (e.g., GPT-4) and a database (e.g., MySQL) are installed on the server. The generative AI model is used to analyze messages from fraudsters. The database is used to store collected evidence information.

[1099] Data processing and calculation

[1100] When the server receives a message from a fraudster, it first inputs the message into a generative AI model. The generative AI model analyzes the message's content and determines whether it is likely to be fraudulent. The server then collects metadata, such as the message's content, the sender's IP address, and the time of sending, and stores this information in a database. Finally, if the server suspects fraud based on the analysis results, it reports the collected evidence to the police.

[1101] Specific examples

[1102] For example, consider the case where a fraudster sends a phishing email to a user. The server collects the content of the phishing email, the sender's IP address, and the time of sending. The generative AI model analyzes this information and determines that it is suspected to be phishing. The server then reports the analysis results and the collected evidence to the police.

[1103] Prompt Sentence Examples

[1104] An example of a prompt sentence to input to the generative AI model is as follows:

[1105] Please analyze this message for possible fraud. The message reads as follows:

[1106] [Message content]

[1107] In this way, a system can be constructed in which the server can efficiently collect information from fraudsters and report it to the police. The flow of the identification process in the third embodiment will be explained with reference to FIG.

[1108] Step 1:

[1109] The server receives messages from fraudsters. The input is the message sent by the fraudster. The server periodically checks for new emails from the mail server and retrieves emails that are suspected to be fraudulent. The output is the received message data.

[1110] Step 2:

[1111] The server inputs the received message into the generative AI model for analysis. The input is the received message data. The generative AI model (e.g., GPT-4) analyzes the content of the message and determines whether it is fraudulent. The output is the analysis result. Specifically, the following prompt sentence is input into the generative AI model:

[1112] Please analyze this message for possible fraud. The message reads as follows:

[1113] [Message content]

[1114] Step 3:

[1115] The server collects not only the message content but also metadata such as the sender's IP address and the time of sending. The input is the received message data. The server extracts the sender's IP address and the time of sending from the email header information and collects this information as evidence. The output is the collected evidence information.

[1116] Step 4:

[1117] The server stores the collected evidence information in a database. The input is the collected evidence information. The server connects to a database (e.g., MySQL) and executes SQL queries to store the evidence information. The output is the evidence information stored in the database.

[1118] Step 5:

[1119] The server determines whether fraud is suspected based on the analysis results of the generative AI model. The input is the analysis results of the generative AI model. The server checks the analysis results and flags any suspected fraud. The output is the determination result of whether fraud is suspected.

[1120] Step 6:

[1121] If the server determines that fraud is suspected, it reports the collected evidence to the police. The input is the collected evidence and the judgment result. The server creates and sends a request to the police API to send the evidence. The output is the evidence reported to the police.

[1122] In this way, the server can efficiently collect, analyze, and report information from fraudsters to law enforcement.

[1123] (Application example 3)

[1124] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[1125] In modern society, fraud crimes are becoming increasingly sophisticated, requiring victims to respond quickly and effectively when they receive fraudulent messages. However, traditional methods require time and effort to analyze fraudulent messages and gather evidence, and reporting to the police is often delayed. This poses a challenge in identifying fraudsters and preventing crimes.

[1126] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1127] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing fraudulent messages and collecting evidence if there is a possibility of fraud, and means for transmitting the collected evidence to the police's API. This makes it possible to automatically analyze fraudulent messages when they are received, quickly collect evidence, and report it to the police.

[1128] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1129] "Natural communication" refers to interfaces and processes that allow humans and machines to interact seamlessly.

[1130] "Fictitious contacts" refer to digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[1131] A "fraud criminal" is someone who attempts to fraudulently obtain money or property by deceiving others.

[1132] "Means of approach detection" refers to technologies and methods that analyze messages and behavioral patterns from fraudsters to determine the likelihood of fraud.

[1133] "Fraudster tactics" refers to techniques and methods used to induce fraudsters to provide more information and to gather evidence.

[1134] "Evidence of criminal activity" refers to data such as message content, source IP address, and time of transmission that can be used to prove fraudulent activities committed by fraudsters.

[1135] "Means of reporting to police" refers to the techniques and methods used to provide collected evidence to police.

[1136] "Means for analyzing fraudulent messages" refers to techniques or methods for analyzing received messages and determining whether they are fraudulent.

[1137] "Means of sending to police API" refers to the technology or method for automatically sending collected evidence to police systems.

[1138] As an embodiment of the present invention, the following system is constructed.

[1139] First, the server has a means for natural communication using artificial intelligence. This artificial intelligence uses a generative AI model to collect information through dialogue with fraudsters. The server also has a means for generating fictitious contact information, which allows it to create email addresses and social networking service accounts to lure fraudsters.

[1140] Next, the server is equipped with a means for detecting approaches from fraudsters. This means determines the possibility of fraud by analyzing the content of messages and behavioral patterns from fraudsters. The server also includes a means for letting fraudsters go, encouraging them to provide more information.

[1141] Furthermore, the server has a means for collecting evidence of criminal activity. This evidence includes the fraudulent message, the source IP address, the time of transmission, etc. The collected evidence is provided to the police through a means for reporting. Specifically, the server has a means for analyzing the fraudulent message and collecting evidence if there is a possibility of fraud. The collected evidence is automatically reported to the police through a means for sending it to the police's API.

[1142] To implement this system, programs are written using programming languages ​​such as Python. The server automatically analyzes fraudulent messages when they are received, quickly collects evidence, and reports them to the police. The hardware used is a smartphone, and the software uses Python. The AI ​​model uses a pre-trained generative AI model.

[1143] For example, if a user receives a message saying, "Your bank account has been frozen. Click here for more information," the system automatically analyzes the message and determines it may be fraudulent. It then collects the message content, the sending IP address, and the time of receipt as evidence and reports it to the police.

[1144] An example of a prompt is as follows:

[1145] "When a fraudulent message is received, please create an application that analyzes the message and, if there is a possibility of fraud, collects evidence and reports it to the police. The message content, source IP address, and time of receipt will be collected as evidence and sent to the police API."

[1146] The above is an embodiment of the present invention.

[1147] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1148] Step 1:

[1149] A user receives a fraudulent message.

[1150] Input: Scam message

[1151] Output: Content of the scam message

[1152] Specific operation: The user's smartphone receives the fraudulent message and sends its contents to the server.

[1153] Step 2:

[1154] The server analyzes the fraudulent message.

[1155] Input: The content of the scam message

[1156] Output: Possibility of fraud judgement result

[1157] How it works: The server uses a generative AI model to analyze the message content and determine whether it is potentially fraudulent.

[1158] Step 3:

[1159] If the server determines that fraud is possible, it will collect evidence.

[1160] Input: Fraudulent message content, source IP address, and time of receipt

[1161] Output: Collected evidence (message content, source IP address, received time)

[1162] Specific operation: The server records the content of the fraudulent message, the sending IP address, and the time of receipt, and collects these as evidence.

[1163] Step 4:

[1164] The server sends the collected evidence to the police API.

[1165] Input: Collected evidence (message content, source IP address, time of receipt)

[1166] Output: Police report result (success or failure)

[1167] What it does: The server sends the collected evidence to the police API and checks whether the report was successful.

[1168] Step 5:

[1169] The server notifies the user of the report results.

[1170] Input: Police report result (success or failure)

[1171] Output: Notification to the user

[1172] Specific operation: The server notifies the user of the results of the police report. If the report is successful, it notifies the user by saying "The report has been submitted to the police." If the report is unsuccessful, it notifies the user by saying "The report failed."

[1173] 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.

[1174] "Example 1"

[1175] One embodiment of the present invention is an artificial intelligence system incorporating an emotion engine. This system recognizes emotions from a user's text messages and voice and adjusts the AI's response based on the emotions. For example, if a user sends a message expressing anger or frustration, the emotion engine recognizes this and the AI ​​generates a response that calms the user. Alternatively, if a user sends a message expressing joy or satisfaction, the emotion engine recognizes this and the AI ​​generates a response that shares the user's joy.

[1176] "Example 2"

[1177] The emotion engine can also track changes in a user's emotions over time, allowing the AI ​​to generate more appropriate responses that reflect the user's changing emotions. For example, if a user initially expresses joy but becomes frustrated over time, the emotion engine will recognize this change and the AI ​​will generate a response that reflects the user's frustration.

[1178] "Example 3"

[1179] Furthermore, the emotion engine has a learning function for recognizing the user's emotions. This learning function allows the emotion engine to learn how the user expresses their emotions and improve the accuracy of emotion recognition over time. For example, if a particular user uses specific words and expressions to express joy, the emotion engine can learn this and become able to more accurately recognize the joy of that user.

[1180] The processing flow of each embodiment will be described below.

[1181] "Example 1"

[1182] Step 1: A text message or voice message from the user is entered into the system.

[1183] Step 2: The emotion engine recognizes the user's emotion from the input.

[1184] Step 3: Based on the recognized emotion, the AI ​​generates a response.

[1185] "Example 2"

[1186] Step 1: A text message or voice message from the user is entered into the system.

[1187] Step 2: The emotion engine recognizes the user's emotions from the input and tracks changes in those emotions over time.

[1188] Step 3: Generate an AI response that corresponds to the change in emotion.

[1189] "Example 3"

[1190] Step 1: A text message or voice message from the user is entered into the system.

[1191] Step 2: The emotion engine recognizes the user's emotions from the input and learns how to express those emotions.

[1192] Step 3: Based on the learned emotional expression methods, the emotion engine improves the accuracy of emotion recognition.

[1193] Example 1

[1194] Next, a description will be given of Example 1 of Form 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."

[1195] Conventional AI systems struggled to communicate naturally and were unable to properly recognize user emotions and generate responses. Furthermore, they lacked the means to detect approaches from fraudsters and respond appropriately, making it difficult to prevent fraud and collect evidence. This made it difficult to ensure user safety and led to a risk of increasing fraud victims.

[1196] 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.

[1197] In this invention, the server includes means for conducting natural communication using artificial intelligence, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing messages using natural language processing technology, means for recognizing user emotions using an emotion engine, and means for generating responses based on the recognized emotions. This enables natural communication with users, effectively detecting approaches from fraudsters, and responding appropriately. Furthermore, recognizing user emotions and generating responses enables more human-like interactions and improves user satisfaction.

[1198] "Artificial intelligence" is the technology that enables computer systems to learn, reason, and self-correct by imitating human intelligence.

[1199] "Natural communication" refers to a dialogue between a human and a computer system using natural language, exchanging information in a manner similar to a dialogue between humans.

[1200] "Fictitious contacts" are non-existent contact information and digital accounts, such as email addresses or social media accounts, created to deceive fraudsters.

[1201] A "fraudster" is a person or organization that attempts to fraudulently obtain money or information by deceiving others.

