system

Abstract

To address cases where elderly users, in particular, become victims of telephone-based fraud, we will provide a system that uses voice recognition to detect fraud risks in real time and issue appropriate warnings. [Solution] A system including means for acquiring voice data from a terminal during a call and converting it into text data using speech recognition technology; means for extracting specific keywords or phrases that suggest fraud; means for scoring and determining the likelihood of fraud; and means for issuing a warning to the user based on the determination result to encourage appropriate action.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Cases where users, especially the elderly, are involved in frauds committed via telephone are increasing. In particular, due to being skillfully engaged in conversation, the users themselves are confused in judgment, which is the cause. Such fraud methods are evolving day by day, and it is difficult for users to take countermeasures on their own. Therefore, a system that can detect the risk of fraud in real time and give appropriate warnings is required.

Means for Solving the Problems

[0005] This invention provides a means for analyzing call content by acquiring voice data and converting it into text data using speech recognition technology. It also includes means for extracting specific keywords and phrases that suggest fraud in real time and further scoring and determining the likelihood of fraud. Based on the determination result, the system provides a system that can prevent users from becoming victims of fraud by issuing a warning to the user and encouraging them to consult with family or end the call. This system improves speech recognition accuracy by performing noise reduction and volume equalization processing, and achieves highly accurate determination by using a machine learning model for scoring fraud risk.

[0006] "Audio data" refers to sound information recorded during a telephone call, and can be either real-time or pre-recorded audio information.

[0007] "Speech recognition technology" is a technology that analyzes speech data and converts it into text data, extracting spoken content as text.

[0008] "Text data" refers to data that represents audio data as a string of characters, and includes information such as letters, numbers, and special symbols.

[0009] A "keyword" refers to a significant word or phrase that suggests fraudulent activity, and is a particularly noteworthy element within a context where fraud is suspected.

[0010] "Scoring" is a process of quantifying and evaluating the likelihood of fraud, and is a method used to determine the degree of fraud risk.

[0011] "Determination" refers to the act of determining whether or not there is a possibility of fraud based on the collected information, and ultimately leads to the result that is notified to the user.

[0012] A "warning" is a notification that informs users of the risk of fraud and encourages them to take precautions beforehand.

[0013] "Noise reduction" is a process that removes unwanted noise from audio data to improve the accuracy of speech recognition.

[0014] "Volume equalization processing" is a process that makes audio data volume uniform, making it easier to hear, and is a technology used to improve recognition accuracy.

[0015] A "machine learning model" is a collection of algorithms that learn patterns from data and perform predictions and classifications. This technology is used to determine the likelihood of fraud with high accuracy. [Brief explanation of the drawing]

[0016] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11]It is a sequence diagram showing the processing flow of the data processing system in Embodiment 1. [Figure 12] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] It is a sequence diagram showing the processing flow of the data processing system in Embodiment 2 when the emotion engine is combined. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when the emotion engine is combined.

Mode for Carrying Out the Invention

[0017] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

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

[0019] In the following embodiments, the numbered processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units 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), and the like.

[0020] In the following embodiments, the numbered RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0021] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

[0022] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0024] [First Embodiment]

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

[0026] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0027] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0028] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0029] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0030] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0031] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

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

[0033] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0034] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0035] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0036] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0037] The special fraud prevention system according to the present invention provides a mechanism that analyzes the content of a phone call in real time when a user receives a call and determines the possibility of fraud with high accuracy. The system performs a series of processes including voice data acquisition, voice recognition, text data conversion, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0038] First, the device records the call content as soon as it receives a call and sends this audio data to the server in real time. The server uses speech recognition technology to convert this audio data into text data, applying noise reduction and volume equalization processing to improve the accuracy of the text conversion.

[0039] Next, the server extracts keywords such as "son," "bank," and "payment required" from the converted text data. At this point, it uses natural language processing techniques to analyze the context based on typical scam phrases. If it determines that there is a possibility of fraud, it inputs the fraud risk into a machine learning model and scores the corresponding risk.

[0040] If this score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will put the caller on hold and display a message to the user saying, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0041] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is a bank, and we urgently need money." This system detects the keywords "bank" and "need money," scores the fraud risk, and if it determines the risk is high, sends a warning to the user. This process allows the user to act quickly and without hesitation, preventing them from becoming a victim of fraud.

[0042] As described above, the present invention aims to create an environment in which many users, including the elderly, can use the telephone safely.

[0043] The following describes the processing flow.

[0044] Step 1:

[0045] The device automatically records the conversation when the user receives a call. Upon starting the recording, the audio data is sent to the server in real time.

[0046] Step 2:

[0047] The server initiates speech recognition processing on the received audio data. This process improves audio quality by applying noise reduction and volume equalization. Subsequently, the audio data is converted into text-based data.

[0048] Step 3:

[0049] The server analyzes the generated text data and searches for pre-configured keywords and phrases. For example, it detects terms that suggest fraud, such as "son," "grandson," "bank," and "payment required."

[0050] Step 4:

[0051] The server analyzes the context of detected keywords and scores their likelihood of being a scam. This process uses a machine learning model based on learning from past scam patterns.

[0052] Step 5:

[0053] The server evaluates the scoring results and immediately notifies the device if it determines that the fraud risk is high. A fraud warning is triggered when the score exceeds a predetermined threshold.

[0054] Step 6:

[0055] If the device receives a warning that a call is likely to be a scam, it will put the caller on hold. Furthermore, it will display a warning message to the user stating, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0056] Step 7:

[0057] The user reassesss the situation based on the displayed warning message. If necessary, they can hang up the phone and take action such as consulting with family or calling the police.

[0058] (Example 1)

[0059] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0060] In modern society, abusive behaviors, particularly those conducted via telephone, are on the rise, and many users, especially the elderly, are becoming victims. To address this problem, there is a need for a system that can efficiently analyze call content and quickly and automatically issue warnings about abusive behavior. Furthermore, it is necessary to reduce false positives by improving the accuracy of voice recognition and the accuracy of abusive behavior detection.

[0061] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0062] In this invention, the server includes means for acquiring voice information and converting that information into symbolic data using information recognition technology, means for extracting specific symbol sequences or expressions that suggest special behavior, and means for determining the possibility of special behavior through information processing. This makes it possible to quickly detect the possibility of special behavior from the content of a call and issue an appropriate warning to the user.

[0063] "Voice information" refers to data related to phone calls and other sounds, including recorded sounds and sound information received in real time.

[0064] "Information recognition technology" refers to the technology of converting speech information into symbolic data, and mainly includes speech recognition algorithms and related software.

[0065] "Symbolic data" refers to text or other digital data that has been converted from audio information.

[0066] "Special acts" refer to fraudulent or deceptive acts against users, and encompass all fraudulent activities that may result in harm.

[0067] A "symbol sequence" refers to a specific word or phrase within text or digital data, meaning a sequence of characters that has a specific meaning or context.

[0068] "Information processing" refers to the process of analyzing data and deriving conclusions or judgments from it, usually using algorithms and computational methods.

[0069] "Automated learning structure" refers to models and algorithms that utilize machine learning techniques, and are technologies that have the ability to automatically improve performance and accuracy based on data.

[0070] "False positive" refers to a situation where a special behavior is mistakenly identified as existing, and it signifies an error that occurs due to inaccurate detection.

[0071] This special fraud prevention system has the function of evaluating the possibility of special activity through real-time processing of voice information and warning the user.

[0072] The device records the call content as soon as the user receives a call and transmits the audio information to the server in real time. High-precision microphones and noise-canceling technology are used for recording to obtain accurate audio information.

[0073] The server uses advanced speech recognition technology to convert received audio information into symbolic data. Specifically, it performs noise reduction and volume equalization to improve the quality of the audio data. This process employs specific noise reduction algorithms and speech recognition software that analyzes the audio waveform.

[0074] Furthermore, the server utilizes natural language processing technology to extract symbol sequences related to specific behaviors from the symbolic data. During this process, a generative AI model performs machine learning to evaluate the likelihood of a specific behavior occurring. This automated learning structure is optimized based on historical data, minimizing false positives.

[0075] If the detected risk exceeds a predetermined threshold, the terminal immediately issues a warning to the user and, if necessary, puts the caller on hold. The warning uses visual messages and audible alerts to draw the user's attention.

[0076] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is an urgent matter. We need to receive a money transfer." This system detects the sequence of symbols "money" and "transfer," and if it is deemed high-risk through scoring using a generative AI model, it automatically displays a warning.

[0077] Examples of prompts include, "How do you determine the risk of fraud when a money transfer request comes from a bank?" and "How can you detect potential unusual behavior in a phone call in real time?"

[0078] This system ensures that users can understand the risks and protect themselves from fraud.

[0079] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0080] Step 1:

[0081] The device records phone calls when a user answers one. The input is the audio from the phone call, and the output is the recorded audio data. When recording begins, the device can emit a notification sound stating, "This call is being recorded for security purposes."

[0082] Step 2:

[0083] The terminal transmits recorded audio data to the server in real time. The input is the recorded audio data, and the output is the transfer of the audio data to the server. This transfer is protected by a secure communication protocol, maintaining data confidentiality.

[0084] Step 3:

[0085] The server applies noise reduction and equalizes the volume of the received audio data. This results in consistent audio quality and improved speech recognition accuracy. The input is the transmitted audio data, and the output is the processed, high-quality audio data. This process involves signal processing using a specific algorithm.

[0086] Step 4:

[0087] The server uses speech recognition technology to convert high-quality audio data into symbolic data (text). The input is processed audio data, and the output is text data. This conversion uses state-of-the-art speech recognition software to transform the audio data into symbolic data.

[0088] Step 5:

[0089] The server uses natural language processing techniques to extract symbol sequences that suggest specific actions from text data. The input is the transformed text data, and the output is the extracted symbol sequence (specific keywords or phrases). In this process, the natural language processing algorithm plays a role in understanding the context.

[0090] Step 6:

[0091] The server uses a generative AI model to score the risk of specific actions based on the extracted symbol sequence. The input is the extracted symbol sequence, and the output is the risk score. The generative AI model is trained on historical data, enabling highly accurate scoring.

[0092] Step 7:

[0093] If the risk score exceeds a set threshold, the server sends a warning to the terminal, and the terminal displays a warning message to the user. The input is the risk score, and the output is a warning to the user. Specifically, a message such as "This call may be for unusual activity. Please talk to your family, end the call, and report to the relevant authorities" is displayed.

[0094] (Application Example 1)

[0095] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0096] In recent years, telephone-based fraud has been on the rise, and many users have fallen victim to it. In particular, elderly people and others unfamiliar with information technology have difficulty responding to fraudulent tactics. Therefore, there is a need for a system that can detect potential fraud in real time during a call and quickly warn the user.

[0097] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0098] In this invention, the server includes means for acquiring voice information and converting it into text information using voice recognition technology, means for extracting specific terms and expressions that suggest fraud, means for determining the likelihood of fraud using an evaluation score, and means for monitoring voice information and displaying a warning on the screen if there is a possibility of fraud. This makes it possible to immediately display a warning to the user if there is suspicion of fraud during a call, thereby preventing damage.

[0099] "Voice information" refers to data transmitted through voice communication such as telephone calls.

[0100] "Speech recognition technology" is a technology that analyzes input speech data and converts it into corresponding text data.

[0101] "Textual information" refers to data in text format converted using speech recognition technology.

[0102] "Specific terms and expressions" refer to key keywords and phrases that suggest fraudulent activity.

[0103] "Evaluation score" is an indicator used to quantify and show the likelihood of fraud.

[0104] A "warning display" is a warning message shown on the screen to alert the user.

[0105] To realize this application, the system is built around servers, terminals, and users.

[0106] The server utilizes speech recognition technology to process audio information in real time, converting audio data into text. During this process, it employs noise reduction and volume equalization algorithms to improve conversion accuracy. Furthermore, it uses natural language processing technology to extract specific terms and expressions from the text, and evaluates the likelihood of fraud based on a score. The server also utilizes a learning model to perform this evaluation, ensuring high reliability.

[0107] When the device receives warning information from the server, it displays a warning to the user in real time. For example, if the user receives a call saying, "This is a bank, and we urgently need money," the device will immediately display a warning to the user saying, "This call may be a scam."

[0108] Users can take appropriate action based on this warning. This allows them to quickly address suspected fraud during a phone call and prevent becoming a victim of fraud.

