system

Abstract

We provide the system. [Solution] A means for collecting conversation content, emails, or chats in real time, A means of analyzing the collected data and determining the risk of customer harassment, A means of issuing a warning based on the judgment result, A system including feedback on warnings and means for generating appropriate alternative text.

Description

Technical Field

[0002]

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is 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 in 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] In recent years, high-pressure behaviors from customers, so-called customer harassment (cushara), have been increasing in companies, causing psychological burdens and turnover among employees. Although companies are urgently required to take measures against this problem, it is difficult to determine individual cases in real time and respond appropriately. In addition, some customers may inadvertently take high-pressure actions, and appropriate feedback and guidance to calm emotions are required. Therefore, there is a need to develop a system that automatically determines cushara and its countermeasures, and provides feedback to customers.

Means for Solving the Problems

[0005] The present invention solves the above-mentioned problems with a system that includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data to determine the risk of customer harassment; means for issuing a warning based on the determination result; means for providing feedback on the warning and generating appropriate alternative text.

[0006] Specifically, by using a conversation content collection method that includes speech recognition technology to convert audio data into text data, telephone and meeting conversations can be analyzed in real time. In addition, by using an artificial intelligence model that performs sentiment analysis and detects specific keywords, the text of emails and chats can be analyzed with high accuracy, and the risk of customer harassment can be immediately determined. This makes it possible to quickly provide appropriate feedback to both employees and customers and reduce the risk of customer harassment.

[0007] "Conversation content" refers to the content of a conversation that took place either verbally or in person.

[0008] "Email" refers to a method of communication in the form of messages sent and received over computer networks such as the internet.

[0009] "Chat" refers to a means of communication that involves exchanging text messages in real time on computers or mobile devices.

[0010] "Real-time" refers to a state where data is processed without delay from the moment it is generated.

[0011] "Means of collection" refers to devices or methods for acquiring information and storing or transferring it for processing.

[0012] "Means of analysis" refers to devices and methods for interpreting data and detecting specific patterns or anomalies.

[0013] "Customer harassment" refers to unreasonable demands or aggressive behavior from customers towards company employees.

[0014] "Means of determining risk" refers to devices or methods that evaluate data to determine whether or not a particular risk exists.

[0015] "Warning mechanisms" refer to devices or methods that provide notification when a problem is detected.

[0016] "Feedback" refers to providing some kind of response or opinion regarding an action or result.

[0017] "Means for generating alternative text" refers to devices or methods for replacing the content of the original text with more appropriate words.

[0018] "Speech recognition means" refers to devices or methods that analyze speech signals and convert them into text. [Brief explanation of the drawing]

[0019] [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 the data processing device and 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] Shows an emotion map where multiple emotions are mapped. [Figure 10] 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

[0020] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described according to the accompanying drawings.

[0021] First, the language used in the following description will be explained.

[0022] In the following embodiments, a processor with a reference number (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.

[0023] In the following embodiments, signed RAM (Random Access Memory) is a memory that temporarily stores information and is used as work memory by the processor.

[0024] 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.

[0025] 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).

[0026] 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."

[0027] [First Embodiment]

[0028] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0029] 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.

[0030] 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).

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0036] 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.

[0037] 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.

[0038] 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.

[0039] 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".

[0040] overview

[0041] This invention is a system that collects conversation content, emails, or chats in real time, analyzes the risk of customer harassment, and issues warnings. Furthermore, it is characterized by generating appropriate feedback and alternative text along with the warning and providing them to the user.

[0042] System Configuration

[0043] 1. User's terminal

[0044] It collects data from conversations, emails, and chats.

[0045] Includes speech recognition functionality for converting audio data into text data.

[0046] It communicates with the server and sends data.

[0047] 2. Server

[0048] Receive data sent from the device.

[0049] The system analyzes data in real time to determine the risk of customer harassment.

[0050] Based on the judgment, a warning message and feedback are generated.

[0051] Generate appropriate alternative text and send it to the user's device.

[0052] Program processing

[0053] 1. Data Collection

[0054] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[0055] 2. Real-time analysis

[0056] The server analyzes the received text data and assesses the risk of customer harassment. This assessment includes sentiment analysis and detection of specific keywords. Based on the analysis results, a risk level is determined.

[0057] 3. Generating warnings and feedback

[0058] If the server determines that there is a high risk of customer harassment, it will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[0059] Specific example

[0060] Example: Detecting customer harassment via email

[0061] 1. The user composes an emotional email and clicks the send button.

[0062] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0063] 2. The device sends the email body to the server.

[0064] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[0065] 3. The server analyzes the content of the email and determines the risk of customer harassment.

[0066] Judgment result: High risk

[0067] 4. The server sends feedback to the user's terminal along with a warning message.

[0068] Warning message: "Suspected customer harassment. Risk level: High"

[0069] Feedback: "This kind of expression should be avoided."

[0070] 5. The server generates an appropriate alternative sentence and provides it to the user.

[0071] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[0072] Example: Detecting customer harassment in chat

[0073] 1. The user types an aggressive message in the chat and clicks the send button.

[0074] Example: "Your response is too slow. What are you doing?"

[0075] 2. The device sends the chat content to the server.

[0076] Sent text: "Your response is too slow. What are you doing?"

[0077] 3. The server analyzes the chat content and assesses the risk of customer harassment.

[0078] Assessment result: Medium risk

[0079] 4. The server sends a warning and feedback to the user's terminal.

[0080] Warning message: "Caution is advised. Risk level: Medium"

[0081] Feedback: "This expression may be considered rude."

[0082] 5. The server generates an appropriate alternative sentence and provides it to the user.

[0083] Alternative suggestion: "Please respond promptly if any problems arise."

[0084] In this way, this system can assess the risk of customer harassment in real time and provide appropriate feedback and alternative text, thereby promoting proper communication between companies and customers and reducing the psychological burden on employees.

[0085] The following describes the processing flow.

[0086] Step 1:

[0087] The user begins typing an email, chat message, or voice call. The device prepares to collect this data.

[0088] Step 2:

[0089] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[0090] Step 3:

[0091] The device collects email body text, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[0092] Step 4:

[0093] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[0094] Step 5:

[0095] The server inputs the received data into an analysis tool. This analysis tool includes sentiment analysis and specific keyword detection algorithms.

[0096] Step 6:

[0097] The server uses analysis tools to analyze incoming data in real time. As a result, the risk level of customer harassment is determined. The risk level is evaluated in three stages: high, medium, and low.

[0098] Step 7:

[0099] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[0100] Step 8:

[0101] The server generates appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[0102] Step 9:

[0103] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[0104] Step 10:

[0105] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[0106] Step 11:

[0107] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[0108] In this way, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions.