[1202] "Approach detection" refers to technologies and algorithms that analyze contacts and messages from fraudsters to identify their intent.

[1203] "Letting fraudsters run wild" refers to intentionally maintaining contact with fraudsters in order to monitor their behavior and gather evidence.

[1204] "Means of collecting evidence of criminal activity" refers to technology that records the actions and messages of fraudsters and collects data for later use in legal proceedings.

[1205] A "police reporting method" is a method for providing collected evidence to law enforcement agencies and requesting appropriate action.

[1206] "Natural language processing technology" is a technology that enables computers to understand and generate human language, and involves analyzing and generating text.

[1207] An "emotion engine" is a technology that recognizes emotions from a user's text or voice and adjusts responses based on those emotions.

[1208] A "response generator" is a technique or algorithm that creates an appropriate reply based on the user's input and sentiment.

[1209] The present invention is a system for natural communication using artificial intelligence, which can detect approaches from fraudsters and respond appropriately. It can also recognize the user's emotions and generate responses based on those emotions. This system is implemented using the following hardware and software.

[1210] Hardware and software used

[1211] 1. Server:

[1212] The server hosts a generative AI model (e.g., OpenAI's GPT-4) and parses messages using natural language processing techniques.

[1213] The server uses an emotion engine to recognize the user's emotions and generate an appropriate response.

[1214] The server runs a fraud detection algorithm to detect approaches from fraudsters.

[1215] 2. Terminal:

[1216] A terminal is a device that allows a user to input and send messages, such as a smartphone or a computer.

[1217] The terminal receives the response from the server and displays it to the user.

[1218] 3. User:

[1219] Users interact with the system using terminals: they send text and voice messages and receive responses from the system.

[1220] Program processing

[1221] The server receives messages sent by users and analyzes the content of the messages using natural language processing techniques. Specifically, it performs tokenization, grammatical analysis, and semantic analysis of the messages. The server then uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the content and tone of the messages to identify the user's emotional state.

[1222] Based on the recognized emotion, the server uses a generative AI model to generate an appropriate response. The generated response is then tailored to create a natural dialogue according to the user's emotion. If an approach from a fraudster is detected, the server generates a fictitious contact and allows the fraudster to escape. This allows for the collection of evidence of criminal activity and the reporting of the collected evidence to the police.

[1223] Specific examples

[1224] For example, if a user sends a message saying "I'm very tired today," the following processing occurs:

[1225] 1. A user types "I'm so tired today" in a chat app and sends it.

[1226] 2. The server receives the message and performs tokenization, grammar analysis, and semantic analysis.

[1227] 3. The server uses an emotion engine to recognize "fatigue."

[1228] 4. The server generates a response such as "Take it easy today."

[1229] 5. The server sends the response to the user's device.

[1230] 6. The user sees the response "Take it easy today" on their device.

[1231] Prompt Sentence Examples

[1232] Examples of prompts to be input to a generative AI model include:

[1233] User: I'm very tired today.

[1234] AI: Please take it easy today.

[1235] In this way, AI can communicate naturally according to the user's emotions.

[1236] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1237] Step 1:

[1238] The server launches the AI ​​model.

[1239] The server loads and initializes a generative AI model (e.g., OpenAI's GPT-4). The server reads the model's parameters and settings and prepares it for interaction.

[1240] Input: AI model library and configuration file

[1241] Output: Initialized AI model

[1242] Specific operation: The server imports the AI ​​model library and initializes the model. The server loads the model settings.

[1243] Step 2:

[1244] A user sends a message.

[1245] Users use devices such as smartphones and personal computers to send text messages or voice messages to the system.

[1246] Input: User's text or voice message

[1247] Output: Message sent from terminal to server

[1248] Specific behavior: A user opens a chat app, types a message, and presses the send button.

[1249] Step 3:

[1250] The server receives the message and performs NLP processing.

[1251] The server receives messages sent by users and analyzes the content of the messages using natural language processing (NLP) techniques, specifically tokenizing, grammatical analysis, and semantic analysis of the messages.

[1252] Input: Message from the user

[1253] Output: Structure and meaning of the parsed message

[1254] Specific operation: The server calls the API to receive the message, and performs tokenization, grammatical analysis, and semantic analysis on the message.

[1255] Step 4:

[1256] The server uses an emotion engine to recognize emotions.

[1257] The server recognizes emotions from the user's messages using an emotion engine, which analyzes the content and tone of the messages to identify the user's emotional state.

[1258] Input: Parsed message structure and meaning

[1259] Output: Perceived emotional state

[1260] Specific operation: The server invokes the emotion engine to perform emotion analysis of the message and identify the emotional state.

[1261] Step 5:

[1262] The server generates an appropriate response.

[1263] The server generates appropriate responses based on the recognized emotions, using a generative AI model to engage in natural dialogue based on the user's emotions.

[1264] Input: Perceived emotional state

[1265] Output: The generated response

[1266] Specific operation: The server inputs a prompt sentence into the generative AI model, obtains the generated response, and formats it into an appropriate format.

[1267] Step 6:

[1268] The server generates fictitious contacts (in case of fraud detection).

[1269] If a fraudster attempts to make contact, the server uses fraud detection algorithms to detect the attempt and generate fictitious contact information.

[1270] Input: Results of the fraud detection algorithm

[1271] Output: Generated fictitious contacts

[1272] What it does: The server runs a fraud detection algorithm to detect fraud attempts and generate fictitious contacts.

[1273] Step 7:

[1274] The server sends the response to the user.

[1275] The server sends the generated response to the user's device, where the user can check the response from the AI.

[1276] Input: Generated response

[1277] Output: The response sent to the user's terminal

[1278] Specific operation: The server calls the API to send a response, and sends the response to the user's device. The user checks the response on the device.

[1279] (Application example 1)

[1280] Next, a description will be given of Application Example 1 of Form 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."

[1281] Conventional AI systems have had difficulty effectively detecting approaches from fraudsters and protecting users. Furthermore, they lacked the ability to recognize users' emotions and generate appropriate responses based on those emotions, making it difficult to reduce user stress. This has led to a need to improve user safety and satisfaction.

[1282] 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.

[1283] In this invention, the server includes means for using artificial intelligence to perform natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing a user's emotions and adjusting a response based on the emotions, means for detecting possible fraud and displaying a warning to the user, and means for generating an appropriate response based on the emotions. This makes it possible to effectively detect approaches from fraudsters, protect the user, and generate an appropriate response according to the user's emotions.

[1284] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1285] "Natural communication" refers to using language and interacting in a similar way to humans.

[1286] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[1287] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[1288] "Approach detection methods" are technologies used to identify and detect contact and activity from fraudsters.

[1289] "Fraudsters' lethal measures" are methods used by fraudsters to gather evidence while making it appear as if they are continuing their criminal activities.

[1290] "Evidence of criminal activity" is data or information that shows fraudulent activity committed by fraudsters.

[1291] A "police reporting method" is a method for providing collected evidence to law enforcement.

[1292] "Means for recognizing emotions" refers to technology that analyzes and identifies emotions from a user's text messages and voice.

[1293] A "means for tailoring responses" is a technique for generating appropriate responses based on recognized emotions.

[1294] "Potential fraud detection methods" are technologies that analyze messages and behavioral patterns to identify fraud risks.

[1295] A "means for displaying a warning" is a method for informing users of the risk of fraud.

[1296] "Means for generating appropriate responses" refers to technology that creates responses that correspond to the user's emotions and circumstances.

[1297] A system for implementing the present invention includes means for using artificial intelligence to communicate naturally, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing a user's emotions and adjusting responses based on the emotions, means for detecting possible fraud and displaying a warning to the user, and means for generating an appropriate response based on the emotions.

[1298] The server uses OpenAI's API to detect possible fraud and TextBlob to analyze the user's emotions. If there is a possibility of fraud, the server displays a warning to the user and generates an appropriate response based on the user's emotions. The hardware used is a smartphone, and the software used is OpenAI API and TextBlob.

[1299] For example, if a user receives the message "Your bank account has been fraudulently accessed. Please take action now," the server will detect the possibility of fraud and warn the user. Alternatively, if a user sends the message "I'm so happy today!", the server will use an emotion engine to recognize joy and generate an appropriate response.

[1300] Examples of prompts to input to a generative AI model include:

[1301] Fraud Detection: "Is this message potentially fraudulent?: Your bank account has been compromised. Take action now."

[1302] Emotional response: "The user is expressing the emotion of happiness. Generate an appropriate response."

[1303] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1304] Step 1:

[1305] The user receives the message.

[1306] The user's terminal sends the received message to the server.

[1307] Input: Message received by the user

[1308] Output: Message sent to the server

[1309] Step 2:

[1310] The server analyzes the received messages to detect possible fraud.

[1311] The server uses the OpenAI API to prompt the message content and determine the likelihood of fraud.

[1312] Input: The message sent to the server

[1313] Output: Judgment result on likelihood of fraud

[1314] Step 3:

[1315] The server will warn the user if there is a possibility of fraud.

[1316] The server generates a warning message and sends it to the user's terminal.

[1317] Input: Determination result regarding the possibility of fraud

[1318] Output: A warning message that is displayed to the user.

[1319] Step 4:

[1320] The server analyzes the sentiment of the received messages.

[1321] The server uses the TextBlob to parse the message for emotion and determine the type and intensity of the emotion.

[1322] Input: The message sent to the server

[1323] Output: Analysis results on the type and intensity of emotions

[1324] Step 5:

[1325] The server generates an appropriate response based on the emotion.

[1326] The server uses a generative AI model to input emotional prompts and generate appropriate responses.

[1327] Input: Analysis results on type and intensity of emotions

[1328] Output: The appropriate response generated

[1329] Step 6:

[1330] The server sends the generated response to the user's terminal.

[1331] The user's terminal displays the received response to the user.

[1332] Input: The appropriate response generated

[1333] Output: The response that is displayed to the user

[1334] Example 2

[1335] Next, a description will be given of Example 2 of Form 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."

[1336] Conventional fraud detection systems only analyze the content of messages and behavioral patterns from fraudsters, and are unable to respond to changes in users' emotions. This makes it difficult to understand how users feel about fraud and take appropriate action. In addition, there are limited means to improve the accuracy of fraud detection. This makes it difficult to fully ensure user safety.

[1337] 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.

[1338] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go free, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for tracking changes in the user's emotions, and means for generating responses corresponding to the user's emotions. This makes it possible to detect approaches from fraudsters with high accuracy and to respond appropriately according to the user's emotions.

[1339] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and perform tasks such as learning, reasoning, and problem-solving.

[1340] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[1341] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to assess the likelihood of fraud.

[1342] "Methods of letting fraudsters go free" refers to a technique for collecting more evidence by not taking immediate action against fraudsters but by allowing them to continue certain behaviors.