[0109] An example of a prompt might be, "Design a call content analysis algorithm to determine the likelihood of fraud." This prompt can provide assistance in analyzing voice data using a generative AI model.

[0110] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0111] Step 1:

[0112] When the device receives a phone call, it records the call audio in real time and sends it to the server as audio data. The input is the phone call audio, and the output is audio data. Since this audio data is transferred directly to the server, the specific operation involves starting the recording process at the beginning of the call and sending the data to the server over the network.

[0113] Step 2:

[0114] The server applies speech recognition technology to the received audio data to remove noise and equalize the volume. This converts the audio data into text information. The input is audio data, and the output is the refined text information. Specifically, the speech recognition engine analyzes the data and converts it into clear text.

[0115] Step 3:

[0116] The server extracts specific terms and expressions from the converted text information and calculates a score to assess the likelihood of fraud. The input is transliterated text information, and the output is a fraud risk assessment score. Specifically, it uses natural language processing to identify keywords, and a learning model calculates a risk score based on these keywords.

[0117] Step 4:

[0118] If the fraud risk assessment score exceeds a certain threshold, the server sends a warning signal to the terminal. The input is the fraud risk assessment score, and the output is a signal indicating the need for a warning. Specifically, the operation includes determining that the threshold has been exceeded, creating a warning signal, and transmitting it over the network.

[0119] Step 5:

[0120] The terminal receives a warning signal from the server and displays a warning on the user screen. The input is a warning signal, and the output is a visual warning message. Specifically, this includes displaying a warning message as a pop-up on the screen to attract the user's attention.

[0121] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0122] The special fraud prevention system according to the present invention analyzes voice data in real time to determine the possibility of fraud, and also recognizes the user's emotions to further improve accuracy. This system performs a series of processes including voice data acquisition, voice recognition, text data conversion, emotion analysis using an emotion engine, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0123] First, the device records the conversation when the user receives a call and sends the audio data to the server in real time. The server uses speech recognition technology to convert the audio data into text data, and applies noise reduction and volume equalization processing as needed to improve the accuracy of the analysis.

[0124] Subsequently, the server uses an emotion engine to determine the user's emotional state from text and audio data. At this stage, it analyzes the user's tone of voice, speaking speed, and phrasing to obtain emotional information such as whether the user is nervous or suspicious.

[0125] Next, the server detects pre-configured keywords from the generated text data and extracts phrases that may suggest fraud. Then, when scoring the fraud risk, it takes into account the user's emotional state obtained by the emotion engine to make a more refined judgment.

[0126] If the fraud risk score exceeds a predetermined threshold, the device will issue a user-specific warning. Depending on the user's emotional state, for example, if the user is feeling significantly anxious, a particularly emphasized warning message will be displayed to reduce psychological burden while still providing attention.

[0127] For example, if a user receives a call and explains that they need money because their son was in an accident, the system detects keywords such as "money" and "accident," and the emotion engine recognizes that the user is speaking in a clearly confused tone. As a result, the system warns of a high risk of fraud and displays a cautionary message to the user, helping them to handle the situation calmly.

[0128] Thus, by combining an emotion engine, the present invention provides an environment that can more safely protect users, including the elderly and those who are not comfortable with phone calls, from telephone fraud.

[0129] The following describes the processing flow.

[0130] Step 1:

[0131] The device starts recording the conversation when the user receives a call. The recorded audio data is streamed to the server in real time.

[0132] Step 2:

[0133] The server uses speech recognition technology to convert speech data into text data. During the conversion process, noise reduction and volume equalization are applied to improve the quality of the speech.

[0134] Step 3:

[0135] The server uses an emotion engine to analyze the user's emotions from voice and text data. It evaluates factors such as voice tone, speed, and stress level to identify the user's emotional state.

[0136] Step 4:

[0137] The server searches the converted text data for keywords such as "son," "accident," and "need money." These keywords may suggest fraud, so natural language processing techniques are used to detect them.

[0138] Step 5:

[0139] The server scores the fraud risk based on the extracted keywords and sentiment analysis results. It uses a machine learning model to quantify the risk level based on historical data.

[0140] Step 6:

[0141] The server issues a warning if the fraud risk score exceeds a set threshold. In this case, it instructs the terminal to prepare to play a hold tone for the caller.

[0142] Step 7:

[0143] The device displays a warning message to the user. The message is tailored to the user's emotional state and provides instructions such as, "This call may be a scam. Stay calm and check with your family or report it to the police."

[0144] Step 8:

[0145] The user should follow the instructions displayed on the device. If they determine the call is suspicious, they should immediately end the call and take action to ensure their safety.

[0146] Through this process, the system detects the risk of telephone fraud in real time, provides optimal warnings to users, and offers support to protect them from becoming victims of fraud.

[0147] (Example 2)

[0148] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0149] In recent years, telephone-based fraud has been increasing, and vulnerable groups such as the elderly are particularly susceptible to becoming victims. Existing countermeasures lack the speedy and accurate means to identify fraud, and warnings do not take into account the user's emotional state, potentially leading to inappropriate decisions. Therefore, there is a need for a system that can identify fraudulent activity in real time and issue appropriate warnings.

[0150] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0151] In this invention, the server includes means for acquiring voice information and converting it into text information using analysis techniques, means for analyzing specific words and expressions to detect the possibility of fraudulent activity, and means for analyzing the user's emotional state and reflecting it in the risk assessment. This enables real-time identification of fraudulent activity and the issuance of warnings that take the user's emotions into consideration.

[0152] "Audio information" refers to sound information that is acquired as audio data and used for analysis.

[0153] "Analysis technology" refers to the techniques used to analyze acquired data and convert it into a format that humans can understand.

[0154] "Textual information" refers to information in text format obtained as a result of converting audio information using analysis technology.

[0155] "Words and expressions" refer to words, phrases with meaning, or specific expressions used within a text.

[0156] "Fraudulent activity" refers to actions that violate laws or norms, particularly inappropriate behavior including fraud.

[0157] "Detecting possibility" means confirming the likelihood of a particular event or state occurring.

[0158] "Emotional state" refers to the state of a person's feelings or emotions, such as joy or anxiety.

[0159] "Risk assessment" is the process of analyzing and judging potential dangers and losses.

[0160] "Issuing a warning" is the act of notifying others that there is a risk.

[0161] This invention provides a fraud prevention system that utilizes voice, which analyzes voice information from incoming phone calls to a user in real time and issues a warning if there is suspicion of fraudulent activity.

[0162] The device records voice data from the user and transmits the voice data to the server via the internet connection. At this stage, the device collects the voice using its built-in microphone and recording application.

[0163] The server converts the received audio data into text information using speech recognition technology such as Google® Cloud Speech-to-Text. In this step, noise reduction and volume equalization are performed to improve analysis accuracy. Noise reduction is the process of reducing background noise, and volume equalization is the process of adjusting the volume of the audio.

[0164] Subsequently, the server utilizes an emotion analysis engine such as IBM Watson® Tone Analyzer to understand the user's emotional state based on the obtained text information. This identifies whether the user is experiencing emotions such as anxiety or confusion.

[0165] Next, the server uses a pre-configured analysis model to extract keywords that suggest fraud. In this process, keywords such as "money" and "accident" are categorized, and a risk assessment is performed in combination with emotional states.

[0166] If the server detects a potential scam and the risk exceeds a certain threshold, the device will issue a warning to the user. This warning may be presented as an audio or on-screen message, encouraging the user to remain calm.

[0167] For example, if a user receives a call saying, "My son has been in an accident," the system integrates keywords such as "accident" and "money" with sentiment data and warns of a high risk of fraud.

[0168] An example of a prompt when using a generative AI model is an input in the form of, "Analyze the tone of this voice to identify the emotion: 'It's terrible, I've been in an accident.'" This prompt allows the model to perform practical analysis based on the voice data.

[0169] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0170] Step 1:

[0171] The device begins recording the conversation when the user receives a call. This process uses the built-in microphone to capture audio data and saves it as a recording file. The input is live audio, and the output is an audio file. A key feature is that the recording occurs almost in real time.

[0172] Step 2:

[0173] The device sends the recorded audio file to the server via the internet. At this stage, the audio file is encrypted for security purposes. The input is the recorded file, and the output is the audio data securely transferred to the server.

[0174] Step 3:

[0175] The server converts the received audio data into text data using a speech recognition service (e.g., Google Cloud Speech-to-Text). This process simultaneously performs noise reduction and volume equalization, transforming the input audio data into clearer audio data. The final output is noise-reduced text data.

[0176] Step 4:

[0177] The server performs sentiment analysis using text and audio data. This process uses a sentiment analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state. Input is text data and voice tone, and output is information about the user's emotional state. Specifically, it infers emotional characteristics from voice tone and word choice.

[0178] Step 5:

[0179] The server extracts keywords that may suggest fraud from the generated text data and integrates them with sentiment analysis results to assess fraud risk. This process searches for pre-defined keywords such as "money" and "accident," and scores the risk if applicable. The input is text data and sentiment state, and the output is a risk score.

[0180] Step 6:

[0181] If the risk score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will provide a clear warning via voice or text message based on the user's emotional state. The input is the risk score and emotional information, and the output is a warning message. The warning is adjusted to encourage the user to act calmly.

[0182] (Application Example 2)

[0183] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0184] In modern society, telephone-based fraud is on the rise, and individuals are particularly vulnerable to becoming victims. To effectively combat such fraud, a system is needed that can detect potential fraud in real time and provide users with appropriate warnings. However, conventional systems cannot consider emotional states when determining the likelihood of fraud, leading to false positives and missed opportunities.

[0185] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0186] In this invention, the server includes means for analyzing voice data and document data and grasping an individual's emotional state using an emotion engine; means for taking emotional information into account when calculating fraud risk to make a more accurate determination; and means for evaluating the possibility of fraud and issuing a warning to alert the user. This makes it possible to judge the possibility of fraud more precisely and for users to receive warnings quickly.

[0187] "Audio data" refers to acoustic signals based on human speech, expressed in a digital format, and analyzed by a speech recognition system.

[0188] "Acoustic recognition technology" is a technology that automatically converts human speech into text information, and includes a process of extracting meaning from audio data.

[0189] "Text information" refers to character information generated from audio data, and is string data used for analysis and processing.

[0190] An "emotion engine" is a technology that analyzes an individual's emotional state from audio and document data and extracts specific emotional information.

[0191] "Fraud risk" is an indicator that assesses the likelihood of fraud based on information extracted from voice calls and other sources, and serves as a criterion for issuing warnings.

[0192] An "artificial intelligence model" is a computer program equipped with algorithms that learn from large datasets and perform predictions and analyses.

[0193] A "warning" is a notification that alerts users to potential fraud and provides information to help them prevent becoming a victim.

[0194] The system for implementing this invention is installed as an application on the user's smartphone and has the function of detecting the risk of phone fraud in real time. First, when the user receives a call, the terminal records the call audio and immediately sends the audio data to the server. The server uses acoustic recognition technology to convert the audio data into text information. Google Speech-to-Text is one example of a speech recognition engine that could be used here. Pre-processing such as noise reduction and volume equalization is also applied to the audio conversion process to improve the accuracy of the results.

[0195] The server then uses an emotion engine to analyze the user's emotional state from voice data and text information. Suitable emotion engines for this purpose include IBM Watson Tone Analyzer. Based on this emotional information, an artificial intelligence model detects specific terms and phrases that suggest fraud and calculates the fraud risk. In this process, emotional information, such as whether the user is confused, is a crucial factor in assessing fraud risk.

[0196] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. For example, if a user receives a phone call claiming "there is an outstanding bill," and emotional information detects that the user is experiencing high levels of anxiety, the keyword "outstanding" will further increase the risk of fraud. In response, the device will warn the user with a message stating, "This may be a scam. Do not directly verify the details; please verify the information through official channels."

[0197] An example of a prompt message could be: "I want to analyze the call content in real time and determine the risk of fraud involving the keyword 'unpaid.' To do this, please tell me how to convert the audio data into text, identify emotional distress, and issue a warning." This is how you can instruct the generative AI model.

[0198] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0199] Step 1:

[0200] When a user receives a call, the device records the call audio in real time. This recorded data is then input. The recorded audio data is programmed to be sent to the server immediately.