[0109] (Example 1)

[0110] 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."

[0111] In today's business environment, communication between companies and customers takes place through a variety of means. However, excessive demands and inappropriate remarks from customers, i.e., customer harassment, can increase the psychological burden on employees and lead to a deterioration of the working environment. Therefore, it is necessary to monitor conversation content in real time, detect the risk of customer harassment early, and take appropriate measures. However, current systems have problems such as delays in analyzing collected data and inability to accurately assess risks.

[0112] 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.

[0113] In this invention, the server includes means for collecting conversation content, electronic messages, or instant messages in real time; means for analyzing the collected data and evaluating the risk of customer harassment; and means for issuing warnings based on the evaluation results. This makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[0114] "Conversation content" refers to the content of communication conducted through voice.

[0115] "Electronic messages" refer to messages sent and received in digital format, such as emails and text messages.

[0116] "Instant messaging" refers to real-time chats and instant messages.

[0117] "Methods for collecting information in real time" refers to technologies for collecting conversation content and messages almost instantly.

[0118] "Analysis" refers to the process of evaluating collected data using mechanical and algorithmic methods to extract meaning and emotion.

[0119] "Customer harassment" refers to acts that cause psychological distress to employees through excessive demands or inappropriate remarks from customers.

[0120] "Means of risk assessment" refers to technologies used to determine the risk level of customer harassment through data analysis.

[0121] "Warning mechanisms" refer to technologies used to send warning messages to users when there is a risk.

[0122] "Speech recognition means" refers to technology for converting speech data into text data.

[0123] A "machine intelligence model" refers to a technology that uses artificial intelligence to analyze data and recognize specific patterns or emotions.

[0124] "Alternative text" refers to appropriate wording that is recommended to be used instead of the original message.

[0125] "Feedback" refers to suggestions, proposals, or specific advice regarding a warning.

[0126] The system in this invention is built to collect conversation content, electronic messages, or instant messages in real time, determine the risk of customer harassment, and issue warnings. This system includes a user terminal, a server, and a generative AI model.

[0127] 1. User terminal

[0128] The user terminal is a device that collects data from conversations, electronic messages, or instant messages. It also includes a speech recognition system that converts voice data into text data. Speech recognition software such as "Google® Cloud Speech-to-Text" is used for speech recognition. It also has the function of collecting text data entered by the user and sending it to the server.

[0129] Specific example: A user types the message "You can't even fulfill such a simple request? Do you even have any motivation?" into the chat window and clicks the send button.

[0130] 2. Server

[0131] The server is a central processing unit that analyzes text data received from user terminals. It analyzes the received data in real time and assesses the risk of customer harassment. The analysis uses natural language processing (NLP) technologies such as "SpaCy" and "IBM Watson®" for sentiment analysis and detection of specific keywords. Based on the analysis results, it determines the risk level and generates warning messages and feedback as needed. Furthermore, it uses generative AI models (e.g., OpenAI®'s GPT-3®) to generate appropriate alternative text.

[0132] Specific example: The server receives a message saying, "You can't even fulfill such a simple request? Are you even motivated?" and the analysis results indicate it's "high risk." Subsequently, it generates a warning message, "Suspected customer harassment. Risk level: high," and appropriate feedback, "Such language should be avoided," and creates alternative text, "Is there anything you didn't understand in this section? I'd be happy to explain in more detail."

[0133] Example of a prompt

[0134] A user is attempting to send the following message. Please check if this message poses a risk of customer harassment and generate a warning message and appropriate alternative text. Original message: "You can't even fulfill such a simple request? Are you even motivated?"

[0135] The above describes the embodiments for carrying out the present invention. This system makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[0136] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0137] Step 1:

[0138] The user creates a conversation, e-message, or instant message and clicks the send button. The user's device then collects the created message.

[0139] Input: User-entered conversation, e-message, or instant message (e.g., "You can't even fulfill such a simple request? Are you even motivated?")

[0140] Output: Message data stored on the user's terminal

[0141] Specific action: The user types a message in the chat window and clicks the send button.

[0142] Step 2:

[0143] The user terminal prepares to send the collected message data to the server in real time. If voice data is included, speech recognition is used to convert the voice to text.

[0144] Input: Voice data or text data

[0145] Output: Text data

[0146] Specific operation: A speech recognition engine (e.g., Google Cloud Speech-to-Text) converts speech to text. If the data is text, it is used as is.

[0147] Step 3:

[0148] The user terminal sends text data to the server.

[0149] Input: Text data (Example: "You can't even meet such a simple request? Do you even have any motivation?")

[0150] Output: Text data sent to the server

[0151] Specific operation: The user terminal uses its communication function to send the converted text data to the server.

[0152] Step 4:

[0153] The server analyzes the received text data. The server uses natural language processing technologies (e.g., SpaCy, IBM Watson) to perform sentiment analysis and detect specific keywords.

[0154] Input: Text data

[0155] Output: Analysis results (sentiment score, keyword detection results)

[0156] Specific operation: The server uses SpaCy or IBM Watson to analyze the sentiment of incoming messages and detect specific keywords.

[0157] Step 5:

[0158] The server assesses the risk of customer harassment based on the analysis results. This assessment includes sentiment scores and keyword detection results.

[0159] Input: Analysis results (sentiment score, keyword detection results)

[0160] Output: Risk assessment results (high risk, medium risk, low risk, etc.)

[0161] Specific operation: The server evaluates the risk level of customer harassment based on sentiment scores and keyword detection results (e.g., determines it as "high risk").

[0162] Step 6:

[0163] The server generates a warning message based on the risk assessment results. If the risk is high, it also generates specific improvement suggestions and alternative text. A generative AI model is used to create appropriate feedback and alternative text.

[0164] Input: Risk assessment results

[0165] Output: Warning message, feedback, alternative text

[0166] Specific actions: The server generates a warning message "Suspected customer harassment. Risk level: High," feedback "Such language should be avoided," and alternative text "Is there anything you don't understand about this part? We'd be happy to explain in more detail."

[0167] Step 7:

[0168] The server sends generated warning messages, feedback, and alternative text to the user's terminal.

[0169] Input: Warning message, feedback, alternative text

[0170] Output: Warning messages, feedback, and alternative text sent to the user's terminal.

[0171] Specific operation: The server uses its communication function to send warning messages, feedback, and alternative text to the user's terminal.

[0172] Through the above processing steps, this system can assess the risk of customer harassment in real time and immediately provide necessary countermeasures.