[1343] "Means for collecting evidence of criminal activity" refers to technology that records the actions and message content of fraudsters and stores them in a form that can later be used in legal proceedings.

[1344] "Means for reporting collected evidence to the police" refers to techniques for providing collected evidence in an appropriate format to the police or other law enforcement agencies.

[1345] The "means for tracking changes in user emotions" is a technology that collects user input and behavioral data and analyzes changes in user emotions using an emotion analysis model.

[1346] "Means for generating responses that correspond to the user's emotions" refers to technology that uses a generative AI model to create messages that match the user's emotions and provide an appropriate response.

[1347] MODE FOR CARRYING OUT THE INVENTION

[1348] The present invention is a system for detecting approaches from fraudsters and responding to changes in a user's emotions. A specific embodiment of this system will be described below.

[1349] A system to detect approaches from fraudsters

[1350] The server analyzes the content and behavioral patterns of messages from fraudsters. Specifically, the server uses an AI model to analyze the wording, topic, and sending time of messages from fraudsters. This analysis uses natural language processing (NLP) technology. For example, Python libraries such as NLTK and spaCy can be used. This allows the server to understand the behavioral patterns of fraudsters and detect messages that are likely to be fraudulent.

[1351] Examples:

[1352] An example of a scam message: "Your account has been compromised. Click this link to check now."

[1353] An example prompt you might use is: "Please rate the likelihood that this message is a scam."

[1354] Tracking user emotional changes with an emotion engine

[1355] The device tracks changes in the user's emotions over time. The emotion engine uses a sentiment analysis model to analyze emotions from the user's input and behavior. For example, it can use Microsoft Azure's Text Analytics API or Google Cloud Natural Language API. This allows the device to recognize changes in the user's emotions in real time and generate appropriate responses.

[1356] Examples:

[1357] An example of a change in user emotions: At first, the user expresses joy, saying, "This service is great!", but later expresses dissatisfaction, saying, "The recent update has made it difficult to use."

[1358] An example prompt you might use is: "Generate an appropriate response if the user's emotion changes from happy to frustrated."

[1359] System configuration

[1360] This system is implemented using the following hardware and software.

[1361] Server: A server with a powerful processor and large memory capacity is used. The server analyzes fraudulent messages and detects fraudulent patterns.

[1362] Device: A device is used to collect user input and behavioral data and track emotional changes. The device has sufficient processing power to run the emotion engine.

[1363] Software: Use software libraries and APIs such as Python, NLTK, spaCy, Microsoft Azure Text Analytics API, Google Cloud Natural Language API, etc.

[1364] Example

[1365] A specific example of this system will be described below.

[1366] 1. Analysis of the fraudulent message:

[1367] The server retrieves messages received by the user.

[1368] The server uses NLTK to tokenize messages and spaCy to classify topics.

[1369] The server uses a generative AI model to assess the likelihood that a message is fraudulent.

[1370] 2. Tracking user emotional changes:

[1371] The device collects user input and behavioral data in real time.

[1372] The device uses the Microsoft Azure Text Analytics API to analyze user sentiment and the Google Cloud Natural Language API to classify sentiment.

[1373] The device uses a generative AI model to generate appropriate responses that correspond to the user's emotions.

[1374] In this way, the system can detect approaches from fraudsters with high accuracy and respond appropriately based on the user's emotions.

[1375] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1376] Processing steps of a system to detect approaches from fraudsters

[1377] Step 1: Receiving a message

[1378] The server retrieves messages received by users, which can be collected from different platforms such as email, SMS, and chat apps.

[1379] Input: Message received by the user

[1380] Output: Message data stored on the server

[1381] Specific behavior:

[1382] The server retrieves new mail from the email server using the IMAP protocol.

[1383] Use the chat app's API to get new messages.

[1384] Step 2: Parse the message

[1385] The server analyzes the received messages using natural language processing (NLP) techniques, specifically extracting information such as the message's wording, topic, and time of sending.

[1386] Input: Message data stored on the server

[1387] Output: Parsed message feature data

[1388] Specific behavior:

[1389] The server uses the NLTK library in Python to tokenize the messages.

[1390] Use spaCy to classify the topic of messages.

[1391] Step 3: Detect fraud patterns

[1392] The server uses the analysis results to detect fraud patterns, and the AI ​​model is trained using a dataset of past fraudulent messages.

[1393] Input: Parsed message feature data

[1394] Output: Message data assessed for fraudulent potential

[1395] Specific behavior:

[1396] The server uses a trained generative AI model (e.g., BERT or GPT-3) to assess the likelihood that a message is fraudulent.

[1397] If there is a high possibility of fraud, a warning will be sent to the user.

[1398] Processing step of tracking user emotion changes by emotion engine

[1399] Step 1: Collecting User Input

[1400] The device collects user input and behavior data, including interactions such as text entry, clicking, and scrolling.

[1401] Input: User input and behavioral data

[1402] Output: User behavior data stored on the device

[1403] Specific behavior:

[1404] The terminal captures text entered by the user in real time.

[1405] Log user click and scroll data.

[1406] Step 2: Sentiment Analysis

[1407] The device analyzes the user's emotions based on the collected data, and the emotion engine uses an emotion analysis model to identify the user's emotions.

[1408] Input: User behavior data stored on the device

[1409] Output: Analyzed user emotion data

[1410] Specific behavior:

[1411] The device uses Microsoft Azure's Text Analytics API to extract sentiment from the user's text input.

[1412] Classify user sentiment using the Google Cloud Natural Language API.

[1413] Step 3: Track your emotions

[1414] The device tracks changes in the user's emotions over time, allowing it to understand how the user's emotions are changing.

[1415] Input: Parsed user emotion data

[1416] Output: User emotion data stored in time series

[1417] Specific behavior:

[1418] The terminal stores the user's emotion data in a time-series database.

[1419] It monitors emotional changes in real time and generates alerts if abnormal changes are detected.

[1420] Step 4: Generate an appropriate response

[1421] The device generates appropriate responses based on the user's emotional changes, using a generative AI model to create messages that match the user's emotions.

[1422] Input: User emotion data stored in time series

[1423] Output: The generated response message

[1424] Specific behavior:

[1425] The device uses a trained generative AI model (e.g., GPT-3) to generate responses that correspond to the user's emotions.

[1426] The generated response is displayed to the user.

[1427] (Application example 2)

[1428] Next, a description will be given of Application Example 2 of Form 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."

[1429] In recent years, the sophisticated methods used by fraudsters have become more prevalent, increasing the risk of users receiving fraudulent messages. Furthermore, there have been many cases where users have been unable to respond appropriately to fraudulent messages and have become victims. Furthermore, there is a lack of systems that provide appropriate advice and warnings in response to changes in users' emotions, making it difficult to ensure user safety.

[1430] 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.

[1431] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing messages received by users in real time, means for detecting potentially fraudulent messages and issuing warnings, means for tracking changes in the user's emotions, and means for providing appropriate advice and warnings. This allows users to respond quickly and appropriately when they receive fraudulent messages, making it possible to prevent fraud damage before it occurs.

[1432] "Artificial intelligence" is the technology that enables computer systems to learn, reason, and self-correct by imitating human intelligence.

[1433] "Natural communication" refers to interfaces and means that allow humans and machines to converse seamlessly.

[1434] "Fictitious contacts" refer to digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[1435] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[1436] "Means of approach detection" refers to technologies and methods that analyze the content of messages and behavioral patterns from fraudsters to identify potential fraud.

[1437] "Fraudster tactics" refers to techniques and methods used to intentionally target fraudsters and gather further evidence.

[1438] "Means of collecting evidence of criminal activity" refers to technologies and methods that record the actions and messages of fraudsters and store them in a form that can later be used in legal proceedings.

[1439] "Means of reporting to law enforcement" refers to the techniques and methods for providing collected evidence to police or law enforcement agencies in an appropriate format.

[1440] "Real-time analysis means" refers to technologies and methods that instantly analyze messages received by users and assess their likelihood of fraud.

[1441] "Warning measures" refers to techniques or methods for alerting users when a potentially fraudulent message is detected.

[1442] "Means for tracking emotional changes" refers to techniques or methods for monitoring a user's emotional state over time and recording those changes.

[1443] "Means for providing appropriate advice and warnings" refers to technologies and methods for providing optimal countermeasures or warnings based on the user's emotional state and the content of the message.

[1444] As an embodiment of the present invention, the fraud detection system is implemented as an application installed on a smartphone, as will be described in detail below.

[1445] System Program

[1446] The system includes a program with the following main functions:

[1447] 1. Natural communication using artificial intelligence:

[1448] The server uses a generative AI model to interact naturally with the user, generating appropriate responses to user input.

[1449] 2. Generate fictitious contacts:

[1450] The server generates fictitious email addresses and social media accounts to attract fraudsters, and these contacts are used to monitor the fraudsters' activities.

[1451] 3. Detecting fraudster approaches:

[1452] The server analyzes messages received by users in real time to detect potentially fraudulent messages, including analyzing message content, time of sending, and behavioral patterns.

[1453] 4. Letting fraudsters run wild:

[1454] The server will intentionally react to fraudsters and collect further evidence, a process that will provide a detailed record of fraudsters' actions.

[1455] 5. Collecting evidence of criminal activity:

[1456] The server records the actions and messages of fraudsters and stores them in a form that can later be used in legal proceedings.

[1457] 6. Police Report:

[1458] The server provides the collected evidence to police and law enforcement agencies in an appropriate format.

[1459] 7. Tracking user emotional changes:

[1460] The server monitors the user's emotional state over time and records changes, allowing it to provide appropriate advice and warnings in response to changes in the user's emotions.

[1461] Hardware and software used

[1462] Hardware: Smartphone

[1463] Software: Python, TextBlob library, generative AI model

[1464] Data processing and calculation

[1465] The server performs the following data processing and calculations to analyze messages received by users in real time.

[1466] 1. Message content analysis:

[1467] Analyzes the text of messages to detect potentially fraudulent keywords and patterns.

[1468] 2. Sentiment analysis:

[1469] Use the TextBlob library to analyze the sentiment of the message and assess the emotional state of the user.

[1470] 3. Behavioral Pattern Analysis:

[1471] Analyze the time and frequency of messages sent to identify fraudsters' behavioral patterns.

[1472] Specific examples

[1473] Received message: "Congratulations! You've won a prize."

[1474] The app responds: "Warning: This message may be a scam."

[1475] Prompt Sentence Examples

[1476] Develop an application that analyzes messages received by users in real time, detects potentially fraudulent messages, and issues a warning. Also, analyze the sentiment of the message and add a warning feature if the message is likely to evoke negative sentiment.

[1477] The above is an embodiment of the present invention.

[1478] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1479] Step 1:

[1480] The server retrieves messages received by the user. The input is the message received by the user, and the output is the message text to be analyzed. Specifically, the server retrieves messages from the user's mailbox or SNS account.