[0201] Step 2:

[0202] The server takes the received audio data as input and converts it into text information using acoustic recognition technology. Specifically, it uses an audio recognition engine such as Google Speech-to-Text to convert the data into text. The output is the converted text information.

[0203] Step 3:

[0204] The server takes text information as input and performs noise reduction and volume equalization. This preprocessing is done to improve the accuracy of the data. The output is clear, preprocessed text data.

[0205] Step 4:

[0206] Using an emotion engine, the server analyzes the user's emotional state from text data. Preprocessed text data is used as input. Specifically, it extracts user emotion information using tools such as IBM Watson Tone Analyzer. The output is the analyzed emotion information.

[0207] Step 5:

[0208] The server uses emotional information and text data as input to detect specific terms and phrases that suggest fraud using an artificial intelligence model. It performs calculations based on the input data to calculate fraud risk. The output is a fraud risk score.

[0209] Step 6:

[0210] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. Specifically, it will display a warning message on the screen to alert the user. The warning message will be generated based on the "generated AI model, prompt text," etc., according to the prompt text.

[0211] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0212] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0213] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0214] [Second Embodiment]

[0215] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0216] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0217] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0218] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0219] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0220] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0221] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0222] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0223] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0224] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0225] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0226] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0227] The special fraud prevention system according to the present invention provides a mechanism that analyzes the content of a phone call in real time when a user receives a call and determines the possibility of fraud with high accuracy. The system performs a series of processes including voice data acquisition, voice recognition, text data conversion, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0228] First, the device records the call content as soon as it receives a call and sends this audio data to the server in real time. The server uses speech recognition technology to convert this audio data into text data, applying noise reduction and volume equalization processing to improve the accuracy of the text conversion.

[0229] Next, the server extracts keywords such as "son," "bank," and "payment required" from the converted text data. At this point, it uses natural language processing techniques to analyze the context based on typical scam phrases. If it determines that there is a possibility of fraud, it inputs the fraud risk into a machine learning model and scores the corresponding risk.

[0230] If this score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will put the caller on hold and display a message to the user saying, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0231] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is a bank, and we urgently need money." This system detects the keywords "bank" and "need money," scores the fraud risk, and if it determines the risk is high, sends a warning to the user. This process allows the user to act quickly and without hesitation, preventing them from becoming a victim of fraud.

[0232] As described above, the present invention aims to create an environment in which many users, including the elderly, can use the telephone safely.

[0233] The following describes the processing flow.

[0234] Step 1:

[0235] The device automatically records the conversation when the user receives a call. Upon starting the recording, the audio data is sent to the server in real time.

[0236] Step 2:

[0237] The server initiates speech recognition processing on the received audio data. This process improves audio quality by applying noise reduction and volume equalization. Subsequently, the audio data is converted into text-based data.

[0238] Step 3:

[0239] The server analyzes the generated text data and searches for pre-configured keywords and phrases. For example, it detects terms that suggest fraud, such as "son," "grandson," "bank," and "payment required."

[0240] Step 4:

[0241] The server analyzes the context of detected keywords and scores their likelihood of being a scam. This process uses a machine learning model based on learning from past scam patterns.

[0242] Step 5:

[0243] The server evaluates the scoring results and immediately notifies the device if it determines that the fraud risk is high. A fraud warning is triggered when the score exceeds a predetermined threshold.

[0244] Step 6:

[0245] If the device receives a warning that a call is likely to be a scam, it will put the caller on hold. Furthermore, it will display a warning message to the user stating, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0246] Step 7:

[0247] The user reassesss the situation based on the displayed warning message. If necessary, they can hang up the phone and take action such as consulting with family or calling the police.

[0248] (Example 1)

[0249] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0250] In modern society, abusive behaviors, particularly those conducted via telephone, are on the rise, and many users, especially the elderly, are becoming victims. To address this problem, there is a need for a system that can efficiently analyze call content and quickly and automatically issue warnings about abusive behavior. Furthermore, it is necessary to reduce false positives by improving the accuracy of voice recognition and the accuracy of abusive behavior detection.

[0251] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0252] In this invention, the server includes means for acquiring voice information and converting that information into symbolic data using information recognition technology, means for extracting specific symbol sequences or expressions that suggest special behavior, and means for determining the possibility of special behavior through information processing. This makes it possible to quickly detect the possibility of special behavior from the content of a call and issue an appropriate warning to the user.

[0253] "Voice information" refers to data related to phone calls and other sounds, including recorded sounds and sound information received in real time.

[0254] "Information recognition technology" refers to the technology of converting speech information into symbolic data, and mainly includes speech recognition algorithms and related software.

[0255] "Symbolic data" refers to text or other digital data that has been converted from audio information.

[0256] "Special acts" refer to fraudulent or deceptive acts against users, and encompass all fraudulent activities that may result in harm.

[0257] A "symbol sequence" refers to a specific word or phrase within text or digital data, meaning a sequence of characters that has a specific meaning or context.

[0258] "Information processing" refers to the process of analyzing data and deriving conclusions or judgments from it, usually using algorithms and computational methods.

[0259] "Automated learning structure" refers to models and algorithms that utilize machine learning techniques, and are technologies that have the ability to automatically improve performance and accuracy based on data.

[0260] "False positive" refers to a situation where a special behavior is mistakenly identified as existing, and it signifies an error that occurs due to inaccurate detection.

[0261] This special fraud prevention system has the function of evaluating the possibility of special activity through real-time processing of voice information and warning the user.

[0262] The device records the call content as soon as the user receives a call and transmits the audio information to the server in real time. High-precision microphones and noise-canceling technology are used for recording to obtain accurate audio information.

[0263] The server uses advanced speech recognition technology to convert received audio information into symbolic data. Specifically, it performs noise reduction and volume equalization to improve the quality of the audio data. This process employs specific noise reduction algorithms and speech recognition software that analyzes the audio waveform.

[0264] Furthermore, the server utilizes natural language processing technology to extract symbol sequences related to specific behaviors from the symbolic data. During this process, a generative AI model performs machine learning to evaluate the likelihood of a specific behavior occurring. This automated learning structure is optimized based on historical data, minimizing false positives.

[0265] If the detected risk exceeds a predetermined threshold, the terminal immediately issues a warning to the user and, if necessary, puts the caller on hold. The warning uses visual messages and audible alerts to draw the user's attention.

[0266] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is an urgent matter. We need to receive a money transfer." This system detects the sequence of symbols "money" and "transfer," and if it is deemed high-risk through scoring using a generative AI model, it automatically displays a warning.

[0267] Examples of prompts include, "How do you determine the risk of fraud when a money transfer request comes from a bank?" and "How can you detect potential unusual behavior in a phone call in real time?"

[0268] This system ensures that users can understand the risks and protect themselves from fraud.

[0269] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0270] Step 1:

[0271] The device records phone calls when a user answers one. The input is the audio from the phone call, and the output is the recorded audio data. When recording begins, the device can emit a notification sound stating, "This call is being recorded for security purposes."

[0272] Step 2:

[0273] The terminal transmits recorded audio data to the server in real time. The input is the recorded audio data, and the output is the transfer of the audio data to the server. This transfer is protected by a secure communication protocol, maintaining data confidentiality.

[0274] Step 3:

[0275] The server applies noise reduction and equalizes the volume of the received audio data. This results in consistent audio quality and improved speech recognition accuracy. The input is the transmitted audio data, and the output is the processed, high-quality audio data. This process involves signal processing using a specific algorithm.

[0276] Step 4:

[0277] The server uses speech recognition technology to convert high-quality audio data into symbolic data (text). The input is processed audio data, and the output is text data. This conversion uses state-of-the-art speech recognition software to transform the audio data into symbolic data.

[0278] Step 5:

[0279] The server uses natural language processing techniques to extract symbol sequences that suggest specific actions from text data. The input is the transformed text data, and the output is the extracted symbol sequence (specific keywords or phrases). In this process, the natural language processing algorithm plays a role in understanding the context.

[0280] Step 6:

[0281] The server uses a generative AI model to score the risk of specific actions based on the extracted symbol sequence. The input is the extracted symbol sequence, and the output is the risk score. The generative AI model is trained on historical data, enabling highly accurate scoring.

[0282] Step 7:

[0283] When the risk score exceeds the set threshold, the server sends a warning to the terminal, and the terminal displays a warning message to the user. The input is the risk score, and the output is a warning to the user. Specifically, a message such as "This call may be a special act. Please consult your family, end the call, and report it to the relevant authorities" is displayed.

[0284] (Application Example 1)

[0285] Next, Application Example 1 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart glasses 214 are referred to as the "terminal".

[0286] In recent years, fraud using telephones has been increasing, and many users have suffered from it. In particular, people who are not familiar with information devices, such as the elderly, have difficulty dealing with fraud tactics. Therefore, there is a need for a system that can detect the possibility of fraud in real time during a call and quickly warn the user.

[0287] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0288] In this invention, the server includes means for acquiring voice information and converting it into character information using voice recognition technology, means for extracting specific terms and expressions suggesting fraud, means for determining the possibility of fraud in terms of evaluation points, and means for monitoring the voice information and displaying a warning on the screen when there is a possibility of fraud. Thereby, when there is a suspicion of fraud during a call, a warning can be immediately displayed to the user, preventing damage.

[0289] "Voice information" is data transmitted by voice communication such as a telephone.

[0290] "Voice recognition technology" is a technology for analyzing input voice data and converting it into corresponding text data.

[0291] "Textual information" refers to data in text format converted using speech recognition technology.

[0292] "Specific terms and expressions" refer to key keywords and phrases that suggest fraudulent activity.

[0293] "Evaluation score" is an indicator used to quantify and show the likelihood of fraud.

[0294] A "warning display" is a warning message shown on the screen to alert the user.

[0295] To realize this application, the system is built around servers, terminals, and users.

[0296] The server utilizes speech recognition technology to process audio information in real time, converting audio data into text. During this process, it employs noise reduction and volume equalization algorithms to improve conversion accuracy. Furthermore, it uses natural language processing technology to extract specific terms and expressions from the text, and evaluates the likelihood of fraud based on a score. The server also utilizes a learning model to perform this evaluation, ensuring high reliability.

[0297] When the device receives warning information from the server, it displays a warning to the user in real time. For example, if the user receives a call saying, "This is a bank, and we urgently need money," the device will immediately display a warning to the user saying, "This call may be a scam."

[0298] Users can take appropriate action based on this warning. This allows them to quickly address suspected fraud during a phone call and prevent becoming a victim of fraud.

[0299] Examples of prompt texts include "Please design a call content analysis algorithm for determining the possibility of fraud." With this prompt, assistance can be obtained for the analysis of voice data by a generative AI model.

[0300] The flow of the specific process in Application Example 1 will be described using FIG. 12.

[0301] Step 1:

[0302] When the terminal receives a phone call, it records the call voice in real time and transmits it to the server as voice data. The input is the phone call voice, and the output is the voice data. Since this voice data is directly transferred to the server, as a specific operation, the recording process is started at the start of the call, and the data is transmitted to the server via the network.

[0303] Step 2:

[0304] The server applies voice recognition technology to the received voice data to perform noise removal and volume normalization. As a result, the voice data is converted into character information. The input is the voice data, and the output is the normalized character information. As specific operations, it includes the work of the voice recognition engine analyzing the data and converting it into clear text. [[ID=2十一]]

[0305] Step 3:

[0306] The server extracts specific terms and expressions from the converted character information and calculates evaluation points for evaluating the possibility of fraud. The input is the normalized character information, and the output is the evaluation points of the fraud risk. As specific operations, it utilizes natural language processing to identify keywords, and based on this, the learning model calculates the risk score.

[0307] Step 4:

[0308] If the fraud risk assessment score exceeds a certain threshold, the server sends a warning signal to the terminal. The input is the fraud risk assessment score, and the output is a signal indicating the need for a warning. Specifically, the operation includes determining that the threshold has been exceeded, creating a warning signal, and transmitting it over the network.

[0309] Step 5:

[0310] The terminal receives a warning signal from the server and displays a warning on the user screen. The input is a warning signal, and the output is a visual warning message. Specifically, this includes displaying a warning message as a pop-up on the screen to attract the user's attention.