[0173] (Application Example 1)

[0174] 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."

[0175] Customer support and call center staff are required to detect the risk of customer harassment in real time during customer conversations and to take appropriate action quickly based on that risk. Traditional methods require staff to assess the risk and devise countermeasures themselves, which places a significant psychological burden on them and carries the risk of incorrect responses. This challenge needs to be addressed.

[0176] 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.

[0177] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing a warning based on the determination result; means for providing feedback on the warning and generating appropriate alternative text; means for collecting conversation content using speech recognition means that convert speech to text in real time; means for determining the risk of customer harassment using a device that performs sentiment analysis; means for determining the risk level based on the determination result and presenting an appropriate alert; means for displaying feedback on a smart device worn by an employee; and natural language processing means for generating appropriate alternative text. This enables staff to detect the risk of customer harassment in real time and take appropriate action quickly.

[0178] "Conversation content" refers to voice-based communication between customer support or call center staff and customers.

[0179] "Email" refers to messages sent and received electronically via the internet.

[0180] "Chat" refers to real-time text-based communication.

[0181] "Means of real-time collection" refers to methods or devices for instantly acquiring this data as conversations, emails, or chats occur.

[0182] "Means of analyzing collected data" refers to methods or devices for analyzing collected conversations, emails, or chats to assess the risk of customer harassment.

[0183] "Means for determining the risk of customer harassment" refers to a method or device that detects inappropriate words or actions from customers and determines their risk level based on collected and analyzed data.

[0184] "Means of issuing warnings" refers to methods or devices for sending appropriate notifications or warning messages to users when a risk of customer harassment is identified.

[0185] "Feedback" or "means for generating appropriate alternative text" refers to a method or device for generating and proposing appropriate countermeasures or alternative expressions to the user in response to the risk of customer harassment.

[0186] "Speech recognition means" refers to technologies and devices for converting speech data into text data.

[0187] "An emotional analysis device" refers to a method or device for analyzing the emotional aspects of text data and detecting specific emotions or tones.

[0188] "Means for determining risk levels" refers to methods or devices for evaluating the degree of customer harassment risk and quantifying that level based on the results of sentiment analysis.

[0189] "Means of providing appropriate alerts" refers to methods or devices for providing warnings or notifications to employees based on the determined risk level.

[0190] "Means of displaying feedback on smart devices" refers to methods or devices for displaying risk levels or feedback messages on wearable devices or other smart devices worn by users.

[0191] "Natural language processing means" refers to language processing technologies and devices that understand text data and generate appropriate alternative text.

[0192] This invention provides a system for detecting the risk of customer harassment in real time in customer support and call centers, and for taking appropriate action quickly based on that risk. The system includes user terminals, servers, and smart devices worn by employees.

[0193] System Configuration

[0194] 1. User's terminal

[0195] It has the ability to collect data from conversations, emails, and chats.

[0196] It is equipped with speech recognition means for converting audio data into text data.

[0197] It has the function of communicating with a server and sending data.

[0198] 2. Server

[0199] It has the ability to receive data sent from a device and analyze it in real time.

[0200] This includes an emotion analysis device for analyzing data and determining the risk of customer harassment, as well as an artificial intelligence model for detecting specific keywords.

[0201] Based on the assessment results, the risk level is determined, and warning and feedback messages are generated.

[0202] The generated warnings and feedback are sent to the user's device and to smart devices that display the feedback.

[0203] 3. Smart devices

[0204] Employees wear the device, which displays warning and feedback messages sent from the server.

[0205] Hardware and software to be used

[0206] Speech recognition method: The "speech_recognition" library and Google's speech recognition API are used.

[0207] Sentiment analysis device: Uses a pre-trained sentiment classification model (such as BERT) using the "transformers" library.

[0208] Smart devices: Employees can wear smart glasses or use smartphones.

[0209] Servers: High-performance servers are used for processing and communicating this data.

[0210] Processing flow

[0211] 1. Data collection:

[0212] When a user initiates a conversation or chat with a customer, the device collects conversation and chat data in real time, and the voice data is converted into text data using speech recognition technology.

[0213] 2. Real-time analysis:

[0214] The collected text data is sent to a server, which uses sentiment analysis devices and artificial intelligence models to analyze the risk of customer harassment.

[0215] 3. Generating warnings and feedback:

[0216] If the risk level of customer harassment is high, the server will generate a warning and specific feedback message, or appropriate alternative text, depending on the risk level.

[0217] 4. Notifications and Feedback:

[0218] The generated warning and feedback messages are sent to and displayed on the user's device and the employee's smart device, allowing employees to take appropriate action immediately.

[0219] Specific example

[0220] For example, if a support staff member is talking to a customer and the customer makes an aggressive statement such as, "Your response is too slow! What's going on?", this system collects the statement in real time and converts it to text using speech recognition. Then, an emotion analyzer analyzes this text, and if it is determined to be high risk, the server generates a warning message and feedback such as, "Specifically, what is the problem? We'd be happy to help," which is displayed on the staff member's smart glasses.

[0221] Prompts for Generative AI Models

[0222] Design a custom-made smart glasses application to detect customer harassment during customer support conversations and provide appropriate alerts and feedback in real time. Generate Python pseudocode that meets the following requirements.

[0223] 1. Collection of audio data

[0224] 2. Text conversion using speech recognition

[0225] 3. Risk assessment using sentiment analysis

[0226] 4. Generate feedback based on the judgment result.

[0227] The above describes the embodiments for carrying out the present invention. By introducing this system, it becomes possible to detect customer harassment in real time at customer support sites and take prompt and appropriate action, thereby reducing the psychological burden on employees.

[0228] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0229] Step 1:

[0230] The user initiates a conversation with a customer. The input is the voice conversation with the customer. The device collects this voice data. The device captures the voice data in real time using its built-in microphone.

[0231] Step 2:

[0232] The device converts the collected audio data into text data. The input is the audio data collected in step 1. The device uses its speech recognition capabilities (using the speech_recognition library and Google's speech recognition API) to convert the audio data into text data. The output is text data.

[0233] Step 3:

[0234] The text data is sent to the server. The input is the text data generated in step 2. The terminal communicates with the server and sends the text data to the server. The output is that the text data is sent to the server.

[0235] Step 4:

[0236] The server analyzes the received text data. The input is the text data sent to the server in step 3. The server uses a sentiment analyzer (using the transformers library) and an artificial intelligence model that detects specific keywords to perform sentiment analysis and keyword detection on the text data. The output is the risk assessment result.