[1481] Step 2:

[1482] The server analyzes the retrieved message text to detect keywords and patterns that may indicate fraud. The input is the message text, and the output is a flag indicating whether it is likely to be fraudulent. Specifically, the server uses regular expressions to search for fraudulent keywords in the message.

[1483] Step 3:

[1484] The server analyzes the sentiment of a message using the TextBlob library. The input is the message text and the output is a sentiment score. Specifically, the server passes the message text to TextBlob and obtains a positive or negative sentiment score.

[1485] Step 4:

[1486] The server analyzes the time and frequency of message sending to identify fraudsters' behavioral patterns. The input is message metadata (sent time, sender information, etc.), and the output is the evaluation result of the behavioral pattern. Specifically, the server analyzes message timestamps to detect anomalous sending patterns.

[1487] Step 5:

[1488] If the server detects a potentially fraudulent message, it issues a warning to the user. The input is a flag indicating whether the message is potentially fraudulent and an emotion score, and the output is a warning message. Specifically, the server sends a notification to the user's smartphone to warn of the possibility of fraud.

[1489] Step 6:

[1490] The server monitors the user's emotional state over time and records the changes. The input is the history of emotional scores, and the output is the pattern of emotional changes. Specifically, the server periodically records the emotional scores and graphs the changes.

[1491] Step 7:

[1492] The server then provides optimal countermeasures and warnings based on the user's emotional state and the content of the message. The input is the pattern of emotional changes and the message content, and the output is advice or a warning message. Specifically, the server uses a generative AI model to generate an appropriate response for the user.

[1493] The above are the specific processing steps of the fraud detection system.

[1494] Example 3

[1495] Next, a description will be given of Example 3 of Form Example 3. 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."

[1496] Fraud crimes are becoming more sophisticated every year, making it difficult to identify fraudsters and collect evidence using conventional methods. Furthermore, there is a lack of systems to effectively analyze information obtained through communication with fraudsters and report it to the police. Furthermore, there is a need to recognize users' emotions and respond appropriately.

[1497] The identification process by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes a means for natural communication using artificial intelligence, a means for generating fictitious contact information, a means for detecting approaches from fraudsters, a means for letting fraudsters go unpunished, a means for collecting evidence of criminal activity, a means for reporting the collected evidence to the police, a means for receiving messages from fraudsters, a means for storing the content of the received messages, the sender's IP address, and the sending time in a database, a means for analyzing the messages using a natural language processing engine, a means for recognizing the user's emotions using an emotion engine, and a means for formatting the collected data and sending a report to the police. This makes it possible to identify fraudsters, effectively collect evidence, and report to the police. It also recognizes the user's emotions and takes appropriate action.

[1498] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1499] "Natural communication" refers to technology that allows humans and computers to exchange information smoothly and without any sense of awkwardness when interacting with each other.

[1500] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[1501] A "fraudster" is someone who attempts to fraudulently obtain money or information by deceiving others.

[1502] "Approach detection methods" are technologies that analyze the content of messages and behavioral patterns from fraudsters to identify fraud attempts.

[1503] "Methods of letting fraudsters go free" refers to a technique of not taking immediate action against fraudsters, but observing them for a certain period of time to gather more information.

[1504] "Evidence of criminal activity" is information or data that proves the fraudulent activities of a fraudster.

[1505] "Means of reporting to the police" are the methods and techniques used to provide the collected evidence to the police.

[1506] "Means for receiving messages" refers to the technology that allows the server to receive messages sent by fraudsters.

[1507] A "database" is a system that systematically stores information and allows it to be searched and retrieved as needed.

[1508] A "natural language processing engine" is a technology for understanding and analyzing human language.

[1509] An "emotion engine" is a technology for recognizing and analyzing a user's emotions.

[1510] The "means for transmitting the report" is the technology used to format and transmit the collected data electronically to the police.

[1511] This invention is a system for collecting information obtained through communication with fraudsters and reporting it to the police. A specific embodiment of this system will be described below.

[1512] The server uses email and chat applications to receive messages from fraudsters. The SMTP protocol is used to receive emails, and WebSocket is used to receive messages from chat applications. Received messages are temporarily stored in memory.

[1513] The server then stores information such as the content of the received message, the IP address of the sender, and the time of sending in a database. This database uses a relational database management system (RDBMS) such as MySQL. The stored data is stored in an appropriate format for later analysis.

[1514] The server analyzes the received messages using a natural language processing (NLP) engine, which uses generative AI models such as Google's BERT and OpenAI's GPT-3. The results are used to identify fraudulent patterns and messages containing specific keywords, such as "bank account," "fraudulent use," and "click on link."

[1515] Furthermore, the server uses an emotion engine to recognize the user's emotions. This emotion engine extracts emotions from the user's messages and improves the accuracy of emotion recognition through learning functions. For example, if a particular user uses specific words and expressions to express joy, the emotion engine can learn this and more accurately recognize the user's joy.

[1516] Finally, the server reports the collected evidence to the police via email or a dedicated reporting system. The server formats the collected data and creates a report, which is sent to the police via email or uploaded to a dedicated reporting system.

[1517] As a concrete example, consider a case where a fraudster sends a message saying, "Your bank account has been fraudulently accessed. Please click this link now to check." The server receives this message and stores the sender's IP address and the time of sending in a database. The server then uses an NLP engine to analyze the message and identify keywords such as "bank account," "fraudulent access," and "click link." The emotion engine recognizes the emotion the user felt when receiving this message and determines that they are likely feeling fear or anxiety. The collected information is reported to the police and used to identify fraudsters and prevent crime.

[1518] Example prompt sentence:

[1519] "Analyze messages from fraudsters to identify the IP addresses from which they were sent and the time of sending. Also, recognize user sentiment and gather information to report to law enforcement."

[1520] In this way, the server effectively collects information obtained through communication with fraudsters and reports it to the police. The flow of the identification process in the third embodiment will be described with reference to FIG.

[1521] Step 1: Receiving a message

[1522] The server receives messages from fraudsters. The input is email or a message from a chat application. The server receives email using the SMTP protocol and messages from chat applications using WebSocket. The received messages are temporarily stored in memory. The output is the content of the received messages.

[1523] Step 2: Save your data

[1524] The server stores information such as the content of the received message, the sender's IP address, and the time of sending in a database. The input is the content of the message received in step 1, the sender's IP address, and the time of sending. The server uses a relational database management system (RDBMS) such as MySQL to store this information in the database in an appropriate format. The output is the information stored in the database.

[1525] Step 3: Analyze the data

[1526] The server uses a natural language processing (NLP) engine to analyze the collected data. The input is the content of messages stored in a database. The server uses generative AI models such as Google's BERT or OpenAI's GPT-3 to analyze the content of the messages. The analysis results in identifying fraud patterns and specific keywords. The output is the analysis results.

[1527] Step 4: Emotion Recognition

[1528] The server uses an emotion engine to recognize the user's emotions. The input is the user's message. The server inputs the user's message into the emotion engine and extracts the emotion. The emotion engine improves the accuracy of emotion recognition through a learning function. The output is the user's emotion recognition result.

[1529] Step 5: File a police report

[1530] The server reports the collected evidence to the police. The inputs are the analysis results and emotion recognition results. The server formats the collected data and creates a report. The report is sent to the police via email or uploaded to a dedicated reporting system. The output is the report sent to the police.

[1531] As a concrete example, consider the situation where you receive a message from a fraudster saying, "Your bank account has been compromised. Click this link now to check."

[1532] Step 1:

[1533] The server receives a message from the fraudster using the SMTP protocol, which reads: "Your bank account has been compromised. Click this link now to check."

[1534] Step 2:

[1535] The server stores the received message content ("Your bank account has been fraudulently used. Click this link now to check"), the sender's IP address (192.168.1.1), and the sending time (2023-10-01 12:00:00) in a database.

[1536] Step 3:

[1537] The server inputs the message content into an NLP engine and identifies keywords such as "bank account," "fraudulent use," and "click on link." The analysis results indicate that the message is "highly likely to be fraudulent."

[1538] Step 4:

[1539] The server inputs the user's message into the emotion engine and determines that the user is likely feeling fear or anxiety. The emotion recognition result is "fear" or "anxiety."

[1540] Step 5:

[1541] The server formats the collected data and creates a report, which is then sent to the police via email. The report contains information about the likely fraudulent message, the sender's IP address, the time of sending, and the user's emotion recognition results.

[1542] (Application example 3)

[1543] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[1544] Fraud crimes are becoming more sophisticated every year, and there are an increasing number of cases where victims fall victim to fraud without realizing the scam methods. Furthermore, the difficulty of quickly and accurately collecting evidence of fraud and reporting it to the police often delays the identification and arrest of fraudsters. Furthermore, the lack of means to properly recognize the victim's emotions and assess the possibility of fraud makes it difficult to prevent damage before it occurs. To solve these issues, a system is needed that can analyze suspicious messages and calls in real time and issue a warning to users.

[1545] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1546] In this invention, the server includes means for using artificial intelligence to conduct natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing the user's emotions and assessing the possibility of fraud, and means for displaying a warning to the user if fraud is suspected. This makes it possible to analyze messages or calls suspected of fraud in real time and issue a warning to the user. Furthermore, by quickly and accurately collecting evidence of fraud and reporting it to the police, fraudsters can be quickly identified and arrested. Furthermore, by appropriately recognizing the user's emotions and assessing the possibility of fraud, damage can be prevented before it occurs.

[1547] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1548] "Natural communication" refers to interfaces and methods designed to allow humans and machines to interact seamlessly.

[1549] "Fictitious contacts" are digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[1550] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to determine whether or not there is suspicion of fraud.

[1551] "Methods of letting fraudsters get away" are techniques for extracting information and collecting evidence from fraudsters without raising suspicion.

[1552] "Means for collecting evidence of criminal activity" refers to technology that records information obtained from fraudsters, such as messages, the IP address of the sender, and the time of sending.

[1553] "Means of reporting collected evidence to the police" refers to techniques for providing collected evidence of fraud to the police in an appropriate format.

[1554] The "means for recognizing a user's emotions and assessing the possibility of fraud" is a technology that analyzes a user's emotions and assesses the possibility of fraud based on those emotions.

[1555] The "means for displaying a warning to the user when fraud is suspected" is a technique for displaying a warning to the user when it is determined that fraud is suspected.

[1556] A system for implementing this invention has the following configuration: a server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact points, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for recognizing a user's emotions and evaluating the possibility of fraud, and means for displaying a warning to the user if fraud is suspected.