[0311] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0312] The special fraud prevention system according to the present invention analyzes voice data in real time to determine the possibility of fraud, and also recognizes the user's emotions to further improve accuracy. This system performs a series of processes including voice data acquisition, voice recognition, text data conversion, emotion analysis using an emotion engine, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0313] First, the device records the conversation when the user receives a call and sends the audio data to the server in real time. The server uses speech recognition technology to convert the audio data into text data, and applies noise reduction and volume equalization processing as needed to improve the accuracy of the analysis.

[0314] Subsequently, the server uses an emotion engine to determine the user's emotional state from text and audio data. At this stage, it analyzes the user's tone of voice, speaking speed, and phrasing to obtain emotional information such as whether the user is nervous or suspicious.

[0315] Next, the server detects pre-configured keywords from the generated text data and extracts phrases that may suggest fraud. Then, when scoring the fraud risk, it takes into account the user's emotional state obtained by the emotion engine to make a more refined judgment.

[0316] If the fraud risk score exceeds a predetermined threshold, the device will issue a user-specific warning. Depending on the user's emotional state, for example, if the user is feeling significantly anxious, a particularly emphasized warning message will be displayed to reduce psychological burden while still providing attention.

[0317] For example, if a user receives a call and explains that they need money because their son was in an accident, the system detects keywords such as "money" and "accident," and the emotion engine recognizes that the user is speaking in a clearly confused tone. As a result, the system warns of a high risk of fraud and displays a cautionary message to the user, helping them to handle the situation calmly.

[0318] Thus, by combining an emotion engine, the present invention provides an environment that can more safely protect users, including the elderly and those who are not comfortable with phone calls, from telephone fraud.

[0319] The following describes the processing flow.

[0320] Step 1:

[0321] The device starts recording the conversation when the user receives a call. The recorded audio data is streamed to the server in real time.

[0322] Step 2:

[0323] The server uses speech recognition technology to convert speech data into text data. During the conversion process, noise reduction and volume equalization are applied to improve the quality of the speech.

[0324] Step 3:

[0325] The server uses an emotion engine to analyze the user's emotions from voice and text data. It evaluates factors such as voice tone, speed, and stress level to identify the user's emotional state.

[0326] Step 4:

[0327] The server searches the converted text data for keywords such as "son," "accident," and "need money." These keywords may suggest fraud, so natural language processing techniques are used to detect them.

[0328] Step 5:

[0329] The server scores the fraud risk based on the extracted keywords and sentiment analysis results. It uses a machine learning model to quantify the risk level based on historical data.

[0330] Step 6:

[0331] The server issues a warning if the fraud risk score exceeds a set threshold. In this case, it instructs the terminal to prepare to play a hold tone for the caller.

[0332] Step 7:

[0333] The device displays a warning message to the user. The message is tailored to the user's emotional state and provides instructions such as, "This call may be a scam. Stay calm and check with your family or report it to the police."

[0334] Step 8:

[0335] The user should follow the instructions displayed on the device. If they determine the call is suspicious, they should immediately end the call and take action to ensure their safety.

[0336] Through this process, the system detects the risk of telephone fraud in real time, provides optimal warnings to users, and offers support to protect them from becoming victims of fraud.

[0337] (Example 2)

[0338] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0339] In recent years, telephone-based fraud has been increasing, and vulnerable groups such as the elderly are particularly susceptible to becoming victims. Existing countermeasures lack the speedy and accurate means to identify fraud, and warnings do not take into account the user's emotional state, potentially leading to inappropriate decisions. Therefore, there is a need for a system that can identify fraudulent activity in real time and issue appropriate warnings.

[0340] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0341] In this invention, the server includes means for acquiring voice information and converting it into text information using analysis techniques, means for analyzing specific words and expressions to detect the possibility of fraudulent activity, and means for analyzing the user's emotional state and reflecting it in the risk assessment. This enables real-time identification of fraudulent activity and the issuance of warnings that take the user's emotions into consideration.

[0342] "Audio information" refers to sound information that is acquired as audio data and used for analysis.

[0343] "Analysis technology" refers to the techniques used to analyze acquired data and convert it into a format that humans can understand.

[0344] "Textual information" refers to information in text format obtained as a result of converting audio information using analysis technology.

[0345] "Words and expressions" refer to words, phrases with meaning, or specific expressions used within a text.

[0346] "Fraudulent activity" refers to actions that violate laws or norms, particularly inappropriate behavior including fraud.

[0347] "Detecting possibility" means confirming the likelihood of a particular event or state occurring.

[0348] "Emotional state" refers to the state of a person's feelings or emotions, such as joy or anxiety.

[0349] "Risk assessment" is the process of analyzing and judging potential dangers and losses.

[0350] "Issuing a warning" is the act of notifying others that there is a risk.

[0351] This invention provides a fraud prevention system that utilizes voice, which analyzes voice information from incoming phone calls to a user in real time and issues a warning if there is suspicion of fraudulent activity.

[0352] The device records voice data from the user and transmits the voice data to the server via the internet connection. At this stage, the device collects the voice using its built-in microphone and recording application.

[0353] The server uses speech recognition technology such as Google Cloud Speech-to-Text to convert the received audio data into text information. In this step, noise reduction and volume equalization are performed to improve the accuracy of the analysis. Noise reduction is the process of reducing background noise, and volume equalization is the process of adjusting the volume of the audio.

[0354] The server then utilizes an emotion analysis engine, such as IBM Watson Tone Analyzer, to understand the user's emotional state based on the obtained text information. This helps identify whether the user is experiencing emotions such as anxiety or confusion.

[0355] Next, the server uses a pre-configured analysis model to extract keywords that suggest fraud. In this process, keywords such as "money" and "accident" are categorized, and a risk assessment is performed in combination with emotional states.

[0356] If the server detects a potential scam and the risk exceeds a certain threshold, the device will issue a warning to the user. This warning may be presented as an audio or on-screen message, encouraging the user to remain calm.

[0357] For example, if a user receives a call saying, "My son has been in an accident," the system integrates keywords such as "accident" and "money" with sentiment data and warns of a high risk of fraud.

[0358] An example of a prompt when using a generative AI model is an input in the form of, "Analyze the tone of this voice to identify the emotion: 'It's terrible, I've been in an accident.'" This prompt allows the model to perform practical analysis based on the voice data.

[0359] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0360] Step 1:

[0361] The device begins recording the conversation when the user receives a call. This process uses the built-in microphone to capture audio data and saves it as a recording file. The input is live audio, and the output is an audio file. A key feature is that the recording occurs almost in real time.

[0362] Step 2:

[0363] The device sends the recorded audio file to the server via the internet. At this stage, the audio file is encrypted for security purposes. The input is the recorded file, and the output is the audio data securely transferred to the server.

[0364] Step 3:

[0365] The server converts the received audio data into text data using a speech recognition service (e.g., Google Cloud Speech-to-Text). This process simultaneously performs noise reduction and volume equalization, transforming the input audio data into clearer audio data. The final output is noise-reduced text data.

[0366] Step 4:

[0367] The server performs sentiment analysis using text and audio data. This process uses a sentiment analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state. Input is text data and voice tone, and output is information about the user's emotional state. Specifically, it infers emotional characteristics from voice tone and word choice.

[0368] Step 5:

[0369] The server extracts keywords that may suggest fraud from the generated text data and integrates them with sentiment analysis results to assess fraud risk. This process searches for pre-defined keywords such as "money" and "accident," and scores the risk if applicable. The input is text data and sentiment state, and the output is a risk score.

[0370] Step 6:

[0371] If the risk score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will provide a clear warning via voice or text message based on the user's emotional state. The input is the risk score and emotional information, and the output is a warning message. The warning is adjusted to encourage the user to act calmly.

[0372] (Application Example 2)

[0373] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0374] In modern society, telephone-based fraud is on the rise, and individuals are particularly vulnerable to becoming victims. To effectively combat such fraud, a system is needed that can detect potential fraud in real time and provide users with appropriate warnings. However, conventional systems cannot consider emotional states when determining the likelihood of fraud, leading to false positives and missed opportunities.

[0375] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0376] In this invention, the server includes means for analyzing voice data and document data and grasping an individual's emotional state using an emotion engine; means for taking emotional information into account when calculating fraud risk to make a more accurate determination; and means for evaluating the possibility of fraud and issuing a warning to alert the user. This makes it possible to judge the possibility of fraud more precisely and for users to receive warnings quickly.

[0377] "Audio data" refers to acoustic signals based on human speech, expressed in a digital format, and analyzed by a speech recognition system.

[0378] "Acoustic recognition technology" is a technology that automatically converts human speech into text information, and includes a process of extracting meaning from audio data.

[0379] "Text information" refers to character information generated from audio data, and is string data used for analysis and processing.

[0380] An "emotion engine" is a technology that analyzes an individual's emotional state from audio and document data and extracts specific emotional information.

[0381] "Fraud risk" is an indicator that assesses the likelihood of fraud based on information extracted from voice calls and other sources, and serves as a criterion for issuing warnings.

[0382] An "artificial intelligence model" is a computer program equipped with algorithms that learn from large datasets and perform predictions and analyses.

[0383] A "warning" is a notification that alerts users to potential fraud and provides information to help them prevent becoming a victim.

[0384] The system for implementing this invention is installed as an application on the user's smartphone and has the function of detecting the risk of phone fraud in real time. First, when the user receives a call, the terminal records the call audio and immediately sends the audio data to the server. The server uses acoustic recognition technology to convert the audio data into text information. Google Speech-to-Text is one example of a speech recognition engine that could be used here. Pre-processing such as noise reduction and volume equalization is also applied to the audio conversion process to improve the accuracy of the results.

[0385] The server then uses an emotion engine to analyze the user's emotional state from voice data and text information. Suitable emotion engines for this purpose include IBM Watson Tone Analyzer. Based on this emotional information, an artificial intelligence model detects specific terms and phrases that suggest fraud and calculates the fraud risk. In this process, emotional information, such as whether the user is confused, is a crucial factor in assessing fraud risk.

[0386] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. For example, if a user receives a phone call claiming "there is an outstanding bill," and emotional information detects that the user is experiencing high levels of anxiety, the keyword "outstanding" will further increase the risk of fraud. In response, the device will warn the user with a message stating, "This may be a scam. Do not directly verify the details; please verify the information through official channels."

[0387] An example of a prompt message could be: "I want to analyze the call content in real time and determine the risk of fraud involving the keyword 'unpaid.' To do this, please tell me how to convert the audio data into text, identify emotional distress, and issue a warning." This is how you can instruct the generative AI model.

[0388] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0389] Step 1:

[0390] When a user receives a call, the device records the call audio in real time. This recorded data is then input. The recorded audio data is programmed to be sent to the server immediately.

[0391] Step 2:

[0392] The server takes the received audio data as input and converts it into text information using acoustic recognition technology. Specifically, it uses an audio recognition engine such as Google Speech-to-Text to convert the data into text. The output is the converted text information.

[0393] Step 3:

[0394] The server takes text information as input and performs noise reduction and volume equalization. This preprocessing is done to improve the accuracy of the data. The output is clear, preprocessed text data.

[0395] Step 4:

[0396] Using an emotion engine, the server analyzes the user's emotional state from text data. Preprocessed text data is used as input. Specifically, it extracts user emotion information using tools such as IBM Watson Tone Analyzer. The output is the analyzed emotion information.

[0397] Step 5:

[0398] The server uses emotional information and text data as input to detect specific terms and phrases that suggest fraud using an artificial intelligence model. It performs calculations based on the input data to calculate fraud risk. The output is a fraud risk score.

[0399] Step 6:

[0400] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. Specifically, it will display a warning message on the screen to alert the user. The warning message will be generated based on the "generated AI model, prompt text," etc., according to the prompt text.

[0401] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0402] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0403] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0404] [Third Embodiment]

[0405] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0406] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0407] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0408] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0409] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0410] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0411] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0412] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0413] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0414] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0415] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0416] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0417] The special fraud prevention system according to the present invention provides a mechanism that analyzes the content of a phone call in real time when a user receives a call and determines the possibility of fraud with high accuracy. The system performs a series of processes including voice data acquisition, voice recognition, text data conversion, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0418] First, the device records the call content as soon as it receives a call and sends this audio data to the server in real time. The server uses speech recognition technology to convert this audio data into text data, applying noise reduction and volume equalization processing to improve the accuracy of the text conversion.