[0237] Step 5:

[0238] The server determines the risk level based on the risk assessment results and generates warnings and feedback. The input is the risk assessment results obtained in step 4. The server determines the risk level (high, medium, low) according to the assessment results and generates appropriate warning messages and feedback messages, as well as appropriate alternative text using natural language processing. The output is the generated warning messages and feedback messages.

[0239] Step 6:

[0240] The server sends the generated warning and feedback messages to the user's terminal and smart device. The input is the warning and feedback messages generated in step 5. The server sends these messages to the terminal and smart device via communication. The output is the warning and feedback displayed on the user's terminal and smart device.

[0241] Step 7:

[0242] The user's terminal and smart device display warnings and feedback. The input consists of the warning and feedback messages sent from the server in step 6. The terminal and smart device use their display functions to present the risk level, warnings, feedback, and alternative text to the user. The output is the visually displayed message.

[0243] The above outlines the processing steps of a system that detects customer harassment risks in real time and provides appropriate feedback.

[0244] 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.

[0245] overview

[0246] This invention relates to a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it provides more accurate risk assessment and feedback. This system analyzes user emotions in real time and proposes appropriate measures to reduce the risk of customer harassment.

[0247] System Configuration

[0248] 1. User's terminal

[0249] It collects data from conversations, emails, and chats.

[0250] Includes speech recognition functionality to convert audio data into text data.

[0251] It communicates with the server and sends data.

[0252] 2. Server

[0253] Receive data sent from the device.

[0254] The system analyzes data in real time to determine the risk of customer harassment.

[0255] We use an emotion engine that recognizes user emotions to perform emotional analysis.

[0256] Based on the judgment result, a warning message and feedback are generated.

[0257] Generate appropriate alternative text and send it to the user's device.

[0258] Program processing

[0259] 1. Data Collection

[0260] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[0261] 2. Real-time analysis

[0262] The server inputs the received data into an analysis tool. The analysis tool includes sentiment analysis and specific keyword detection algorithms, and in addition, it operates an emotion engine that recognizes the user's emotions.

[0263] 3. Emotion recognition

[0264] The emotion engine analyzes the user's emotions from incoming data. For audio data, it analyzes tone and speed of voice; for text data, it analyzes context and word choice to determine whether the user is emotional.

[0265] 4. Risk Assessment

[0266] The server determines the risk level of customer harassment based on analysis results, including those from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[0267] 5. Generating warnings and feedback

[0268] If the risk level is determined to be high or medium, the server will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[0269] Specific example

[0270] Example: Detecting customer harassment via email

[0271] 1. The user composes an emotional email and clicks the send button.

[0272] Example: "Can't even meet such a simple request? Do you have any motivation?"

[0273] 2. The terminal sends the body of the email to the server.

[0274] The sent text: "Can't even meet such a simple request? Do you have any motivation?"

[0275] 3. The server analyzes the content of the email, and the sentiment engine recognizes the user's sentiment.

[0276] Sentiment analysis result: High level of anger

[0277] 4. Based on the overall analysis result, the server determines the risk of kasukhara.

[0278] Judgment result: High risk

[0279] 5. The server sends feedback to the user's terminal together with a warning message.

[0280] Warning message: "There is a suspicion of kasukhara. Risk level: High"

[0281] Feedback: "Such expressions should be avoided."

[0282] 6. The server generates an appropriate alternative sentence and provides it to the user.

[0283] Alternative proposal: "Was there anything unclear in this part? Please let me explain it in detail." <9000898> Example: Detection of kasukhara in chat

[0285] 1. The user enters a high-pressure message in the chat and clicks the send button.

[0286] Example: "Your response is too slow. What are you doing?"

[0287] 2. The terminal sends the chat content to the server.

[0288] Transmitted text: "Your response is too slow. What are you doing?"

[0289] 3. The server analyzes the chat content, and the emotion engine recognizes the user's emotion.

[0290] Emotion analysis result: High level of irritated emotion

[0291] 4. Based on the overall analysis result, the server evaluates the kas哈拉 risk.

[0292] Judgment result: Medium risk

[0293] 5. The server sends a warning and feedback to the user's terminal.

[0294] Warning message: "Attention is required. Risk level: Medium"

[0295] Feedback: "This expression may be impolite."

[0296] 6. The server generates an appropriate alternative sentence and provides it to the user.

[0297] Alternative proposal: "If a problem occurs, please respond promptly."

[0298] In this way, by combining the emotion engine, this system can accurately grasp the user's emotion, determine the kas哈拉 risk in real time, and provide appropriate feedback and alternative expressions. This can optimize the communication between the company and the customer and reduce the psychological burden of employees.

[0299] The following describes the processing flow.

[0300] Step 1:

[0301] The user starts an email, chat message, or conversation. The terminal prepares to collect this data in real time.

[0302] Step 2:

[0303] The user creates an email and clicks the send button. Or, the user enters a chat message and clicks the send button. In the case of a voice call, the conversation is started.

[0304] Step 3:

[0305] The terminal collects the body of the email, the chat message, or the voice data. The voice data is converted into text data using a voice recognition function.

[0306] Step 4:

[0307] The collected data (text and converted voice text) is sent from the terminal to the server. The server receives the data.

[0308] Step 5:

[0309] The server inputs the received data into an analysis tool. The analysis tool includes a sentiment analysis engine and a specific keyword detection algorithm.

[0310] Step 6:

[0311] The user's sentiment is analyzed by the sentiment engine. Specifically, in the case of voice data, the sentiment is recognized based on the tone and speed of the voice, and in the case of text data, based on the context and diction.

[0312] Step 7:

[0313] The server integrates analysis results, including those from the emotion engine, to determine the risk level of customer harassment. The risk level is evaluated in three stages: high, medium, and low.

[0314] Step 8:

[0315] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[0316] Step 9:

[0317] The server generates a warning message along with appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[0318] Step 10:

[0319] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[0320] Step 11:

[0321] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[0322] Step 12:

[0323] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[0324] As described above, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions. By combining it with an emotion engine, it can accurately grasp the user's emotions and enable more accurate risk assessment and countermeasures.

[0325] (Example 2)

[0326] 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".

[0327] There is a problem of increasing psychological burden on employees due to customer harassment. Furthermore, it is difficult to determine the risk of customer harassment in real time, making it difficult to take appropriate action and provide feedback. To solve this problem, a system is needed that analyzes user emotions in real time and proposes quick and appropriate countermeasures.

[0328] 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.

[0329] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for analyzing the user's emotions using an emotion engine; means for evaluating the risk level in three stages: high, medium, and low; and means for inputting the data into an analysis tool. This makes it possible to evaluate the risk of customer harassment in real time and provide rapid and appropriate feedback and alternative text.