[1557] Hardware and Software Configuration

[1558] 1. Hardware:

[1559] Smartphone (iOS or Android)

[1560] Smart glasses (e.g. Google Glass)

[1561] 2. Software:

[1562] AI models (e.g. GPT-4)

[1563] Emotion engine (e.g. Affectiva SDK)

[1564] Message analysis engine (e.g. IBM Watson)

[1565] Data processing and calculation

[1566] The server analyzes messages and call content received by the user in real time. It uses an AI model (GPT-4) to analyze the message content and determine whether it is suspected of fraud. If fraud is suspected, it collects the sender's IP address, the time of sending, and the message content. It uses an emotion engine (Affectiva SDK) to analyze the user's emotions and further evaluate the possibility of fraud. If fraud is suspected, it displays a warning to the user. If necessary, it reports the collected evidence to the police.

[1567] Specific examples

[1568] When a user receives a message on their smartphone, the fraud prevention AI assistant analyzes the message content and, if it detects a suspected fraud, displays a warning to the user, while also collecting evidence of the fraud and reporting it to the police.

[1569] Prompt Sentence Examples

[1570] Analyze messages received by users to determine whether they are suspected to be fraudulent. If so, collect the sending IP address, time of sending, and message content to generate data for reporting to law enforcement. Also, analyze user sentiment to assess the likelihood of fraud.

[1571] In this way, the fraud prevention AI assistant protects users from fraud and provides a system that helps prevent crime.

[1572] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1573] Step 1:

[1574] The user receives the message.

[1575] Input: Message content, sender information

[1576] Output: Message content, sender information

[1577] What it does: The user's smartphone or smart glasses receives a new message and sends the message's content and sender information to the server.

[1578] Step 2:

[1579] The server analyzes the message content.

[1580] Input: Message content

[1581] Output: Suspected fraud?

[1582] How it works: The server uses an AI model (GPT-4) to analyze the message content and determine whether it is suspected of fraud. If it is, it passes the information to the next step.

[1583] Step 3:

[1584] If the server suspects fraud, it will collect evidence.

[1585] Input: Message content, sender information, sending time

[1586] Output: Collected evidence data

[1587] Specific operation: If the server suspects fraud, it records the sender's IP address, the time of sending, and the message content, and saves them as evidence.

[1588] Step 4:

[1589] The server analyzes the user's emotions.

[1590] Input: User reaction data (voice, facial expressions, etc.)

[1591] Output: User's sentiment rating

[1592] How it works: The server uses an emotion engine (Affectiva SDK) to analyze the user's reaction data and evaluate the user's emotions, which can then be used to further assess the likelihood of fraud.

[1593] Step 5:

[1594] The server displays a warning to the user if fraud is suspected.

[1595] Input: Suspected fraud, user sentiment rating

[1596] Output: Warning message

[1597] Specific behavior: If the server determines that fraud is suspected and the user's emotion rating indicates the possibility of fraud, a warning message will be displayed on the user's smartphone or smart glasses.

[1598] Step 6:

[1599] Report any evidence collected by the server to the police.

[1600] Input: Collected evidence data

[1601] Output: Report data sent to police

[1602] Specific operations: The server converts the collected evidence data into an appropriate format, generates data for reporting to the police, and sends it.

[1603] In this way, the fraud prevention AI assistant protects users from fraud and provides a system that helps prevent crime.

[1604] 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.

[1605] 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> ) 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.

[1606] Another example of generative AI is Gemini (internet search engine). <url: https: gemini.google.com ?hl="ja">) are listed.

[1607] 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.

[1608] [Third embodiment]

[1609] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1610] 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.

[1611] 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).

[1612] 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.

[1613] 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.

[1614] 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).

[1615] 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.

[1616] 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.

[1617] 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.

[1618] 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.

[1619] 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.

[1620] 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.

[1621] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[1622] "Example 1"

[1623] One embodiment of the present invention is a system in which artificial intelligence (AI) can communicate naturally. This AI can use natural language processing (NLP) technology to have conversations similar to those of humans. The AI ​​also has the ability to generate fictitious contact information. This allows the AI ​​to detect attempts by fraudsters to contact the AI ​​and allow them to escape.

[1624] "Example 2"

[1625] Another embodiment of the present invention is a system for detecting approaches from fraudsters. This system detects approaches from fraudsters by analyzing the content of messages and behavioral patterns from fraudsters. Specifically, AI analyzes the wording, topic, and sending time of messages from fraudsters, and from this information, it understands the behavioral patterns of fraudsters.

[1626] "Example 3"

[1627] Furthermore, one embodiment of the present invention is a system that collects evidence of criminal activity and reports it to the police. AI collects information obtained through communication with fraudsters as evidence and reports it to the police as necessary. Specifically, AI collects information such as messages from fraudsters, the sender's IP address, and the time of transmission, and provides this information to the police. This can be useful in identifying fraudsters and preventing crimes.

[1628] The processing flow of each embodiment will be described below.

[1629] "Example 1"

[1630] Step 1: AI uses natural language processing (NLP) techniques to engage in human-like dialogue.

[1631] Step 2: The AI ​​generates fictitious contacts. These contacts can be email addresses or social media accounts.

[1632] It is a digital account such as a credit card.

[1633] Step 3: If a fraudster attempts to contact the AI, the AI ​​will detect the attempt.

[1634] Step 4: The AI ​​lets the fraudsters go free while collecting evidence of their criminal activity.

[1635] "Example 2"

[1636] Step 1: AI analyzes the content of messages and behavioral patterns from fraudsters.

[1637] Step 2: The AI ​​analyzes the scammers' messages for wording, topic, and time of sending.

[1638] Step 3: Use this information to understand the behavioral patterns of fraudsters and detect their approaches.

[1639] "Example 3"

[1640] Step 1: The AI ​​collects evidence from communications with fraudsters.

[1641] Step 2: The AI ​​collects information from the fraudsters, such as the message, the IP address from which it was sent, and the time of sending.

[1642] Step 3: The AI ​​will provide the collected information to law enforcement, which can help identify fraudsters and prevent crime.

[1643] Example 1

[1644] Next, a description will be given of Example 1 of Form 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."

[1645] Conventional AI systems have had difficulty effectively detecting approaches from fraudsters and responding appropriately. They also lacked the means to generate fictitious contact information to lure fraudsters and collect evidence of criminal activity. Furthermore, they needed to be able to detect signs of fraud and generate appropriate responses while engaging in natural communication.

[1646] 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.

[1647] In this invention, the server includes means for conducting natural communication using artificial intelligence, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing input data using natural language processing technology, means for generating responses using a generative AI model, means for executing an algorithm for detecting signs of fraud, and means for sending the generated responses or fictitious contact information to the terminal. This makes it possible to effectively detect approaches from fraudsters and deal with them appropriately.

[1648] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1649] "Natural communication" is communication that mimics the dialogue and conversation that humans have on a daily basis and that occurs without any sense of incongruity.

[1650] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[1651] A "fraudster" is someone who attempts to fraudulently obtain money or information by deceiving others.

[1652] "Approach detection methods" are techniques and methods used to identify and detect contact or attempts by fraudsters.

[1653] "Means to keep fraudsters going" are methods used to encourage fraudsters to continue committing crimes and to monitor their behavior.

[1654] "Means for collecting evidence of criminal activity" refers to techniques or methods that record the actions or communications of fraudsters so that they can later be used as evidence.

[1655] A "police reporting method" is a method of providing collected evidence to law enforcement agencies to assist in the investigation or prosecution of a crime.

[1656] "Natural language processing technology" is a technology that allows computers to understand and analyze human language.

[1657] A "generative AI model" is an artificial intelligence model that generates new text or responses based on input data.

[1658] A "fraud indicator detection algorithm" is a computational procedure designed to identify patterns or characteristics that indicate possible fraudulent activity.

[1659] A "terminal" is a device used by a user, such as a computer or smartphone.

[1660] MODE FOR CARRYING OUT THE INVENTION

[1661] This invention is a system that uses artificial intelligence to communicate naturally, detect approaches from fraudsters, and respond appropriately. This system works in cooperation with three parties: a server, a terminal, and a user.

[1662] Hardware and software used

[1663] Hardware: High-performance servers (e.g., virtual servers from cloud service providers)

[1664] Software: Natural language processing engines (e.g., natural language processing APIs), generative AI models (e.g., generative AI models), fraud detection algorithms

[1665] System configuration

[1666] 1. Server:

[1667] The server executes a program for natural communication using artificial intelligence.

[1668] The server analyzes the input data from the user using natural language processing techniques.

[1669] The server uses a generative AI model to generate an appropriate response.

[1670] The server runs algorithms that detect signs of fraud and detect approaches from fraudsters.

[1671] The server generates fictitious contact details to allow fraudsters to operate.

[1672] The server collects evidence of criminal activity and reports the collected evidence to the police.

[1673] 2. Terminal:

[1674] The terminal provides an interface for the user to enter input.

[1675] The terminal transmits the user's input data to the server.

[1676] The terminal receives the response or the fictitious contact from the server and displays it to the user.

[1677] 3. User:

[1678] A user uses a terminal to input questions or requests to the system.

[1679] The user checks the response from the server displayed on the terminal.

[1680] Specific examples

[1681] For example, if a user types "Hello, how's the weather today?", the device sends this input to the server. The server uses a natural language processing engine to parse the input data and uses a generative AI model to generate the response "Hello, the weather is sunny today." After verifying that there are no signs of fraud, the server sends this response to the device, which displays it to the user.

[1682] On the other hand, if a fraudster types "I want to add a new contact, how do I do that?", the server will detect the signs of fraud and generate a fictitious contact and send it to the device, which will then display this fictitious contact to the fraudster.

[1683] Prompt Sentence Examples

[1684] "Hello, how's the weather today?"

[1685] "I want to add a new contact, how do I do that?"

[1686] "I've been getting a lot of scam calls lately, what should I do?"

[1687] In this way, the present invention makes it possible to effectively detect approaches from fraudsters and deal with them appropriately.

[1688] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1689] Step 1:

[1690] The user enters input from the terminal.

[1691] The user uses the terminal's interface to input a question or request, for example, "Hello, what's the weather like today?" The input data is sent to the terminal in text form.

[1692] Step 2:

[1693] The terminal sends the input data to the server.

[1694] The terminal sends the text data entered by the user to the server via the Internet. At this time, the data is encrypted before being sent. The input data reaches the server.

[1695] Step 3:

[1696] The server receives the input data and analyzes it using a natural language processing (NLP) engine.

[1697] The server receives input data sent from the device. The received data is analyzed using a natural language processing engine. Specifically, the intent and sentiment of the input data are extracted. For example, the input "Hello, how's the weather today?" is analyzed and determined to be a question about the weather.

[1698] Step 4:

[1699] The server generates a response using a generative AI model based on the analysis results.

[1700] The server uses the generative AI model to generate an appropriate response based on the analysis results of the NLP engine. For example, if the analysis result is a question about the weather, the generative AI model will generate the response "Hello, the weather is sunny today." The generated response is stored in text format on the server.