[0419] Next, the server extracts keywords such as "son," "bank," and "payment required" from the converted text data. At this point, it uses natural language processing techniques to analyze the context based on typical scam phrases. If it determines that there is a possibility of fraud, it inputs the fraud risk into a machine learning model and scores the corresponding risk.

[0420] If this score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will put the caller on hold and display a message to the user saying, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0421] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is a bank, and we urgently need money." This system detects the keywords "bank" and "need money," scores the fraud risk, and if it determines the risk is high, sends a warning to the user. This process allows the user to act quickly and without hesitation, preventing them from becoming a victim of fraud.

[0422] As described above, the present invention aims to create an environment in which many users, including the elderly, can use the telephone safely.

[0423] The following describes the processing flow.

[0424] Step 1:

[0425] The device automatically records the conversation when the user receives a call. Upon starting the recording, the audio data is sent to the server in real time.

[0426] Step 2:

[0427] The server initiates speech recognition processing on the received audio data. This process improves audio quality by applying noise reduction and volume equalization. Subsequently, the audio data is converted into text-based data.

[0428] Step 3:

[0429] The server analyzes the generated text data and searches for pre-configured keywords and phrases. For example, it detects terms that suggest fraud, such as "son," "grandson," "bank," and "payment required."

[0430] Step 4:

[0431] The server analyzes the context of detected keywords and scores their likelihood of being a scam. This process uses a machine learning model based on learning from past scam patterns.

[0432] Step 5:

[0433] The server evaluates the scoring results and immediately notifies the device if it determines that the fraud risk is high. A fraud warning is triggered when the score exceeds a predetermined threshold.

[0434] Step 6:

[0435] If the device receives a warning that a call is likely to be a scam, it will put the caller on hold. Furthermore, it will display a warning message to the user stating, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0436] Step 7:

[0437] The user reassesss the situation based on the displayed warning message. If necessary, they can hang up the phone and take action such as consulting with family or calling the police.

[0438] (Example 1)

[0439] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0440] In modern society, abusive behaviors, particularly those conducted via telephone, are on the rise, and many users, especially the elderly, are becoming victims. To address this problem, there is a need for a system that can efficiently analyze call content and quickly and automatically issue warnings about abusive behavior. Furthermore, it is necessary to reduce false positives by improving the accuracy of voice recognition and the accuracy of abusive behavior detection.

[0441] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0442] In this invention, the server includes means for acquiring voice information and converting that information into symbolic data using information recognition technology, means for extracting specific symbol sequences or expressions that suggest special behavior, and means for determining the possibility of special behavior through information processing. This makes it possible to quickly detect the possibility of special behavior from the content of a call and issue an appropriate warning to the user.

[0443] "Voice information" refers to data related to phone calls and other sounds, including recorded sounds and sound information received in real time.

[0444] "Information recognition technology" refers to the technology of converting speech information into symbolic data, and mainly includes speech recognition algorithms and related software.

[0445] "Symbolic data" refers to text or other digital data that has been converted from audio information.

[0446] "Special acts" refer to fraudulent or deceptive acts against users, and encompass all fraudulent activities that may result in harm.

[0447] A "symbol sequence" refers to a specific word or phrase within text or digital data, meaning a sequence of characters that has a specific meaning or context.

[0448] "Information processing" refers to the process of analyzing data and deriving conclusions or judgments from it, usually using algorithms and computational methods.

[0449] "Automated learning structure" refers to models and algorithms that utilize machine learning techniques, and are technologies that have the ability to automatically improve performance and accuracy based on data.

[0450] "False positive" refers to a situation where a special behavior is mistakenly identified as existing, and it signifies an error that occurs due to inaccurate detection.

[0451] This special fraud prevention system has the function of evaluating the possibility of special activity through real-time processing of voice information and warning the user.

[0452] The device records the call content as soon as the user receives a call and transmits the audio information to the server in real time. High-precision microphones and noise-canceling technology are used for recording to obtain accurate audio information.

[0453] The server uses advanced speech recognition technology to convert received audio information into symbolic data. Specifically, it performs noise reduction and volume equalization to improve the quality of the audio data. This process employs specific noise reduction algorithms and speech recognition software that analyzes the audio waveform.

[0454] Furthermore, the server utilizes natural language processing technology to extract symbol sequences related to specific behaviors from the symbolic data. During this process, a generative AI model performs machine learning to evaluate the likelihood of a specific behavior occurring. This automated learning structure is optimized based on historical data, minimizing false positives.

[0455] If the detected risk exceeds a predetermined threshold, the terminal immediately issues a warning to the user and, if necessary, puts the caller on hold. The warning uses visual messages and audible alerts to draw the user's attention.

[0456] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is an urgent matter. We need to receive a money transfer." This system detects the sequence of symbols "money" and "transfer," and if it is deemed high-risk through scoring using a generative AI model, it automatically displays a warning.

[0457] Examples of prompts include, "How do you determine the risk of fraud when a money transfer request comes from a bank?" and "How can you detect potential unusual behavior in a phone call in real time?"

[0458] This system ensures that users can understand the risks and protect themselves from fraud.

[0459] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0460] Step 1:

[0461] The device records phone calls when a user answers one. The input is the audio from the phone call, and the output is the recorded audio data. When recording begins, the device can emit a notification sound stating, "This call is being recorded for security purposes."

[0462] Step 2:

[0463] The terminal transmits recorded audio data to the server in real time. The input is the recorded audio data, and the output is the transfer of the audio data to the server. This transfer is protected by a secure communication protocol, maintaining data confidentiality.

[0464] Step 3:

[0465] The server applies noise reduction and equalizes the volume of the received audio data. This results in consistent audio quality and improved speech recognition accuracy. The input is the transmitted audio data, and the output is the processed, high-quality audio data. This process involves signal processing using a specific algorithm.

[0466] Step 4:

[0467] The server uses speech recognition technology to convert high-quality audio data into symbolic data (text). The input is processed audio data, and the output is text data. This conversion uses state-of-the-art speech recognition software to transform the audio data into symbolic data.

[0468] Step 5:

[0469] The server uses natural language processing techniques to extract symbol sequences that suggest specific actions from text data. The input is the transformed text data, and the output is the extracted symbol sequence (specific keywords or phrases). In this process, the natural language processing algorithm plays a role in understanding the context.

[0470] Step 6:

[0471] The server uses a generative AI model to score the risk of specific actions based on the extracted symbol sequence. The input is the extracted symbol sequence, and the output is the risk score. The generative AI model is trained on historical data, enabling highly accurate scoring.

[0472] Step 7:

[0473] If the risk score exceeds a set threshold, the server sends a warning to the terminal, and the terminal displays a warning message to the user. The input is the risk score, and the output is a warning to the user. Specifically, a message such as "This call may be for unusual activity. Please talk to your family, end the call, and report to the relevant authorities" is displayed.

[0474] (Application Example 1)

[0475] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0476] In recent years, telephone-based fraud has been on the rise, and many users have fallen victim to it. In particular, elderly people and others unfamiliar with information technology have difficulty responding to fraudulent tactics. Therefore, there is a need for a system that can detect potential fraud in real time during a call and quickly warn the user.

[0477] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0478] In this invention, the server includes means for acquiring voice information and converting it into text information using voice recognition technology, means for extracting specific terms and expressions that suggest fraud, means for determining the likelihood of fraud using an evaluation score, and means for monitoring voice information and displaying a warning on the screen if there is a possibility of fraud. This makes it possible to immediately display a warning to the user if there is suspicion of fraud during a call, thereby preventing damage.

[0479] "Voice information" refers to data transmitted through voice communication such as telephone calls.

[0480] "Speech recognition technology" is a technology that analyzes input speech data and converts it into corresponding text data.

[0481] "Textual information" refers to data in text format converted using speech recognition technology.

[0482] "Specific terms and expressions" refer to key keywords and phrases that suggest fraudulent activity.

[0483] "Evaluation score" is an indicator used to quantify and show the likelihood of fraud.

[0484] A "warning display" is a warning message shown on the screen to alert the user.

[0485] To realize this application, the system is built around servers, terminals, and users.

[0486] The server utilizes speech recognition technology to process audio information in real time, converting audio data into text. During this process, it employs noise reduction and volume equalization algorithms to improve conversion accuracy. Furthermore, it uses natural language processing technology to extract specific terms and expressions from the text, and evaluates the likelihood of fraud based on a score. The server also utilizes a learning model to perform this evaluation, ensuring high reliability.

[0487] When the device receives warning information from the server, it displays a warning to the user in real time. For example, if the user receives a call saying, "This is a bank, and we urgently need money," the device will immediately display a warning to the user saying, "This call may be a scam."

[0488] Users can take appropriate action based on this warning. This allows them to quickly address suspected fraud during a phone call and prevent becoming a victim of fraud.

[0489] An example of a prompt might be, "Design a call content analysis algorithm to determine the likelihood of fraud." This prompt can provide assistance in analyzing voice data using a generative AI model.

[0490] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0491] Step 1:

[0492] When the device receives a phone call, it records the call audio in real time and sends it to the server as audio data. The input is the phone call audio, and the output is audio data. Since this audio data is transferred directly to the server, the specific operation involves starting the recording process at the beginning of the call and sending the data to the server over the network.

[0493] Step 2:

[0494] The server applies speech recognition technology to the received audio data to remove noise and equalize the volume. This converts the audio data into text information. The input is audio data, and the output is the refined text information. Specifically, the speech recognition engine analyzes the data and converts it into clear text.

[0495] Step 3:

[0496] The server extracts specific terms and expressions from the converted text information and calculates a score to assess the likelihood of fraud. The input is transliterated text information, and the output is a fraud risk assessment score. Specifically, it uses natural language processing to identify keywords, and a learning model calculates a risk score based on these keywords.

[0497] Step 4:

[0498] If the fraud risk assessment score exceeds a certain threshold, the server sends a warning signal to the terminal. The input is the fraud risk assessment score, and the output is a signal indicating the need for a warning. Specifically, the operation includes determining that the threshold has been exceeded, creating a warning signal, and transmitting it over the network.

[0499] Step 5:

[0500] The terminal receives a warning signal from the server and displays a warning on the user screen. The input is a warning signal, and the output is a visual warning message. Specifically, this includes displaying a warning message as a pop-up on the screen to attract the user's attention.

[0501] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0502] The special fraud prevention system according to the present invention analyzes voice data in real time to determine the possibility of fraud, and also recognizes the user's emotions to further improve accuracy. This system performs a series of processes including voice data acquisition, voice recognition, text data conversion, emotion analysis using an emotion engine, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0503] First, the device records the conversation when the user receives a call and sends the audio data to the server in real time. The server uses speech recognition technology to convert the audio data into text data, and applies noise reduction and volume equalization processing as needed to improve the accuracy of the analysis.

[0504] Subsequently, the server uses an emotion engine to determine the user's emotional state from text and audio data. At this stage, it analyzes the user's tone of voice, speaking speed, and phrasing to obtain emotional information such as whether the user is nervous or suspicious.

[0505] Next, the server detects pre-configured keywords from the generated text data and extracts phrases that may suggest fraud. Then, when scoring the fraud risk, it takes into account the user's emotional state obtained by the emotion engine to make a more refined judgment.

[0506] If the fraud risk score exceeds a predetermined threshold, the device will issue a user-specific warning. Depending on the user's emotional state, for example, if the user is feeling significantly anxious, a particularly emphasized warning message will be displayed to reduce psychological burden while still providing attention.

[0507] For example, if a user receives a call and explains that they need money because their son was in an accident, the system detects keywords such as "money" and "accident," and the emotion engine recognizes that the user is speaking in a clearly confused tone. As a result, the system warns of a high risk of fraud and displays a cautionary message to the user, helping them to handle the situation calmly.

[0508] Thus, by combining an emotion engine, the present invention provides an environment that can more safely protect users, including the elderly and those who are not comfortable with phone calls, from telephone fraud.

[0509] The following describes the processing flow.

[0510] Step 1:

[0511] The device starts recording the conversation when the user receives a call. The recorded audio data is streamed to the server in real time.

[0512] Step 2:

[0513] The server uses speech recognition technology to convert speech data into text data. During the conversion process, noise reduction and volume equalization are applied to improve the quality of the speech.

[0514] Step 3:

[0515] The server uses an emotion engine to analyze the user's emotions from voice and text data. It evaluates factors such as voice tone, speed, and stress level to identify the user's emotional state.