[0330] "Conversation content" refers to all communication content in audio and text format, primarily including information in dialogue and chat formats.

[0331] "Email" refers to messages containing data such as text, images, and audio that are sent and received over the internet.

[0332] "Chat" refers to online conversational communication where text messages are exchanged in real time.

[0333] "Real-time data collection" means collecting data simultaneously with user actions and processing it without delay.

[0334] "Data analysis" refers to the techniques used to process and analyze collected information to identify useful insights and risks.

[0335] "Customer harassment" refers to acts that cause employees psychological distress due to excessive demands or intimidating behavior from customers.

[0336] "Risk assessment" refers to the process of evaluating whether a particular action or statement carries a risk of customer harassment through sentiment analysis and keyword detection.

[0337] A "warning" is a message from the system to alert the user, and may include feedback based on the risk level.

[0338] "Feedback" refers to evaluations and suggestions for improvement regarding a user's actions and comments.

[0339] "Alternative text" refers to improved versions of the original statement, suggested to avoid customer harassment.

[0340] An "emotion engine" refers to an algorithm or model used to analyze a user's emotional state, sensing emotions based on voice and text data.

[0341] "Risk level" refers to a three-tiered evaluation (high, medium, low) that indicates how high the risk of customer harassment is.

[0342] "Analysis tools" refer to software and algorithms used for data analysis and sentiment analysis.

[0343] Modes for carrying out the invention

[0344] System Overview

[0345] This invention is a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it can provide more accurate risk assessment and feedback. This system can analyze user emotions in real time and propose appropriate measures to reduce the risk of customer harassment.

[0346] System Configuration

[0347] 1. User's device:

[0348] This includes means of collecting data from conversations, emails, and chats.

[0349] Includes speech recognition means for converting audio data into text data.

[0350] Includes means for communicating with a server and transmitting data.

[0351] 2. Server:

[0352] Includes means for receiving data transmitted from a terminal.

[0353] This includes methods for analyzing data in real time and determining the risk of customer harassment.

[0354] This includes means for performing emotional analysis using an emotion engine that recognizes the user's emotions.

[0355] Includes means for generating warning messages and feedback based on the judgment result.

[0356] Includes means for generating appropriate alternative text and sending it to the user's terminal.

[0357] Specific hardware and software to be used

[0358] We will use "Google Cloud Speech-to-Text API" or "IBM Watson Speech to Text" as speech recognition software.

[0359] We use Google Cloud Natural Language API and IBM Watson Natural Language Understanding for sentiment analysis and text analysis.

[0360] Standard server and cloud computing environments are used for data processing and analysis.

[0361] Specific examples of operation

[0362] Example 1: Detecting customer harassment via email

[0363] 1. The user composes an emotional email and clicks the send button.

[0364] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0365] 2. The device sends the email body to the server.

[0366] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[0367] 3. The server sends the email content to the Google Cloud Natural Language API for sentiment analysis.

[0368] Emotional analysis results: High level of anger.

[0369] 4. The server determines the risk of customer harassment based on the overall analysis results.

[0370] Judgment result: High risk

[0371] 5. The server sends feedback to the user's terminal along with a warning message.

[0372] Warning message: "Suspected customer harassment. Risk level: High"

[0373] Feedback: "This kind of expression should be avoided."

[0374] 6. The server generates appropriate alternative text and provides it to the user.

[0375] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[0376] Example 2: Detection of customer harassment in chat

[0377] 1. The user types an aggressive message in the chat and clicks the send button.

[0378] Example: "Your response is too slow. What are you doing?"

[0379] 2. The device sends the chat content to the server.

[0380] Sent text: "Your response is too slow. What are you doing?"

[0381] 3. The server sends the chat content to "IBM Watson Natural Language Understanding" for sentiment analysis.

[0382] Emotional analysis results: High level of irritation.

[0383] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[0384] Assessment result: Medium risk

[0385] 5. The server sends a warning and feedback to the user's terminal.

[0386] Warning message: "Caution is advised. Risk level: Medium"

[0387] Feedback: "This expression may be considered rude."

[0388] 6. The server generates appropriate alternative text and provides it to the user.

[0389] Alternative suggestion: "Please respond promptly if any problems arise."

[0390] Example of a prompt

[0391] Examples of prompt statements to input into the generative AI model are as follows:

[0392] User-generated email: "You can't even fulfill such a simple request? Do you even have any motivation?"

[0393] Please suggest the feedback and alternative sentences that the system will generate.

[0394] conclusion

[0395] This system optimizes communication between companies and customers, reduces the psychological burden on employees, and effectively manages the potential risk of customer harassment.

[0396] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0397] Step 1: Data Collection

[0398] The user composes an email or chat message and clicks the send button.

[0399] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0400] The device collects user input data and extracts email content and chat messages. If audio data is present, it is converted into text using speech recognition technology.

[0401] Input: User's email or chat message

[0402] Output: Text data ("You can't even meet such a simple request? Do you even have any motivation?")

[0403] Step 2: Data transmission

[0404] The device sends the collected data to the server.

[0405] Input: Text data

[0406] Output: Sending data to the server

[0407] Step 3: Data Reception

[0408] The server receives data sent from the terminal.

[0409] Input: Text data from the terminal

[0410] Output: Received data

[0411] Step 4: Emotion Analysis

[0412] The server inputs the received data into an analysis tool. The analysis tool uses either "Google Cloud Natural Language API" or "IBM Watson Natural Language Understanding".

[0413] Input: Received data (text data)

[0414] Data processing: An emotion engine analyzes emotions from context and word choice.

[0415] Output: Emotion analysis results (e.g., high level of anger)

[0416] Step 5: Risk Assessment

[0417] The server determines the risk level of customer harassment based on the analysis results from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[0418] Input: Sentiment analysis results

[0419] Data processing: Assess risk based on the intensity of emotions and specific keywords.

[0420] Output: Risk level (high, medium, low)

[0421] Step 6: Generate warning message

[0422] If the server is determined to have a high or medium risk level, it will generate a warning message and feedback.

[0423] Input: Risk level

[0424] Data processing: Generate warning messages and feedback based on risk level.

[0425] Output: Warning messages and feedback

[0426] Step 7: Send a warning message

[0427] The server sends the generated warning message and feedback to the user's terminal.

[0428] Input: Warning messages and feedback

[0429] Output: Sent to the user's terminal

[0430] Step 8: Generate alternative text

[0431] The server generates appropriate alternative text and provides it to the user.

[0432] Input: Risk level and sentiment analysis results

[0433] Data processing: Generate appropriate and gentle alternative text based on sentiment analysis results.