[1701] Step 5:

[1702] The server runs fraud detection algorithms to detect signs of fraud.

[1703] Before sending the generated response to the user, the server runs a fraud detection algorithm. This algorithm detects whether the input data contains any signs of fraud. For example, if the input "I want to add a new contact, how do I do this?" indicates signs of fraud, the server will detect this. The detection results are stored on the server.

[1704] Step 6:

[1705] Send a server-generated response or fictitious contact to the device.

[1706] If fraud signs are detected, the server generates a fictitious contact and sends it to the device. If no fraud signs are detected, the server sends the generated response to the device as is. The transmitted data is encrypted before reaching the device.

[1707] Step 7:

[1708] The terminal displays the response from the server to the user.

[1709] The terminal displays the response or fictitious contact information received from the server to the user. The user checks the displayed information and decides the next action. For example, the response "Hello, the weather is sunny today" is displayed.

[1710] (Application example 1)

[1711] Next, a description will be given of Application Example 1 of Form 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."

[1712] In recent years, fraud crimes have been increasing, especially frauds using digital communication. It is difficult to respond to such fraud crimes quickly and effectively using conventional methods. In addition, there are limited means to identify fraudsters and collect evidence. This has led to an increase in the number of victims, which has become a social problem.

[1713] 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.

[1714] In this invention, the server includes means for using artificial intelligence to perform natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for performing sentiment analysis, and means for notifying users using the fictitious contact information. This makes it possible to quickly detect fraudsters, collect evidence, and take appropriate action.

[1715] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1716] "Natural communication" refers to the natural verbal exchanges that humans engage in on a daily basis, and in particular to dialogues that are realized using natural language processing technology.

[1717] "Fictitious contacts" are contact information that does not exist but is created to deceive fraudsters.

[1718] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[1719] "Approach detection methods" refers to technologies and methods used to detect contact or messages from fraudsters.

[1720] "Making fraudsters vulnerable" refers to deliberately maintaining contact with fraudsters in order to monitor their behavior and gather evidence.

[1721] "Evidence of criminal activity" refers to data or information that proves the fraudulent activities of fraudsters.

[1722] "Sentiment analysis" is a technique for analyzing emotions and the intensity of emotions from text data.

[1723] "Means of notifying users" refers to the methods and technologies used to communicate information or warnings detected by the system to users.

[1724] As an embodiment of this invention, we will explain how to install a fraud detection AI assistant system on a smartphone. This system uses artificial intelligence to communicate naturally, detects approaches from fraudsters, and generates fictitious contact information to allow fraudsters to escape.

[1725] System configuration

[1726] The system consists of the following main components:

[1727] 1. Artificial intelligence module: Uses natural language processing technology to communicate naturally with users.

[1728] 2. Fictitious Contact Generation Module: Generates fictitious contacts to deceive fraudsters.

[1729] 3. Fraud detection module: Analyzes message content and behavioral patterns to detect approaches from fraudsters.

[1730] 4. Sentiment Analysis Module: Analyzes the sentiment of the message and assesses its likelihood of fraud.

[1731] 5. Notification module: Notifies users of potential fraud and prompts them to take appropriate action.

[1732] Hardware and software used

[1733] Hardware: Smartphone

[1734] Software: Python, NLTK (Natural Language Toolkit)

[1735] Data processing and calculation

[1736] 1. Receiving a message: The system retrieves the message received by the user.

[1737] 2. Sentiment Analysis: Calculate the sentiment score of the message using NLTK's SentimentIntensityAnalyzer.

[1738] 3. Fraud detection: If the sentiment score exceeds a certain threshold, it is determined to be a possible fraud.

[1739] 4. Fictitious Contact Generation: Randomly generate fictitious contacts if fraud is suspected.

[1740] 5. User Notification: Notify users of potential scams and fictitious contacts.

[1741] Specific examples

[1742] For example, if a user receives a message saying "You've won the lottery! Please send us your bank details," the system can analyze the message and determine that it is likely a scam. It can then generate a fake contact and notify the user, saying "Suspicious activity has been detected. Please contact fake_contact_1@example.com for more information."

[1743] Prompt Sentence Examples

[1744] Examples of prompts to input to a generative AI model include:

[1745] "Analyze the following message for potential fraud: 'You have won a lottery! Please send your bank details.'"

[1746] Using this prompt, a generative AI model is asked to analyze the message and determine whether it is likely to be fraudulent.

[1747] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1748] Step 1:

[1749] The user receives a message. The terminal acquires the message received by the user. This message becomes the input data for the system.

[1750] Step 2:

[1751] The device sends the received message to the sentiment analysis module, which calculates the sentiment score of the message using NLTK's SentimentIntensityAnalyzer. This process outputs the message's sentiment score as positive, negative, or neutral.

[1752] Step 3:

[1753] The server analyzes the emotion scores and determines that fraud is likely if the negative emotion score exceeds a certain threshold. This determination result becomes the input data for the next step.

[1754] Step 4:

[1755] If the server determines that fraud is likely, it invokes a fictitious contact generation module, which randomly generates a fictitious contact and outputs the contact information.

[1756] Step 5:

[1757] The server notifies the user using the generated fictitious contact information. The notification module sends a message to the user informing them of the potential fraud and including the fictitious contact information. The notification is displayed on the user's device.

[1758] Step 6:

[1759] The user is notified and made aware of the potential scam, and can either continue to contact the scammer using the fictitious contact details provided, or take appropriate action.

[1760] Step 7:

[1761] The server monitors interactions between users and fraudsters and collects evidence of criminal activity, which is then stored for later reporting to law enforcement.

[1762] Step 8:

[1763] The server reports the collected evidence to the police. The reporting module organizes the evidence data and sends it to the police in the appropriate format. This process helps identify and apprehend fraudsters.

[1764] Example 2

[1765] Next, a description will be given of Example 2 of Form 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."

[1766] There is a need to quickly and accurately detect approaches from fraudsters and protect users. However, conventional systems have low accuracy in detecting fraudulent messages, putting users at high risk of becoming victims of fraud. In addition, analyzing and evaluating fraudulent messages takes time, making it difficult to respond in real time.

[1767] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving messages sent by fraudsters and storing them in a database, means for analyzing the stored messages using a natural language processing tool, means for extracting message characteristics from the analysis results, means for inputting the extracted characteristics into a generative AI model to evaluate the likelihood of fraud, and means for notifying users of messages that are likely to be fraudulent. This makes it possible to quickly and accurately detect approaches from fraudsters and protect users in real time.

[1768] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1769] "Natural communication" refers to exchanging information using natural language and non-verbal means, as humans do in their daily lives.

[1770] "Fictitious contacts" are non-existent contact methods created to attract fraudsters.

[1771] A "fraud criminal" is someone who attempts to fraudulently obtain money or information by deceiving others.

[1772] "Approach detection methods" are technologies and methods used to identify and detect contact or messages from fraudsters.

[1773] "Methods to let fraudsters go free" refers to the practice of monitoring fraudsters' activities and intentionally allowing them to continue their actions in order to gather evidence.

[1774] "Means of collecting evidence of criminal activity" are methods of recording the actions and messages of fraudsters to gather evidence for later use in legal proceedings.

[1775] A "police reporting method" is a method for providing collected evidence to law enforcement and reporting criminal activity.

[1776] A "database" is a system for storing and managing data in an organized manner.

[1777] "Natural language processing tools" are technologies and software that enable computers to understand and analyze human language.

[1778] A "generative AI model" is a model of artificial intelligence trained to perform a specific task, especially a generative task.

[1779] A "feature extraction method" is a method for identifying and extracting important patterns or attributes from data.

[1780] A "means for assessing the likelihood of fraud" is a technique or method for assessing and scoring the risk of fraud based on the extracted features.

[1781] A "means for notifying a user" is a method for informing a user of a detected risk of fraud.

[1782] This invention is a system for detecting approaches from fraudsters and protecting users. This system operates in cooperation with a server, terminals, and users.

[1783] The server receives messages sent by fraudsters and stores them in a database. Database management systems such as MySQL or PostgreSQL are used for the database. The stored messages are then analyzed using natural language processing tools. Specifically, natural language processing libraries such as "spaCy" and "NLTK" are used to tokenize the messages and tag them by part of speech.

[1784] From the analysis results, the server extracts message features. These features include specific keywords in the message (e.g., "bank," "account," "link," etc.) and the fact that the message was sent late at night. These features are input into a generative AI model. For example, OpenAI's GPT-4 is used as the generative AI model.

[1785] The generative AI model evaluates the likelihood of fraud based on the input features and assigns a score. For example, the following prompt sentence is input into the generative AI model:

[1786] "Please rate the following message for possible fraud. It reads: 'Hello, there's a problem with your bank account. Click this link to check it now.'"

[1787] The generative AI model analyzes the message based on this prompt and scores the likelihood of it being fraudulent. If the score is high, the server sends a notification to the user. Notifications can be sent via email, in-app notifications, or other methods. For example, a push notification could be sent to the user's smartphone, warning them, "You have received a message that is likely to be fraudulent. Please be careful."

[1788] This system makes it possible to quickly and accurately detect approaches from fraudsters and protect users in real time.

[1789] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1790] Step 1: Receiving a message

[1791] The server retrieves messages received by the user. Inputs include messages from the user's email or chat app. The server receives these in real time and passes them on to the next processing step. Specifically, it retrieves messages from the email server or chat server via an API.

[1792] Step 2: Save the message

[1793] The server stores the received message in a database. The input is the message received in step 1. The output is the message stored in the database. Specifically, a database management system such as MySQL or PostgreSQL is used to store the message content and metadata (sender information, receipt time, etc.).

[1794] Step 3: Parse the message

[1795] The server analyzes the stored messages using natural language processing tools. The input is the message stored in step 2. The output is the analysis results. Specifically, it uses "spaCy" and "NLTK" to tokenize the messages and tag them as parts of speech.

[1796] Step 4: Feature extraction

[1797] The server extracts message features from the analysis results. The input is the analysis result obtained in step 3. The output is the extracted features. Specifically, features are extracted such as specific keywords in the message (such as "bank," "account," or "link"), or the fact that the message was sent late at night.

[1798] Step 5: Assess the likelihood of fraud

[1799] The server inputs the extracted features into a generative AI model to assess the likelihood of fraud. The input is the features extracted in step 4. The output is a score indicating the likelihood of fraud. Specifically, the following prompt sentence is input into the generative AI model:

[1800] "Please rate the following message for possible fraud. It reads: 'Hello, there's a problem with your bank account. Click this link to check it now.'"

[1801] The generative AI model analyzes the message based on this prompt and scores it for its likelihood of fraud.

[1802] Step 6: Notify users

[1803] The server notifies the user of messages that are likely to be fraudulent based on the score returned by the generative AI model. The input is the score obtained in step 5. The output is a notification to the user. Specifically, it sends a push notification to the user's smartphone, warning them, "You have received a message that is likely to be fraudulent. Please be careful."