[0516] Step 4:

[0517] The server searches the converted text data for keywords such as "son," "accident," and "need money." These keywords may suggest fraud, so natural language processing techniques are used to detect them.

[0518] Step 5:

[0519] The server scores the fraud risk based on the extracted keywords and sentiment analysis results. It uses a machine learning model to quantify the risk level based on historical data.

[0520] Step 6:

[0521] The server issues a warning if the fraud risk score exceeds a set threshold. In this case, it instructs the terminal to prepare to play a hold tone for the caller.

[0522] Step 7:

[0523] The device displays a warning message to the user. The message is tailored to the user's emotional state and provides instructions such as, "This call may be a scam. Stay calm and check with your family or report it to the police."

[0524] Step 8:

[0525] The user should follow the instructions displayed on the device. If they determine the call is suspicious, they should immediately end the call and take action to ensure their safety.

[0526] Through this process, the system detects the risk of telephone fraud in real time, provides optimal warnings to users, and offers support to protect them from becoming victims of fraud.

[0527] (Example 2)

[0528] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0529] In recent years, telephone-based fraud has been increasing, and vulnerable groups such as the elderly are particularly susceptible to becoming victims. Existing countermeasures lack the speedy and accurate means to identify fraud, and warnings do not take into account the user's emotional state, potentially leading to inappropriate decisions. Therefore, there is a need for a system that can identify fraudulent activity in real time and issue appropriate warnings.

[0530] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0531] In this invention, the server includes means for acquiring voice information and converting it into text information using analysis techniques, means for analyzing specific words and expressions to detect the possibility of fraudulent activity, and means for analyzing the user's emotional state and reflecting it in the risk assessment. This enables real-time identification of fraudulent activity and the issuance of warnings that take the user's emotions into consideration.

[0532] "Audio information" refers to sound information that is acquired as audio data and used for analysis.

[0533] "Analysis technology" refers to the techniques used to analyze acquired data and convert it into a format that humans can understand.

[0534] "Textual information" refers to information in text format obtained as a result of converting audio information using analysis technology.

[0535] "Words and expressions" refer to words, phrases with meaning, or specific expressions used within a text.

[0536] "Fraudulent activity" refers to actions that violate laws or norms, particularly inappropriate behavior including fraud.

[0537] "Detecting possibility" means confirming the likelihood of a particular event or state occurring.

[0538] "Emotional state" refers to the state of a person's feelings or emotions, such as joy or anxiety.

[0539] "Risk assessment" is the process of analyzing and judging potential dangers and losses.

[0540] "Issuing a warning" is the act of notifying others that there is a risk.

[0541] This invention provides a fraud prevention system that utilizes voice, which analyzes voice information from incoming phone calls to a user in real time and issues a warning if there is suspicion of fraudulent activity.

[0542] The device records voice data from the user and transmits the voice data to the server via the internet connection. At this stage, the device collects the voice using its built-in microphone and recording application.

[0543] The server uses speech recognition technology such as Google Cloud Speech-to-Text to convert the received audio data into text information. In this step, noise reduction and volume equalization are performed to improve the accuracy of the analysis. Noise reduction is the process of reducing background noise, and volume equalization is the process of adjusting the volume of the audio.

[0544] The server then utilizes an emotion analysis engine, such as IBM Watson Tone Analyzer, to understand the user's emotional state based on the obtained text information. This helps identify whether the user is experiencing emotions such as anxiety or confusion.

[0545] Next, the server uses a pre-configured analysis model to extract keywords that suggest fraud. In this process, keywords such as "money" and "accident" are categorized, and a risk assessment is performed in combination with emotional states.

[0546] If the server detects a potential scam and the risk exceeds a certain threshold, the device will issue a warning to the user. This warning may be presented as an audio or on-screen message, encouraging the user to remain calm.

[0547] For example, if a user receives a call saying, "My son has been in an accident," the system integrates keywords such as "accident" and "money" with sentiment data and warns of a high risk of fraud.

[0548] An example of a prompt when using a generative AI model is an input in the form of, "Analyze the tone of this voice to identify the emotion: 'It's terrible, I've been in an accident.'" This prompt allows the model to perform practical analysis based on the voice data.

[0549] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0550] Step 1:

[0551] The device begins recording the conversation when the user receives a call. This process uses the built-in microphone to capture audio data and saves it as a recording file. The input is live audio, and the output is an audio file. A key feature is that the recording occurs almost in real time.

[0552] Step 2:

[0553] The device sends the recorded audio file to the server via the internet. At this stage, the audio file is encrypted for security purposes. The input is the recorded file, and the output is the audio data securely transferred to the server.

[0554] Step 3:

[0555] The server converts the received audio data into text data using a speech recognition service (e.g., Google Cloud Speech-to-Text). This process simultaneously performs noise reduction and volume equalization, transforming the input audio data into clearer audio data. The final output is noise-reduced text data.

[0556] Step 4:

[0557] The server performs sentiment analysis using text and audio data. This process uses a sentiment analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state. Input is text data and voice tone, and output is information about the user's emotional state. Specifically, it infers emotional characteristics from voice tone and word choice.

[0558] Step 5:

[0559] The server extracts keywords that may suggest fraud from the generated text data and integrates them with sentiment analysis results to assess fraud risk. This process searches for pre-defined keywords such as "money" and "accident," and scores the risk if applicable. The input is text data and sentiment state, and the output is a risk score.

[0560] Step 6:

[0561] If the risk score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will provide a clear warning via voice or text message based on the user's emotional state. The input is the risk score and emotional information, and the output is a warning message. The warning is adjusted to encourage the user to act calmly.

[0562] (Application Example 2)

[0563] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0564] In modern society, telephone-based fraud is on the rise, and individuals are particularly vulnerable to becoming victims. To effectively combat such fraud, a system is needed that can detect potential fraud in real time and provide users with appropriate warnings. However, conventional systems cannot consider emotional states when determining the likelihood of fraud, leading to false positives and missed opportunities.

[0565] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0566] In this invention, the server includes means for analyzing voice data and document data and grasping an individual's emotional state using an emotion engine; means for taking emotional information into account when calculating fraud risk to make a more accurate determination; and means for evaluating the possibility of fraud and issuing a warning to alert the user. This makes it possible to judge the possibility of fraud more precisely and for users to receive warnings quickly.

[0567] "Audio data" refers to acoustic signals based on human speech, expressed in a digital format, and analyzed by a speech recognition system.

[0568] "Acoustic recognition technology" is a technology that automatically converts human speech into text information, and includes a process of extracting meaning from audio data.

[0569] "Text information" refers to character information generated from audio data, and is string data used for analysis and processing.

[0570] An "emotion engine" is a technology that analyzes an individual's emotional state from audio and document data and extracts specific emotional information.

[0571] "Fraud risk" is an indicator that assesses the likelihood of fraud based on information extracted from voice calls and other sources, and serves as a criterion for issuing warnings.

[0572] An "artificial intelligence model" is a computer program equipped with algorithms that learn from large datasets and perform predictions and analyses.

[0573] A "warning" is a notification that alerts users to potential fraud and provides information to help them prevent becoming a victim.

[0574] The system for implementing this invention is installed as an application on the user's smartphone and has the function of detecting the risk of phone fraud in real time. First, when the user receives a call, the terminal records the call audio and immediately sends the audio data to the server. The server uses acoustic recognition technology to convert the audio data into text information. Google Speech-to-Text is one example of a speech recognition engine that could be used here. Pre-processing such as noise reduction and volume equalization is also applied to the audio conversion process to improve the accuracy of the results.

[0575] The server then uses an emotion engine to analyze the user's emotional state from voice data and text information. Suitable emotion engines for this purpose include IBM Watson Tone Analyzer. Based on this emotional information, an artificial intelligence model detects specific terms and phrases that suggest fraud and calculates the fraud risk. In this process, emotional information, such as whether the user is confused, is a crucial factor in assessing fraud risk.

[0576] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. For example, if a user receives a phone call claiming "there is an outstanding bill," and emotional information detects that the user is experiencing high levels of anxiety, the keyword "outstanding" will further increase the risk of fraud. In response, the device will warn the user with a message stating, "This may be a scam. Do not directly verify the details; please verify the information through official channels."

[0577] An example of a prompt message could be: "I want to analyze the call content in real time and determine the risk of fraud involving the keyword 'unpaid.' To do this, please tell me how to convert the audio data into text, identify emotional distress, and issue a warning." This is how you can instruct the generative AI model.

[0578] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0579] Step 1:

[0580] When a user receives a call, the device records the call audio in real time. This recorded data is then input. The recorded audio data is programmed to be sent to the server immediately.

[0581] Step 2:

[0582] The server takes the received audio data as input and converts it into text information using acoustic recognition technology. Specifically, it uses an audio recognition engine such as Google Speech-to-Text to convert the data into text. The output is the converted text information.

[0583] Step 3:

[0584] The server takes text information as input and performs noise reduction and volume equalization. This preprocessing is done to improve the accuracy of the data. The output is clear, preprocessed text data.

[0585] Step 4:

[0586] Using an emotion engine, the server analyzes the user's emotional state from text data. Preprocessed text data is used as input. Specifically, it extracts user emotion information using tools such as IBM Watson Tone Analyzer. The output is the analyzed emotion information.

[0587] Step 5:

[0588] The server uses emotional information and text data as input to detect specific terms and phrases that suggest fraud using an artificial intelligence model. It performs calculations based on the input data to calculate fraud risk. The output is a fraud risk score.

[0589] Step 6:

[0590] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. Specifically, it will display a warning message on the screen to alert the user. The warning message will be generated based on the "generated AI model, prompt text," etc., according to the prompt text.

[0591] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0592] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0593] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[0594] [Fourth Embodiment]

[0595] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0596] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0597] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0598] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0599] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0600] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0601] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0602] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0603] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0604] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0605] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0606] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0607] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0608] The special fraud prevention system according to the present invention provides a mechanism that analyzes the content of a phone call in real time when a user receives a call and determines the possibility of fraud with high accuracy. The system performs a series of processes including voice data acquisition, voice recognition, text data conversion, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0609] First, the device records the call content as soon as it receives a call and sends this audio data to the server in real time. The server uses speech recognition technology to convert this audio data into text data, applying noise reduction and volume equalization processing to improve the accuracy of the text conversion.

[0610] Next, the server extracts keywords such as "son," "bank," and "payment required" from the converted text data. At this point, it uses natural language processing techniques to analyze the context based on typical scam phrases. If it determines that there is a possibility of fraud, it inputs the fraud risk into a machine learning model and scores the corresponding risk.

[0611] If this score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will put the caller on hold and display a message to the user saying, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0612] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is a bank, and we urgently need money." This system detects the keywords "bank" and "need money," scores the fraud risk, and if it determines the risk is high, sends a warning to the user. This process allows the user to act quickly and without hesitation, preventing them from becoming a victim of fraud.

[0613] As described above, the present invention aims to create an environment in which many users, including the elderly, can use the telephone safely.

[0614] The following describes the processing flow.

[0615] Step 1:

[0616] The device automatically records the conversation when the user receives a call. Upon starting the recording, the audio data is sent to the server in real time.

[0617] Step 2:

[0618] The server initiates speech recognition processing on the received audio data. This process improves audio quality by applying noise reduction and volume equalization. Subsequently, the audio data is converted into text-based data.

[0619] Step 3:

[0620] The server analyzes the generated text data and searches for pre-configured keywords and phrases. For example, it detects terms that suggest fraud, such as "son," "grandson," "bank," and "payment required."

[0621] Step 4:

[0622] The server analyzes the context of detected keywords and scores their likelihood of being a scam. This process uses a machine learning model based on learning from past scam patterns.

[0623] Step 5:

[0624] The server evaluates the scoring results and immediately notifies the device if it determines that the fraud risk is high. A fraud warning is triggered when the score exceeds a predetermined threshold.

[0625] Step 6:

[0626] If the device receives a warning that a call is likely to be a scam, it will put the caller on hold. Furthermore, it will display a warning message to the user stating, "This call may be a scam. Please talk to a family member or end the call and report it to the police."

[0627] Step 7:

[0628] The user reassesss the situation based on the displayed warning message. If necessary, they can hang up the phone and take action such as consulting with family or calling the police.