[0434] Output: Alternative text suggestions

[0435] Step 9: Send alternative text

[0436] The server sends the generated alternative text to the user's terminal.

[0437] Input: Alternative text

[0438] Output: Sent to the user's terminal

[0439] Specific example:

[0440] When a user sends a message via chat saying, "Your response is too slow. What are you doing?",

[0441] 1. The device collects messages and sends them to the server.

[0442] 2. The server receives the data, performs sentiment analysis, and determines that the level of irritation is high.

[0443] 3. The server assesses the risk as moderate and generates a warning message and feedback indicating that attention is needed.

[0444] 4. The server sends a warning message ("Caution is advised. Risk level: Medium") and feedback ("This language may be offensive.") to the user.

[0445] 5. The server generates alternative text ("Please respond promptly if a problem occurs.") and provides it to the user.

[0446] In this way, the system assesses the risk of customer harassment at each step and provides appropriate responses and feedback.

[0447] (Application Example 2)

[0448] 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 device 14 will be referred to as the "terminal."

[0449] In modern businesses, customer harassment is on the rise during customer service interactions, negatively impacting employee psychological burden and overall business efficiency. To prevent such harassment, a system is needed that monitors customer interactions in real time and takes appropriate action. However, conventional systems struggle to accurately grasp user emotions and provide instant feedback, and they also lack sufficient suggestions for appropriate alternative expressions to mitigate risks. As a result, effective harassment prevention remains difficult, and more sophisticated systems are required.

[0450] 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.

[0451] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for providing feedback on the warnings and generating appropriate alternative text; means for analyzing the user's emotions and recognizing their emotional state; and means for sending improvement suggestions when the risk of customer harassment exceeds a certain level. This enables real-time determination of the risk of customer harassment, prompt warnings to employees, appropriate feedback, and suggestions for alternative expressions.

[0452] "Conversation content" refers to the words and expressions used when exchanging information in audio or text format.

[0453] "Email" refers to messages that are sent and received electronically via networks such as the internet.

[0454] "Chat" refers to real-time, text-based communication.

[0455] "Means of real-time collection" refers to devices or programs that have the function of collecting conversation content and text messages and sending them to a server.

[0456] "Means for analyzing collected data" refers to a system that has the function of analyzing collected information and detecting specific patterns or keywords.

[0457] "Means for determining the risk of customer harassment" refers to a system that has processes and technologies for evaluating the possibility of harassment in interactions with customers.

[0458] "Means of issuing warnings" refers to a system that has the function of sending alerts and warnings to users based on the results of risk assessment.

[0459] "Means of generating feedback and appropriate alternative text for warnings" refers to a system that has the function of suggesting directions for improvement or appropriate wording to the user.

[0460] "Means for analyzing user emotions and recognizing emotional states" refers to a system that possesses technologies and processes for performing emotion analysis on collected data and identifying the user's emotional state.

[0461] "Emotional state" refers to the user's psychological state, emotions, and mood.

[0462] "A means of sending improvement suggestions when the risk of customer harassment exceeds a certain level" refers to a system that has the technology or process to notify users of specific countermeasures or alternative expressions when the risk assessment is high.

[0463] System Overview

[0464] This invention provides a system that collects conversation content, emails, or chats in real time, determines the risk of customer harassment, and provides warnings and feedback based on the results. Furthermore, by recognizing the user's emotions, it enables more accurate risk assessment and generation of appropriate alternative text. The aim of this system is to reduce the risk of harassment in everyday business communications such as conversations and chats, and to reduce the psychological burden on companies and employees.

[0465] Hardware and software to be used

[0466] Hardware:

[0467] Smartphone: Used for collecting and sending conversations, chats, and emails.

[0468] Server: Used for data analysis, risk assessment, and feedback generation.

[0469] Robots: They can also be used to collect and transmit communications in security operations.

[0470] software:

[0471] Speech Recognition: Google Cloud Speech-to-Text

[0472] Sentiment Analysis: IBM Watson Natural Language Understanding, Amazon Comprehend Sentiment Analysis

[0473] Cloud services: AWS®, Google Cloud Platform

[0474] System operation

[0475] Data collection:

[0476] When users engage in conversations or chats using their smartphones or robots, the data is collected in real time. The voice data is converted into text data using speech recognition technology and sent to a server.

[0477] Real-time analysis:

[0478] The server inputs the collected text data into a sentiment analysis tool, checking for specific keywords and contexts in real time. This helps identify the risk of customer harassment.

[0479] Emotion recognition:

[0480] The emotion engine is used to analyze the user's emotional state. For audio data, it analyzes the tone and speed of the voice; for text data, it analyzes the structure and word choice of the sentences to identify the user's emotional state.

[0481] Risk assessment:

[0482] Based on the analysis results from the emotion engine, the server determines the risk of customer harassment in three stages: high, medium, and low. If the risk level is determined to be medium or high, immediate action is taken.

[0483] Generating warnings and feedback:

[0484] Depending on the risk level, the server sends a warning message to the user's terminal. The warning includes the risk level and specific improvement suggestions, and appropriate alternative text is also generated and provided to the user.

[0485] Specific example

[0486] Preventing customer harassment in call centers

[0487] 1. When a user (employee) is providing chat support at a call center, they receive emotional messages.

[0488] Example: "Why are you taking so long to respond? Hurry up!"

[0489] 2. The employee's device sends the chat content to the server.

[0490] Sent text: "Why are you taking so long to respond? Hurry up!"

[0491] 3. The server analyzes the chat content, and the emotion engine recognizes the recipient's (customer's) emotions.

[0492] Emotional analysis results: High level of anger.

[0493] 4. The server determines the risk of customer harassment.

[0494] Judgment result: High risk

[0495] 5. The server sends a warning message and feedback to the employee's terminal.

[0496] Warning message: "There is a high risk. Please consider how to mitigate customer anger."

[0497] Feedback: "Explain the situation carefully and patiently until the customer is satisfied."

[0498] 6. The server generates an appropriate alternative sentence and provides it to the employee.

[0499] Alternative suggestion: "We apologize for the delay. We are currently working to process your request as quickly as possible, so please wait a little longer."

[0500] Examples of prompts to input into a generative AI model

[0501] "In customer service, I received a message that conveyed anger. What would be an appropriate response? For example, please suggest a response to the following message."

[0502] In this way, this system can significantly reduce the risk of customer harassment by assessing risks in real time and prompting users to take appropriate action immediately.

[0503] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0504] Step 1: Data Collection

[0505] Input: Data from conversations, emails, and chats that users have with smartphones or robots.