[1804] In this way, a system is realized in which the server, terminal, and user work together to detect approaches from fraudsters and protect users.

[1805] (Application example 2)

[1806] Next, a description will be given of Application Example 2 of Form 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."

[1807] In recent years, fraudulent activities via messages by fraudsters have been increasing, resulting in many cases of users becoming victims. Conventional fraud prevention systems have had problems with low accuracy in detecting fraudulent messages and difficulty in displaying warnings in real time. In addition, there are insufficient means for users to report fraudulent messages, making it difficult to detect and prevent fraud early. To solve these problems, there is a need for a system that can detect fraudulent messages with greater accuracy in real time and display warnings to users.

[1808] 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.

[1809] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for monitoring a user's message application and detecting potentially fraudulent messages in real time, means for displaying a warning for detected potentially fraudulent messages, and means for the user to report fraudulent messages. This enables highly accurate detection of fraudulent messages and display of warnings in real time.

[1810] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1811] "Natural communication" refers to communication using natural language and in a conversational format, as humans do on a daily basis.

[1812] "Fictitious contacts" refer to digital accounts, such as email addresses or social media accounts, that do not exist but are created to lure fraudsters.

[1813] A "fraud criminal" is someone who attempts to fraudulently obtain money or property by deceiving others.

[1814] "Means of approach detection" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to identify potential fraudulent contacts.

[1815] "Fraudster tactics" refers to techniques used to induce fraudsters to continue their activities and to gather evidence.

[1816] "Means of collecting evidence of criminal activity" refers to technology that records the actions and messages of fraudsters and gathers evidence that can later be used in legal proceedings.

[1817] "Law reporting" refers to techniques that provide collected evidence to law enforcement agencies to assist in the arrest and prosecution of fraudsters.

[1818] "Messaging Application" means software that allows a User to send and receive text messages.

[1819] "Real-time detection measures" refers to technology that instantly analyzes potentially fraudulent messages and displays a warning to users.

[1820] "Warning measures" refers to technologies that notify users of potentially fraudulent messages and warn them.

[1821] "Means for reporting fraudulent messages" refers to technology that allows users to report potentially fraudulent messages they receive to the system, contributing to improving the accuracy of AI models.

[1822] As an embodiment of the present invention, a method for installing a fraud prevention assistant system on a smartphone will be described. The system monitors the user's messaging application, detects potentially fraudulent messages in real time, and displays a warning.

[1823] System configuration

[1824] The system consists of the following main components:

[1825] 1. Artificial Intelligence (AI) module: Contains machine learning models for detecting fraudulent messages.

[1826] 2. Message monitoring module: Monitor users' message applications in real time.

[1827] 3. Warning display module: Displays a warning to the user when a potentially fraudulent message is detected.

[1828] 4. Reporting module: Provides an interface for users to report scam messages.

[1829] Hardware and software used

[1830] Hardware: Smartphone

[1831] Software: Python, scikit-learn, joblib

[1832] Data processing and calculation

[1833] 1. Message preprocessing: Lowercase the message and remove special characters.

[1834] 2. Vectorization: Use TfidfVectorizer to convert the message into a numeric vector.

[1835] 3. Predict: Use a pre-trained Logistic Regression model to predict whether the message is fraudulent or not.

[1836] 4. Generate analysis results: Generate analysis results including message sending times and prediction results.

[1837] 5. Display warning: If a potentially fraudulent message is detected, a warning will be displayed to the user.

[1838] Specific examples

[1839] Message received by users: "Your account has been compromised. Click this link now."

[1840] The application detects this and displays a warning: "Warning: A potentially fraudulent message has been detected. Details: {'message': 'Your account has been compromised. Click this link now.', 'send_time': '2023-10-01 12:34:56', 'is_fraud': True}"

[1841] Prompt Sentence Examples

[1842] Generate Python code to develop a fraud prevention assistant app. This app monitors users' messaging apps and detects potentially fraudulent messages in real time. It uses a pre-trained Logistic Regression model and TfidfVectorizer to detect fraudulent messages.

[1843] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1844] Step 1:

[1845] The user receives the message.

[1846] Input: A new message that arrives in the user's messaging application.

[1847] Specific operation: A new message is received by the user's smartphone, and the message monitoring module detects the message.

[1848] Step 2:

[1849] The terminal preprocesses the message.

[1850] Input: The received message.

[1851] Data manipulation: Lowercase the message and remove special characters.

[1852] Output: The preprocessed message.

[1853] What happens: The device converts the message text to lowercase and removes special characters and unnecessary spaces.

[1854] Step 3:

[1855] The terminal vectorizes the preprocessed messages.

[1856] Input: The preprocessed message.

[1857] Data operation: Convert the message into a numeric vector using TfidfVectorizer.

[1858] Output: Vectorized messages.

[1859] Specific operation: The terminal uses TfidfVectorizer to convert the preprocessed message into a numeric vector.

[1860] Step 4:

[1861] The device uses an AI model to predict the vectorized message.

[1862] Input: Vectorized message.

[1863] Data Computation: Use a pre-trained Logistic Regression model to predict whether a message is fraudulent.

[1864] Output: Prediction results of fraudulent messages.

[1865] Specific operation: The device uses a logistic regression model to predict whether the vectorized message is a fraudulent message.

[1866] Step 5:

[1867] The device analyzes the prediction results and displays a warning.

[1868] Input: Predicted results of scam messages.

[1869] Data processing: Generate analysis results, including message sending times and prediction results.

[1870] Output: The warning message that is displayed to the user.

[1871] Specific operation: The device analyzes the prediction results and displays a warning message to the user if there is a possibility of fraud.

[1872] Step 6:

[1873] A user reports a fraudulent message.

[1874] Input: The scam message reported by the user.

[1875] Data processing: Reported messages are recorded and used to improve the accuracy of AI models.

[1876] Output: The updated AI model.

[1877] How it works: A user reports a fraudulent message, and the device records the message and adds it to the training data for the AI ​​model.

[1878] Example 3

[1879] Next, a third embodiment of the third embodiment will be described. 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."

[1880] Fraud crimes are becoming more sophisticated every year, making it difficult to identify fraudsters and collect evidence using conventional methods. It is also necessary to quickly detect approaches from fraudsters and take appropriate action. Furthermore, a system is needed to efficiently report collected evidence to the police.

[1881] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1882] In this invention, the server includes means for using artificial intelligence to conduct natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for storing the collected evidence in a database, means for analyzing messages using a generative AI model, and means for reporting suspected fraud to the police. This makes it possible to quickly detect approaches from fraudsters, efficiently collect and store evidence, and report to the police.

[1883] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1884] "Fictitious contacts" are digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to attract fraudsters.

[1885] "Means for detecting approaches from fraudsters" refers to technology that analyzes the content of messages and behavioral patterns from fraudsters to determine the possibility of fraud.

[1886] "Fraudster tactics" are techniques used to induce fraudsters to continue their activities and gather more evidence.

[1887] "Means of collecting evidence of criminal activity" refers to technology that collects information such as the content of messages from fraudsters, the IP address of the sender, and the time of sending.

[1888] "Means for storing collected evidence in a database" refers to a technique for storing collected evidence information in a database so that it can be referenced later.

[1889] A "generative AI model" is an artificial intelligence model trained to perform natural language processing and data analysis.

[1890] "Means for analyzing messages" refers to technology that uses a generative AI model to analyze the content of received messages and determine whether they are fraudulent.

[1891] "Means of reporting to the police" refers to technology that provides collected evidence information to the police to help identify fraudsters and prevent crime.

[1892] MODE FOR CARRYING OUT THE INVENTION

[1893] The present invention relates to a system for detecting approaches from fraudsters, collecting evidence, and reporting to the police. A specific embodiment of this system will be described below.

[1894] System configuration

[1895] This system consists of three main components: a server, a terminal, and a user. The server receives messages from fraudsters, analyzes them, collects evidence, and reports them to the police. The terminal is a device that users use to access and operate the system. Users use the system to monitor approaches from fraudsters and take necessary action.

[1896] Hardware and software used

[1897] The server is a computer system equipped with a high-performance processor and a large amount of memory. A generative AI model (e.g., GPT-4) and a database (e.g., MySQL) are installed on the server. The generative AI model is used to analyze messages from fraudsters. The database is used to store collected evidence information.

[1898] Data processing and calculation

[1899] When the server receives a message from a fraudster, it first inputs the message into a generative AI model. The generative AI model analyzes the message's content and determines whether it is likely to be fraudulent. The server then collects metadata, such as the message's content, the sender's IP address, and the time of sending, and stores this information in a database. Finally, if the server suspects fraud based on the analysis results, it reports the collected evidence to the police.

[1900] Specific examples

[1901] For example, consider the case where a fraudster sends a phishing email to a user. The server collects the content of the phishing email, the sender's IP address, and the time of sending. The generative AI model analyzes this information and determines that it is suspected to be phishing. The server then reports the analysis results and the collected evidence to the police.

[1902] Prompt Sentence Examples

[1903] An example of a prompt sentence to input to the generative AI model is as follows:

[1904] Please analyze this message for possible fraud. The message reads as follows:

[1905] [Message content]

[1906] In this way, a system can be constructed in which the server can efficiently collect information from fraudsters and report it to the police. The flow of the identification process in the third embodiment will be explained with reference to FIG.

[1907] Step 1:

[1908] The server receives messages from fraudsters. The input is the message sent by the fraudster. The server periodically checks for new emails from the mail server and retrieves emails that are suspected to be fraudulent. The output is the received message data.

[1909] Step 2:

[1910] The server inputs the received message into the generative AI model for analysis. The input is the received message data. The generative AI model (e.g., GPT-4) analyzes the content of the message and determines whether it is fraudulent. The output is the analysis result. Specifically, the following prompt sentence is input into the generative AI model:

[1911] Please analyze this message for possible fraud. The message reads as follows:

[1912] [Message content]

[1913] Step 3:

[1914] The server collects not only the message content but also metadata such as the sender's IP address and the time of sending. The input is the received message data. The server extracts the sender's IP address and the time of sending from the email header information and collects this information as evidence. The output is the collected evidence information.

[1915] Step 4:

[1916] The server stores the collected evidence information in a database. The input is the collected evidence information. The server connects to a database (e.g., MySQL) and executes SQL queries to store the evidence information. The output is the evidence information stored in the database.

[1917] Step 5:

[1918] The server determines whether fraud is suspected based on the analysis results of the generative AI model. The input is the analysis results of the generative AI model. The server checks the analysis results and flags any suspected fraud. The output is the determination result of whether fraud is suspected.

[1919] Step 6:

[1920] If the server determines that fraud is suspected, it reports the collected evidence to the police. The input is the collected evidence and the judgment result. The server creates and sends a request to the police API to send the evidence. The output is the evidence reported to the police.