[0629] (Example 1)

[0630] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0631] In modern society, abusive behaviors, particularly those conducted via telephone, are on the rise, and many users, especially the elderly, are becoming victims. To address this problem, there is a need for a system that can efficiently analyze call content and quickly and automatically issue warnings about abusive behavior. Furthermore, it is necessary to reduce false positives by improving the accuracy of voice recognition and the accuracy of abusive behavior detection.

[0632] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0633] In this invention, the server includes means for acquiring voice information and converting that information into symbolic data using information recognition technology, means for extracting specific symbol sequences or expressions that suggest special behavior, and means for determining the possibility of special behavior through information processing. This makes it possible to quickly detect the possibility of special behavior from the content of a call and issue an appropriate warning to the user.

[0634] "Voice information" refers to data related to phone calls and other sounds, including recorded sounds and sound information received in real time.

[0635] "Information recognition technology" refers to the technology of converting speech information into symbolic data, and mainly includes speech recognition algorithms and related software.

[0636] "Symbolic data" refers to text or other digital data that has been converted from audio information.

[0637] "Special acts" refer to fraudulent or deceptive acts against users, and encompass all fraudulent activities that may result in harm.

[0638] A "symbol sequence" refers to a specific word or phrase within text or digital data, meaning a sequence of characters that has a specific meaning or context.

[0639] "Information processing" refers to the process of analyzing data and deriving conclusions or judgments from it, usually using algorithms and computational methods.

[0640] "Automated learning structure" refers to models and algorithms that utilize machine learning techniques, and are technologies that have the ability to automatically improve performance and accuracy based on data.

[0641] "False positive" refers to a situation where a special behavior is mistakenly identified as existing, and it signifies an error that occurs due to inaccurate detection.

[0642] This special fraud prevention system has the function of evaluating the possibility of special activity through real-time processing of voice information and warning the user.

[0643] The device records the call content as soon as the user receives a call and transmits the audio information to the server in real time. High-precision microphones and noise-canceling technology are used for recording to obtain accurate audio information.

[0644] The server uses advanced speech recognition technology to convert received audio information into symbolic data. Specifically, it performs noise reduction and volume equalization to improve the quality of the audio data. This process employs specific noise reduction algorithms and speech recognition software that analyzes the audio waveform.

[0645] Furthermore, the server utilizes natural language processing technology to extract symbol sequences related to specific behaviors from the symbolic data. During this process, a generative AI model performs machine learning to evaluate the likelihood of a specific behavior occurring. This automated learning structure is optimized based on historical data, minimizing false positives.

[0646] If the detected risk exceeds a predetermined threshold, the terminal immediately issues a warning to the user and, if necessary, puts the caller on hold. The warning uses visual messages and audible alerts to draw the user's attention.

[0647] As a concrete example, consider a scenario where a user receives a phone call and the caller says, "This is an urgent matter. We need to receive a money transfer." This system detects the sequence of symbols "money" and "transfer," and if it is deemed high-risk through scoring using a generative AI model, it automatically displays a warning.

[0648] Examples of prompts include, "How do you determine the risk of fraud when a money transfer request comes from a bank?" and "How can you detect potential unusual behavior in a phone call in real time?"

[0649] This system ensures that users can understand the risks and protect themselves from fraud.

[0650] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0651] Step 1:

[0652] The device records phone calls when a user answers one. The input is the audio from the phone call, and the output is the recorded audio data. When recording begins, the device can emit a notification sound stating, "This call is being recorded for security purposes."

[0653] Step 2:

[0654] The terminal transmits recorded audio data to the server in real time. The input is the recorded audio data, and the output is the transfer of the audio data to the server. This transfer is protected by a secure communication protocol, maintaining data confidentiality.

[0655] Step 3:

[0656] The server applies noise reduction and equalizes the volume of the received audio data. This results in consistent audio quality and improved speech recognition accuracy. The input is the transmitted audio data, and the output is the processed, high-quality audio data. This process involves signal processing using a specific algorithm.

[0657] Step 4:

[0658] The server uses speech recognition technology to convert high-quality audio data into symbolic data (text). The input is processed audio data, and the output is text data. This conversion uses state-of-the-art speech recognition software to transform the audio data into symbolic data.

[0659] Step 5:

[0660] The server uses natural language processing techniques to extract symbol sequences that suggest specific actions from text data. The input is the transformed text data, and the output is the extracted symbol sequence (specific keywords or phrases). In this process, the natural language processing algorithm plays a role in understanding the context.

[0661] Step 6:

[0662] The server uses a generative AI model to score the risk of specific actions based on the extracted symbol sequence. The input is the extracted symbol sequence, and the output is the risk score. The generative AI model is trained on historical data, enabling highly accurate scoring.

[0663] Step 7:

[0664] If the risk score exceeds a set threshold, the server sends a warning to the terminal, and the terminal displays a warning message to the user. The input is the risk score, and the output is a warning to the user. Specifically, a message such as "This call may be for unusual activity. Please talk to your family, end the call, and report to the relevant authorities" is displayed.

[0665] (Application Example 1)

[0666] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0667] In recent years, telephone-based fraud has been on the rise, and many users have fallen victim to it. In particular, elderly people and others unfamiliar with information technology have difficulty responding to fraudulent tactics. Therefore, there is a need for a system that can detect potential fraud in real time during a call and quickly warn the user.

[0668] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0669] In this invention, the server includes means for acquiring voice information and converting it into text information using voice recognition technology, means for extracting specific terms and expressions that suggest fraud, means for determining the likelihood of fraud using an evaluation score, and means for monitoring voice information and displaying a warning on the screen if there is a possibility of fraud. This makes it possible to immediately display a warning to the user if there is suspicion of fraud during a call, thereby preventing damage.

[0670] "Voice information" refers to data transmitted through voice communication such as telephone calls.

[0671] "Speech recognition technology" is a technology that analyzes input speech data and converts it into corresponding text data.

[0672] "Textual information" refers to data in text format converted using speech recognition technology.

[0673] "Specific terms and expressions" refer to key keywords and phrases that suggest fraudulent activity.

[0674] "Evaluation score" is an indicator used to quantify and show the likelihood of fraud.

[0675] A "warning display" is a warning message shown on the screen to alert the user.

[0676] To realize this application, the system is built around servers, terminals, and users.

[0677] The server utilizes speech recognition technology to process audio information in real time, converting audio data into text. During this process, it employs noise reduction and volume equalization algorithms to improve conversion accuracy. Furthermore, it uses natural language processing technology to extract specific terms and expressions from the text, and evaluates the likelihood of fraud based on a score. The server also utilizes a learning model to perform this evaluation, ensuring high reliability.

[0678] When the device receives warning information from the server, it displays a warning to the user in real time. For example, if the user receives a call saying, "This is a bank, and we urgently need money," the device will immediately display a warning to the user saying, "This call may be a scam."

[0679] Users can take appropriate action based on this warning. This allows them to quickly address suspected fraud during a phone call and prevent becoming a victim of fraud.

[0680] An example of a prompt might be, "Design a call content analysis algorithm to determine the likelihood of fraud." This prompt can provide assistance in analyzing voice data using a generative AI model.

[0681] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0682] Step 1:

[0683] When the device receives a phone call, it records the call audio in real time and sends it to the server as audio data. The input is the phone call audio, and the output is audio data. Since this audio data is transferred directly to the server, the specific operation involves starting the recording process at the beginning of the call and sending the data to the server over the network.

[0684] Step 2:

[0685] The server applies speech recognition technology to the received audio data to remove noise and equalize the volume. This converts the audio data into text information. The input is audio data, and the output is the refined text information. Specifically, the speech recognition engine analyzes the data and converts it into clear text.

[0686] Step 3:

[0687] The server extracts specific terms and expressions from the converted text information and calculates a score to assess the likelihood of fraud. The input is transliterated text information, and the output is a fraud risk assessment score. Specifically, it uses natural language processing to identify keywords, and a learning model calculates a risk score based on these keywords.

[0688] Step 4:

[0689] If the fraud risk assessment score exceeds a certain threshold, the server sends a warning signal to the terminal. The input is the fraud risk assessment score, and the output is a signal indicating the need for a warning. Specifically, the operation includes determining that the threshold has been exceeded, creating a warning signal, and transmitting it over the network.

[0690] Step 5:

[0691] The terminal receives a warning signal from the server and displays a warning on the user screen. The input is a warning signal, and the output is a visual warning message. Specifically, this includes displaying a warning message as a pop-up on the screen to attract the user's attention.

[0692] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0693] The special fraud prevention system according to the present invention analyzes voice data in real time to determine the possibility of fraud, and also recognizes the user's emotions to further improve accuracy. This system performs a series of processes including voice data acquisition, voice recognition, text data conversion, emotion analysis using an emotion engine, keyword extraction, fraud risk scoring, result determination, and warning the user.

[0694] First, the device records the conversation when the user receives a call and sends the audio data to the server in real time. The server uses speech recognition technology to convert the audio data into text data, and applies noise reduction and volume equalization processing as needed to improve the accuracy of the analysis.

[0695] Subsequently, the server uses an emotion engine to determine the user's emotional state from text and audio data. At this stage, it analyzes the user's tone of voice, speaking speed, and phrasing to obtain emotional information such as whether the user is nervous or suspicious.

[0696] Next, the server detects pre-configured keywords from the generated text data and extracts phrases that may suggest fraud. Then, when scoring the fraud risk, it takes into account the user's emotional state obtained by the emotion engine to make a more refined judgment.

[0697] If the fraud risk score exceeds a predetermined threshold, the device will issue a user-specific warning. Depending on the user's emotional state, for example, if the user is feeling significantly anxious, a particularly emphasized warning message will be displayed to reduce psychological burden while still providing attention.

[0698] For example, if a user receives a call and explains that they need money because their son was in an accident, the system detects keywords such as "money" and "accident," and the emotion engine recognizes that the user is speaking in a clearly confused tone. As a result, the system warns of a high risk of fraud and displays a cautionary message to the user, helping them to handle the situation calmly.

[0699] Thus, by combining an emotion engine, the present invention provides an environment that can more safely protect users, including the elderly and those who are not comfortable with phone calls, from telephone fraud.

[0700] The following describes the processing flow.

[0701] Step 1:

[0702] The device starts recording the conversation when the user receives a call. The recorded audio data is streamed to the server in real time.

[0703] Step 2:

[0704] The server uses speech recognition technology to convert speech data into text data. During the conversion process, noise reduction and volume equalization are applied to improve the quality of the speech.

[0705] Step 3:

[0706] The server uses an emotion engine to analyze the user's emotions from voice and text data. It evaluates factors such as voice tone, speed, and stress level to identify the user's emotional state.

[0707] Step 4:

[0708] The server searches the converted text data for keywords such as "son," "accident," and "need money." These keywords may suggest fraud, so natural language processing techniques are used to detect them.

[0709] Step 5:

[0710] The server scores the fraud risk based on the extracted keywords and sentiment analysis results. It uses a machine learning model to quantify the risk level based on historical data.

[0711] Step 6:

[0712] The server issues a warning if the fraud risk score exceeds a set threshold. In this case, it instructs the terminal to prepare to play a hold tone for the caller.

[0713] Step 7:

[0714] The device displays a warning message to the user. The message is tailored to the user's emotional state and provides instructions such as, "This call may be a scam. Stay calm and check with your family or report it to the police."

[0715] Step 8:

[0716] The user should follow the instructions displayed on the device. If they determine the call is suspicious, they should immediately end the call and take action to ensure their safety.

[0717] Through this process, the system detects the risk of telephone fraud in real time, provides optimal warnings to users, and offers support to protect them from becoming victims of fraud.

[0718] (Example 2)

[0719] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0720] In recent years, telephone-based fraud has been increasing, and vulnerable groups such as the elderly are particularly susceptible to becoming victims. Existing countermeasures lack the speedy and accurate means to identify fraud, and warnings do not take into account the user's emotional state, potentially leading to inappropriate decisions. Therefore, there is a need for a system that can identify fraudulent activity in real time and issue appropriate warnings.

[0721] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0722] In this invention, the server includes means for acquiring voice information and converting it into text information using analysis techniques, means for analyzing specific words and expressions to detect the possibility of fraudulent activity, and means for analyzing the user's emotional state and reflecting it in the risk assessment. This enables real-time identification of fraudulent activity and the issuance of warnings that take the user's emotions into consideration.

[0723] "Audio information" refers to sound information that is acquired as audio data and used for analysis.