[0506] Specific operation: The user initiates communication, and voice and text data are generated.

[0507] Data processing: The audio data is converted to text data using Google Cloud Speech-to-Text. The text data is then sent directly to the server.

[0508] Output: Text-formatted conversation data, chat data, and email data are sent to the server.

[0509] Step 2: Forecast Analysis

[0510] Input: Text data collected in Step 1.

[0511] Specific operation: The server inputs the collected data into IBM Watson Natural Language Understanding or Amazon Comprehend Sentiment Analysis for real-time analysis.

[0512] Data processing: This involves detecting keywords within text data and performing overall contextual analysis. It also uses an emotion engine to analyze the user's emotional state.

[0513] Output: Sentiment analysis results and a list of specific keywords.

[0514] Step 3: Emotion Recognition

[0515] Input: Sentiment analysis results and specific keyword list analyzed in Step 2.

[0516] Specific operation: The server uses the results of the emotion engine to determine the user's emotional state. In particular, it analyzes the intensity and trend of emotions.

[0517] Data processing: Based on the emotion analysis results, the user's emotional state is classified into categories such as "anger," "sadness," and "joy."

[0518] Output: Classified emotion state data.

[0519] Step 4: Risk Assessment

[0520] Input: Emotional state data and specific keyword list obtained in Step 3.

[0521] Specific operation: The server determines the risk of customer harassment based on the collected data. The risk level is evaluated in three stages: "high," "medium," and "low."

[0522] Data calculation: Calculate harassment risk based on the intensity of emotions and the frequency of specific keywords.

[0523] Output: Risk assessment result (high, medium, low).

[0524] Step 5: Generating warnings and feedback

[0525] Input: Risk assessment results (high, medium, low) and emotional state data.

[0526] Specific operation: Based on the determination result, the server sends a warning message to the user's terminal.

[0527] Data processing: Warning messages will include risk levels along with specific improvement suggestions. Additionally, a generative AI model will be used to generate appropriate alternative text.

[0528] Output: Warning message, improvement suggestion, and appropriate alternative text.

[0529] Step 6: Suggesting alternative text

[0530] Input: Improvement suggestions and alternative text.

[0531] Specific operation: The server uses a generative AI model to provide the user with appropriate alternative text.

[0532] Output: The provided alternative text will be displayed on the user's terminal.

[0533] This allows the entire system to assess the risk of customer harassment in real time and promptly prompt users to take appropriate action.

[0534] 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.

[0535] Data generation model 58 is a type of 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.

[0536] 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.

[0537] [Second Embodiment]

[0538] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0539] 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.

[0540] 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).

[0541] 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.

[0542] 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.

[0543] 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).

[0544] 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.

[0545] 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.

[0546] 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.

[0547] 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.

[0548] 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.

[0549] 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".

[0550] overview

[0551] This invention is a system that collects conversation content, emails, or chats in real time, analyzes the risk of customer harassment, and issues warnings. Furthermore, it is characterized by generating appropriate feedback and alternative text along with the warning and providing them to the user.

[0552] System Configuration

[0553] 1. User's terminal

[0554] It collects data from conversations, emails, and chats.

[0555] Includes speech recognition functionality for converting audio data into text data.

[0556] It communicates with the server and sends data.

[0557] 2. Server

[0558] Receive data sent from the device.

[0559] The system analyzes data in real time to determine the risk of customer harassment.

[0560] Based on the judgment, a warning message and feedback are generated.

[0561] Generate appropriate alternative text and send it to the user's device.

[0562] Program processing

[0563] 1. Data Collection

[0564] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[0565] 2. Real-time analysis

[0566] The server analyzes the received text data and assesses the risk of customer harassment. This assessment includes sentiment analysis and detection of specific keywords. Based on the analysis results, a risk level is determined.

[0567] 3. Generating warnings and feedback

[0568] If the server determines that there is a high risk of customer harassment, it will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[0569] Specific example

[0570] Example: Detecting customer harassment via email

[0571] 1. The user composes an emotional email and clicks the send button.

[0572] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0573] 2. The device sends the email body to the server.

[0574] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[0575] 3. The server analyzes the content of the email and determines the risk of customer harassment.

[0576] Judgment result: High risk

[0577] 4. The server sends feedback to the user's terminal along with a warning message.

[0578] Warning message: "Suspected customer harassment. Risk level: High"

[0579] Feedback: "This kind of expression should be avoided."

[0580] 5. The server generates an appropriate alternative sentence and provides it to the user.

[0581] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[0582] Example: Detecting customer harassment in chat

[0583] 1. The user types an aggressive message in the chat and clicks the send button.

[0584] Example: "Your response is too slow. What are you doing?"

[0585] 2. The device sends the chat content to the server.

[0586] Sent text: "Your response is too slow. What are you doing?"

[0587] 3. The server analyzes the chat content and assesses the risk of customer harassment.

[0588] Assessment result: Medium risk

[0589] 4. The server sends a warning and feedback to the user's terminal.

[0590] Warning message: "Caution is advised. Risk level: Medium"

[0591] Feedback: "This expression may be considered rude."

[0592] 5. The server generates an appropriate alternative sentence and provides it to the user.

[0593] Alternative suggestion: "Please respond promptly if any problems arise."

[0594] In this way, this system can assess the risk of customer harassment in real time and provide appropriate feedback and alternative text, thereby promoting proper communication between companies and customers and reducing the psychological burden on employees.

[0595] The following describes the processing flow.

[0596] Step 1:

[0597] The user begins typing an email, chat message, or voice call. The device prepares to collect this data.

[0598] Step 2:

[0599] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[0600] Step 3:

[0601] The device collects email body text, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[0602] Step 4:

[0603] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[0604] Step 5:

[0605] The server inputs the received data into an analysis tool. This analysis tool includes sentiment analysis and specific keyword detection algorithms.

[0606] Step 6:

[0607] The server uses analysis tools to analyze incoming data in real time. As a result, the risk level of customer harassment is determined. The risk level is evaluated in three stages: high, medium, and low.

[0608] Step 7:

[0609] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[0610] Step 8:

[0611] The server generates appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[0612] Step 9:

[0613] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[0614] Step 10:

[0615] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[0616] Step 11:

[0617] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[0618] In this way, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions.

[0619] (Example 1)

[0620] 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."

[0621] In today's business environment, communication between companies and customers takes place through a variety of means. However, excessive demands and inappropriate remarks from customers, i.e., customer harassment, can increase the psychological burden on employees and lead to a deterioration of the working environment. Therefore, it is necessary to monitor conversation content in real time, detect the risk of customer harassment early, and take appropriate measures. However, current systems have problems such as delays in analyzing collected data and inability to accurately assess risks.