[1921] In this way, the server can efficiently collect, analyze, and report information from fraudsters to law enforcement.

[1922] (Application example 3)

[1923] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[1924] In modern society, fraud crimes are becoming increasingly sophisticated, requiring victims to respond quickly and effectively when they receive fraudulent messages. However, traditional methods require time and effort to analyze fraudulent messages and gather evidence, and reporting to the police is often delayed. This poses a challenge in identifying fraudsters and preventing crimes.

[1925] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1926] In this invention, the server includes means for using artificial intelligence to carry out natural communication, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters go unpunished, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing fraudulent messages and collecting evidence if there is a possibility of fraud, and means for transmitting the collected evidence to the police's API. This makes it possible to automatically analyze fraudulent messages when they are received, quickly collect evidence, and report it to the police.

[1927] "Artificial intelligence" is the technology that enables computer systems to mimic human intelligence and learn, reason, and self-correct.

[1928] "Natural communication" refers to interfaces and processes that allow humans and machines to interact seamlessly.

[1929] "Fictitious contacts" refer to digital accounts, such as email addresses or social networking service accounts, that do not exist but are created to lure fraudsters.

[1930] A "fraud criminal" is someone who attempts to fraudulently obtain money or property by deceiving others.

[1931] "Means of approach detection" refers to technologies and methods that analyze messages and behavioral patterns from fraudsters to determine the likelihood of fraud.

[1932] "Fraudster tactics" refers to techniques and methods used to induce fraudsters to provide more information and to gather evidence.

[1933] "Evidence of criminal activity" refers to data such as message content, source IP address, and time of transmission that can be used to prove fraudulent activities committed by fraudsters.

[1934] "Means of reporting to the police" refers to the techniques and methods used to provide collected evidence to the police.

[1935] "Means for analyzing fraudulent messages" refers to techniques or methods for analyzing received messages and determining whether they are fraudulent.

[1936] "Means of sending to police API" refers to the technology or method for automatically sending collected evidence to police systems.

[1937] As an embodiment of the present invention, the following system is constructed.

[1938] First, the server has a means for natural communication using artificial intelligence. This artificial intelligence uses a generative AI model to collect information through dialogue with fraudsters. The server also has a means for generating fictitious contact information, which allows it to create email addresses and social networking service accounts to lure fraudsters.

[1939] Next, the server is equipped with a means for detecting approaches from fraudsters. This means determines the possibility of fraud by analyzing the content of messages and behavioral patterns from fraudsters. The server also includes a means for letting fraudsters go, encouraging them to provide more information.

[1940] Furthermore, the server has a means for collecting evidence of criminal activity. This evidence includes the fraudulent message, the source IP address, the time of transmission, etc. The collected evidence is provided to the police through a means for reporting. Specifically, the server has a means for analyzing the fraudulent message and collecting evidence if there is a possibility of fraud. The collected evidence is automatically reported to the police through a means for sending it to the police's API.

[1941] To implement this system, programs are written using programming languages ​​such as Python. The server automatically analyzes fraudulent messages when they are received, quickly collects evidence, and reports them to the police. The hardware used is a smartphone, and the software uses Python. The AI ​​model uses a pre-trained generative AI model.

[1942] For example, if a user receives a message saying, "Your bank account has been frozen. Click here for more information," the system automatically analyzes the message and determines it may be fraudulent. It then collects the message content, the sending IP address, and the time of receipt as evidence and reports it to the police.

[1943] An example of a prompt is as follows:

[1944] "When a fraudulent message is received, please create an application that analyzes the message and, if there is a possibility of fraud, collects evidence and reports it to the police. The message content, source IP address, and time of receipt will be collected as evidence and sent to the police API."

[1945] The above is an embodiment of the present invention.

[1946] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1947] Step 1:

[1948] A user receives a fraudulent message.

[1949] Input: Scam message

[1950] Output: Content of the scam message

[1951] Specific operation: The user's smartphone receives the fraudulent message and sends its contents to the server.

[1952] Step 2:

[1953] The server analyzes the fraudulent message.

[1954] Input: The content of the scam message

[1955] Output: Possibility of fraud judgement result

[1956] How it works: The server uses a generative AI model to analyze the message content and determine whether it is potentially fraudulent.

[1957] Step 3:

[1958] If the server determines that fraud is possible, it will collect evidence.

[1959] Input: Fraudulent message content, source IP address, and time of receipt

[1960] Output: Collected evidence (message content, source IP address, received time)

[1961] Specific operation: The server records the content of the fraudulent message, the sending IP address, and the time of receipt, and collects these as evidence.

[1962] Step 4:

[1963] The server sends the collected evidence to the police API.

[1964] Input: Collected evidence (message content, source IP address, time of receipt)

[1965] Output: Police report result (success or failure)

[1966] What it does: The server sends the collected evidence to the police API and checks whether the report was successful.

[1967] Step 5:

[1968] The server notifies the user of the report results.

[1969] Input: Police report result (success or failure)

[1970] Output: Notification to the user

[1971] Specific operation: The server notifies the user of the results of the police report. If the report is successful, it notifies the user by saying "The report has been submitted to the police." If the report is unsuccessful, it notifies the user by saying "The report failed."

[1972] 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.

[1973] "Example 1"

[1974] One embodiment of the present invention is an artificial intelligence system incorporating an emotion engine. This system recognizes emotions from a user's text messages and voice and adjusts the AI's response based on the emotions. For example, if a user sends a message expressing anger or frustration, the emotion engine recognizes this and the AI ​​generates a response that calms the user. Alternatively, if a user sends a message expressing joy or satisfaction, the emotion engine recognizes this and the AI ​​generates a response that shares the user's joy.

[1975] "Example 2"

[1976] The emotion engine can also track changes in a user's emotions over time, allowing the AI ​​to generate more appropriate responses that reflect the user's changing emotions. For example, if a user initially expresses joy but becomes frustrated over time, the emotion engine will recognize this change and the AI ​​will generate a response that reflects the user's frustration.

[1977] "Example 3"

[1978] Furthermore, the emotion engine has a learning function for recognizing the user's emotions. This learning function allows the emotion engine to learn how the user expresses their emotions and improve the accuracy of emotion recognition over time. For example, if a particular user uses specific words and expressions to express joy, the emotion engine can learn this and become able to more accurately recognize the joy of that user.

[1979] The processing flow of each embodiment will be described below.

[1980] "Example 1"

[1981] Step 1: A text message or voice message from the user is entered into the system.

[1982] Step 2: The emotion engine recognizes the user's emotion from the input.

[1983] Step 3: Based on the recognized emotion, the AI ​​generates a response.

[1984] "Example 2"

[1985] Step 1: A text message or voice message from the user is entered into the system.

[1986] Step 2: The emotion engine recognizes the user's emotion from the input and tracks the change in emotion over time.

[1987] Track together.

[1988] Step 3: Generate an AI response that corresponds to the change in emotion.

[1989] "Example 3"

[1990] Step 1: A text message or voice message from the user is entered into the system.

[1991] Step 2: The emotion engine recognizes the user's emotions from the input and learns how to express those emotions.

[1992] Step 3: Based on the learned emotional expression methods, the emotion engine improves the accuracy of emotion recognition.

[1993] Example 1

[1994] Next, a description will be given of Example 1 of Form 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."

[1995] Conventional AI systems struggled to communicate naturally and were unable to properly recognize user emotions and generate responses. Furthermore, they lacked the means to detect approaches from fraudsters and respond appropriately, making it difficult to prevent fraud and collect evidence. This made it difficult to ensure user safety and led to a risk of increasing fraud victims.

[1996] 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.

[1997] In this invention, the server includes means for conducting natural communication using artificial intelligence, means for generating fictitious contact information, means for detecting approaches from fraudsters, means for letting fraudsters off the hook, means for collecting evidence of criminal activity, means for reporting the collected evidence to the police, means for analyzing messages using natural language processing technology, means for recognizing user emotions using an emotion engine, and means for generating responses based on the recognized emotions. This enables natural communication with users, effectively detecting approaches from fraudsters, and responding appropriately. Furthermore, recognizing user emotions and generating responses enables more human-like interactions and improves user satisfaction.

[1998] "Artificial intelligence" is the technology that enables computer systems to learn, reason, and self-correct by imitating human intelligence.

[1999] "Natural communication" refers to a dialogue between a human and a computer system using natural language, exchanging information in a manner similar to a dialogue between humans.

[2000] "Fictitious contacts" are non-existent contact information and digital accounts, such as email addresses or social media accounts, created to deceive fraudsters.

[2001] A "fraudster" is a person or organization that attempts to fraudulently obtain money or information by deceiving others.

[2002] "Approach detection" refers to technologies and algorithms that analyze contacts and messages from fraudsters to identify their intent.

[2003] "Letting fraudsters run wild" refers to intentionally maintaining contact with fraudsters in order to monitor their behavior and gather evidence.

[2004] "Means of collecting evidence of criminal activity" refers to technology that records the actions and messages of fraudsters and collects data for later use in legal proceedings.

[2005] A "police reporting method" is a method for providing collected evidence to law enforcement agencies and requesting appropriate action.

[2006] "Natural language processing technology" is a technology that enables computers to understand and generate human language, and involves analyzing and generating text.

[2007] An "emotion engine" is a technology that recognizes emotions from a user's text or voice and adjusts responses based on those emotions.

[2008] A "response generator" is a technique or algorithm that creates an appropriate reply based on the user's input and sentiment.

[2009] The present invention is a system for natural communication usi...

Claims

1. A means of natural communication using artificial intelligence; means for the artificial intelligence to generate fictitious contacts; means for performing sentiment analysis of messages received by a user; A means for invoking a means for generating a fictitious contact point when it is determined from the result of the sentiment analysis that the message received by the user is likely to be fraudulent; means for transmitting the fictitious contact information to a user terminal; means for detecting approaches from fraudsters to said fictitious contacts; means of inducing fraudsters to continue their criminal activities; means of collecting evidence of criminal activity; a means of reporting the collected evidence to the police; means for receiving messages sent by fraudsters and storing them in a database; means for analyzing the stored messages using natural language processing tools; means for extracting characteristics of the message from the analysis results; means for inputting the extracted features into a generative AI model to assess the likelihood of fraud; a means for notifying a user of a message that is likely to be fraudulent; means for monitoring a user's messaging application to detect potentially fraudulent messages in real time; A means of displaying warnings about potentially fraudulent messages detected in real time; and A system including:

2. The system of claim 1 , wherein the fictitious contact is a digital account including at least one of an email address and a social networking service account.

3. 2. The system according to claim 1, wherein the means for detecting an approach from a fraudster detects the approach by analyzing at least one of the contents of messages and behavioral patterns from the fraudster.

Images