[0724] "Analysis technology" refers to the techniques used to analyze acquired data and convert it into a format that humans can understand.

[0725] "Textual information" refers to information in text format obtained as a result of converting audio information using analysis technology.

[0726] "Words and expressions" refer to words, phrases with meaning, or specific expressions used within a text.

[0727] "Fraudulent activity" refers to actions that violate laws or norms, particularly inappropriate behavior including fraud.

[0728] "Detecting possibility" means confirming the likelihood of a particular event or state occurring.

[0729] "Emotional state" refers to the state of a person's feelings or emotions, such as joy or anxiety.

[0730] "Risk assessment" is the process of analyzing and judging potential dangers and losses.

[0731] "Issuing a warning" is the act of notifying others that there is a risk.

[0732] This invention provides a fraud prevention system that utilizes voice, which analyzes voice information from incoming phone calls to a user in real time and issues a warning if there is suspicion of fraudulent activity.

[0733] The device records voice data from the user and transmits the voice data to the server via the internet connection. At this stage, the device collects the voice using its built-in microphone and recording application.

[0734] The server uses speech recognition technology such as Google Cloud Speech-to-Text to convert the received audio data into text information. In this step, noise reduction and volume equalization are performed to improve the accuracy of the analysis. Noise reduction is the process of reducing background noise, and volume equalization is the process of adjusting the volume of the audio.

[0735] The server then utilizes an emotion analysis engine, such as IBM Watson Tone Analyzer, to understand the user's emotional state based on the obtained text information. This helps identify whether the user is experiencing emotions such as anxiety or confusion.

[0736] Next, the server uses a pre-configured analysis model to extract keywords that suggest fraud. In this process, keywords such as "money" and "accident" are categorized, and a risk assessment is performed in combination with emotional states.

[0737] If the server detects a potential scam and the risk exceeds a certain threshold, the device will issue a warning to the user. This warning may be presented as an audio or on-screen message, encouraging the user to remain calm.

[0738] For example, if a user receives a call saying, "My son has been in an accident," the system integrates keywords such as "accident" and "money" with sentiment data and warns of a high risk of fraud.

[0739] An example of a prompt when using a generative AI model is an input in the form of, "Analyze the tone of this voice to identify the emotion: 'It's terrible, I've been in an accident.'" This prompt allows the model to perform practical analysis based on the voice data.

[0740] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0741] Step 1:

[0742] The device begins recording the conversation when the user receives a call. This process uses the built-in microphone to capture audio data and saves it as a recording file. The input is live audio, and the output is an audio file. A key feature is that the recording occurs almost in real time.

[0743] Step 2:

[0744] The device sends the recorded audio file to the server via the internet. At this stage, the audio file is encrypted for security purposes. The input is the recorded file, and the output is the audio data securely transferred to the server.

[0745] Step 3:

[0746] The server converts the received audio data into text data using a speech recognition service (e.g., Google Cloud Speech-to-Text). This process simultaneously performs noise reduction and volume equalization, transforming the input audio data into clearer audio data. The final output is noise-reduced text data.

[0747] Step 4:

[0748] The server performs sentiment analysis using text and audio data. This process uses a sentiment analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state. Input is text data and voice tone, and output is information about the user's emotional state. Specifically, it infers emotional characteristics from voice tone and word choice.

[0749] Step 5:

[0750] The server extracts keywords that may suggest fraud from the generated text data and integrates them with sentiment analysis results to assess fraud risk. This process searches for pre-defined keywords such as "money" and "accident," and scores the risk if applicable. The input is text data and sentiment state, and the output is a risk score.

[0751] Step 6:

[0752] If the risk score exceeds a predetermined threshold, the device will issue a warning to the user. Specifically, it will provide a clear warning via voice or text message based on the user's emotional state. The input is the risk score and emotional information, and the output is a warning message. The warning is adjusted to encourage the user to act calmly.

[0753] (Application Example 2)

[0754] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0755] In modern society, telephone-based fraud is on the rise, and individuals are particularly vulnerable to becoming victims. To effectively combat such fraud, a system is needed that can detect potential fraud in real time and provide users with appropriate warnings. However, conventional systems cannot consider emotional states when determining the likelihood of fraud, leading to false positives and missed opportunities.

[0756] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0757] In this invention, the server includes means for analyzing voice data and document data and grasping an individual's emotional state using an emotion engine; means for taking emotional information into account when calculating fraud risk to make a more accurate determination; and means for evaluating the possibility of fraud and issuing a warning to alert the user. This makes it possible to judge the possibility of fraud more precisely and for users to receive warnings quickly.

[0758] "Audio data" refers to acoustic signals based on human speech, expressed in a digital format, and analyzed by a speech recognition system.

[0759] "Acoustic recognition technology" is a technology that automatically converts human speech into text information, and includes a process of extracting meaning from audio data.

[0760] "Text information" refers to character information generated from audio data, and is string data used for analysis and processing.

[0761] An "emotion engine" is a technology that analyzes an individual's emotional state from audio and document data and extracts specific emotional information.

[0762] "Fraud risk" is an indicator that assesses the likelihood of fraud based on information extracted from voice calls and other sources, and serves as a criterion for issuing warnings.

[0763] An "artificial intelligence model" is a computer program equipped with algorithms that learn from large datasets and perform predictions and analyses.

[0764] A "warning" is a notification that alerts users to potential fraud and provides information to help them prevent becoming a victim.

[0765] The system for implementing this invention is installed as an application on the user's smartphone and has the function of detecting the risk of phone fraud in real time. First, when the user receives a call, the terminal records the call audio and immediately sends the audio data to the server. The server uses acoustic recognition technology to convert the audio data into text information. Google Speech-to-Text is one example of a speech recognition engine that could be used here. Pre-processing such as noise reduction and volume equalization is also applied to the audio conversion process to improve the accuracy of the results.

[0766] The server then uses an emotion engine to analyze the user's emotional state from voice data and text information. Suitable emotion engines for this purpose include IBM Watson Tone Analyzer. Based on this emotional information, an artificial intelligence model detects specific terms and phrases that suggest fraud and calculates the fraud risk. In this process, emotional information, such as whether the user is confused, is a crucial factor in assessing fraud risk.

[0767] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. For example, if a user receives a phone call claiming "there is an outstanding bill," and emotional information detects that the user is experiencing high levels of anxiety, the keyword "outstanding" will further increase the risk of fraud. In response, the device will warn the user with a message stating, "This may be a scam. Do not directly verify the details; please verify the information through official channels."

[0768] An example of a prompt message could be: "I want to analyze the call content in real time and determine the risk of fraud involving the keyword 'unpaid.' To do this, please tell me how to convert the audio data into text, identify emotional distress, and issue a warning." This is how you can instruct the generative AI model.

[0769] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0770] Step 1:

[0771] When a user receives a call, the device records the call audio in real time. This recorded data is then input. The recorded audio data is programmed to be sent to the server immediately.

[0772] Step 2:

[0773] The server takes the received audio data as input and converts it into text information using acoustic recognition technology. Specifically, it uses an audio recognition engine such as Google Speech-to-Text to convert the data into text. The output is the converted text information.

[0774] Step 3:

[0775] The server takes text information as input and performs noise reduction and volume equalization. This preprocessing is done to improve the accuracy of the data. The output is clear, preprocessed text data.

[0776] Step 4:

[0777] Using an emotion engine, the server analyzes the user's emotional state from text data. Preprocessed text data is used as input. Specifically, it extracts user emotion information using tools such as IBM Watson Tone Analyzer. The output is the analyzed emotion information.

[0778] Step 5:

[0779] The server uses emotional information and text data as input to detect specific terms and phrases that suggest fraud using an artificial intelligence model. It performs calculations based on the input data to calculate fraud risk. The output is a fraud risk score.

[0780] Step 6:

[0781] If the fraud risk score exceeds a set threshold, the device will display a warning to the user. Specifically, it will display a warning message on the screen to alert the user. The warning message will be generated based on the "generated AI model, prompt text," etc., according to the prompt text.

[0782] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0783] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0784] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0785] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0786] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. In the upper and lower directions of the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. Also, the upper side of the concentric circles is where "pleasant" emotions are located, and the lower side is where "unpleasant" emotions are located. In this way, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0787] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0788] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0789] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0790] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0791] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0792] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[0793] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[0794] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0795] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[0796] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0797] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0798] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0799] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0800] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0801] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0802] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0803] The following is further disclosed regarding the embodiments described above.

[0804] (Claim 1)

[0805] A means of acquiring audio data and converting it into text data using speech recognition technology,

[0806] A method for extracting specific keywords or phrases that suggest fraud,

[0807] A method for scoring and determining the likelihood of fraud,

[0808] A means of warning the user based on the judgment result and prompting them to take appropriate action,

[0809] A system that includes this.

[0810] (Claim 2)

[0811] The system according to claim 1, which performs noise reduction and volume equalization processing to improve the accuracy of speech recognition results.

[0812] (Claim 3)

[0813] The system according to claim 1, which uses a machine learning model to score fraud risk.

[0814] "Example 1"

[0815] (Claim 1)

[0816] A means for acquiring audio information and converting that information into symbolic data using information recognition technology,

[0817] A means for extracting specific sequences of symbols or expressions that suggest special acts,

[0818] A means of determining the possibility of special acts through information processing,

[0819] A means of issuing warnings to users based on the judgment results and encouraging appropriate action,

[0820] A system that includes this.

[0821] (Claim 2)

[0822] The system according to claim 1, which performs sound removal and volume adjustment processing to improve the accuracy of the information recognition result.

[0823] (Claim 3)

[0824] The system according to claim 1, which uses an automated learning structure to determine the risk of special conduct.

[0825] "Application Example 1"

[0826] (Claim 1)

[0827] A means for acquiring audio information and converting it into text information using speech recognition technology,

[0828] A means of extracting specific terms and expressions that suggest fraud,

[0829] A method for determining the likelihood of fraud based on an evaluation score,

[0830] A means of issuing warnings to users based on the judgment results and prompting them to take appropriate action,

[0831] A means of monitoring audio information and displaying a warning on the screen if there is a possibility of fraud,

[0832] A system that includes this.

[0833] (Claim 2)

[0834] The system according to claim 1, which performs noise reduction and volume equalization processing to improve the accuracy of speech recognition results.

[0835] (Claim 3)

[0836] The system according to claim 1, which uses a learning model to score the fraud risk.

[0837] "Example 2 of combining an emotion engine"

[0838] (Claim 1)

[0839] A means for acquiring audio information and converting it into text information using analysis technology,

[0840] A means of detecting the possibility of misconduct by analyzing specific words or expressions,

[0841] A means of analyzing the user's emotional state and reflecting it in risk assessment,

[0842] A means of issuing a warning and alerting the user when the value of the risk exceeds the standard,

[0843] A system that includes this.

[0844] (Claim 2)

[0845] The system according to claim 1, which performs acoustic interference reduction and volume control processing to improve the accuracy of speech recognition results.

[0846] (Claim 3)

[0847] The system according to claim 1, which utilizes machine learning methods for risk assessment.

[0848] "Application example 2 when combining with an emotional engine"

[0849] (Claim 1)

[0850] A means of acquiring audio data and converting it into text information using acoustic recognition technology,

[0851] A means of detecting specific terms and phrases that suggest fraud,

[0852] A means of assessing the possibility of fraud and issuing warnings to alert users,

[0853] A means of analyzing audio and document data to understand an individual's emotional state using an emotion engine,

[0854] When calculating fraud risk, a method is needed to make a more accurate judgment by taking emotional information into account.

[0855] A system that includes this.

[0856] (Claim 2)

[0857] The system according to claim 1, which performs noise suppression and volume adjustment processing to improve the accuracy of the recognition result.

[0858] (Claim 3)

[0859] The system according to claim 1, which uses an artificial intelligence model to assess fraud risk and analyzes voice data and emotional data. [Explanation of Symbols]

[0860] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of acquiring audio data and converting it into text data using speech recognition technology, A method for extracting specific keywords or phrases that suggest fraud, A method for scoring and determining the likelihood of fraud, A means of warning the user based on the judgment result and prompting them to take appropriate action, A system that includes this.

2. The system according to claim 1, which performs noise reduction and volume equalization processing to improve the accuracy of speech recognition results.

3. The system according to claim 1, which uses a machine learning model to score fraud risk.

Images