[0622] 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.

[0623] In this invention, the server includes means for collecting conversation content, electronic messages, or instant messages in real time; means for analyzing the collected data and evaluating the risk of customer harassment; and means for issuing warnings based on the evaluation results. This makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[0624] "Conversation content" refers to the content of communication conducted through voice.

[0625] "Electronic messages" refer to messages sent and received in digital format, such as emails and text messages.

[0626] "Instant messaging" refers to real-time chats and instant messages.

[0627] "Methods for collecting information in real time" refers to technologies for collecting conversation content and messages almost instantly.

[0628] "Analysis" refers to the process of evaluating collected data using mechanical and algorithmic methods to extract meaning and emotion.

[0629] "Customer harassment" refers to acts that cause psychological distress to employees through excessive demands or inappropriate remarks from customers.

[0630] "Means of risk assessment" refers to technologies used to determine the risk level of customer harassment through data analysis.

[0631] "Warning mechanisms" refer to technologies used to send warning messages to users when there is a risk.

[0632] "Speech recognition means" refers to technology for converting speech data into text data.

[0633] A "machine intelligence model" refers to a technology that uses artificial intelligence to analyze data and recognize specific patterns or emotions.

[0634] "Alternative text" refers to appropriate wording that is recommended to be used instead of the original message.

[0635] "Feedback" refers to suggestions, proposals, or specific advice regarding a warning.

[0636] The system in this invention is built to collect conversation content, electronic messages, or instant messages in real time, determine the risk of customer harassment, and issue warnings. This system includes a user terminal, a server, and a generative AI model.

[0637] 1. User terminal

[0638] The user terminal is a device that collects data from conversations, electronic messages, or instant messages. It also has a speech recognition system that converts voice data into text data. Speech recognition software such as "Google Cloud Speech-to-Text" is used for speech recognition. It also has the function to collect text data entered by the user and send it to the server.

[0639] Specific example: A user types the message "You can't even fulfill such a simple request? Do you even have any motivation?" into the chat window and clicks the send button.

[0640] 2. Server

[0641] The server is a central processing unit that analyzes text data received from user terminals. It analyzes the received data in real time and assesses the risk of customer harassment. The analysis uses natural language processing (NLP) technologies such as "SpaCy" and "IBM Watson" for sentiment analysis and detection of specific keywords. Based on the analysis results, it determines the risk level and generates warning messages and feedback as needed. Furthermore, it uses generative AI models (e.g., OpenAI's GPT-3) to generate appropriate alternative text.

[0642] Specific example: The server receives a message saying, "You can't even fulfill such a simple request? Are you even motivated?" and the analysis results indicate it's "high risk." Subsequently, it generates a warning message, "Suspected customer harassment. Risk level: high," and appropriate feedback, "Such language should be avoided," and creates alternative text, "Is there anything you didn't understand in this section? I'd be happy to explain in more detail."

[0643] Example of a prompt

[0644] A user is attempting to send the following message. Please check if this message poses a risk of customer harassment and generate a warning message and appropriate alternative text. Original message: "You can't even fulfill such a simple request? Are you even motivated?"

[0645] The above describes the embodiments for carrying out the present invention. This system makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[0646] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0647] Step 1:

[0648] The user creates a conversation, e-message, or instant message and clicks the send button. The user's device then collects the created message.

[0649] Input: User-entered conversation, e-message, or instant message (e.g., "You can't even fulfill such a simple request? Are you even motivated?")

[0650] Output: Message data stored on the user's terminal

[0651] Specific action: The user types a message in the chat window and clicks the send button.

[0652] Step 2:

[0653] The user terminal prepares to send the collected message data to the server in real time. If voice data is included, speech recognition is used to convert the voice to text.

[0654] Input: Voice data or text data

[0655] Output: Text data

[0656] Specific operation: A speech recognition engine (e.g., Google Cloud Speech-to-Text) converts speech to text. If the data is text, it is used as is.

[0657] Step 3:

[0658] The user terminal sends text data to the server.

[0659] Input: Text data (Example: "You can't even meet such a simple request? Do you even have any motivation?")

[0660] Output: Text data sent to the server

[0661] Specific operation: The user terminal uses its communication function to send the converted text data to the server.

[0662] Step 4:

[0663] The server analyzes the received text data. The server uses natural language processing technologies (e.g., SpaCy, IBM Watson) to perform sentiment analysis and detect specific keywords.

[0664] Input: Text data

[0665] Output: Analysis results (sentiment score, keyword detection results)

[0666] Specific operation: The server uses SpaCy or IBM Watson to analyze the sentiment of incoming messages and detect specific keywords.

[0667] Step 5:

[0668] The server assesses the risk of customer harassment based on the analysis results. This assessment includes sentiment scores and keyword detection results.

[0669] Input: Analysis results (sentiment score, keyword detection results)

[0670] Output: Risk assessment results (high risk, medium risk, low risk, etc.)

[0671] Specific operation: The server evaluates the risk level of customer harassment based on sentiment scores and keyword detection results (e.g., determines it as "high risk").

[0672] Step 6:

[0673] The server generates a warning message based on the risk assessment results. If the risk is high, it also generates specific improvement suggestions and alternative text. A generative AI model is used to create appropriate feedback and alternative text.

[0674] Input: Risk assessment results

[0675] Output: Warning message, feedback, alternative text

[0676] Specific actions: The server generates a warning message "Suspected customer harassment. Risk level: High," feedback "Such language should be avoided," and alternative text "Is there anything you don't understand about this part? We'd be happy to explain in more detail."

[0677] Step 7:

[0678] The server sends generated warning messages, feedback, and alternative text to the user's terminal.

[0679] Input: Warning message, feedback, alternative text

[0680] Output: Warning messages, feedback, and alternative text sent to the user's terminal.

[0681] Specific operation: The server uses its communication function to send warning messages, feedback, and alternative text to the user's terminal.

[0682] Through the above processing steps, this system can assess the risk of customer harassment in real time and immediately provide necessary countermeasures.

[0683] (Application Example 1)

[0684] 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 glasses 214 will be referred to as the "terminal."

[0685] Customer support and call center staff are required to detect the risk of customer harassment in real time during customer conversations and to take appropriate action quickly based on that risk. Traditional methods require staff to assess the risk and devise countermeasures themselves, which places a significant psychological burden on them and carries the risk of incorrect responses. This challenge needs to be addressed.

[0686] 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.

[0687] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing a warning based on the determination result; means for providing feedback on the warning and generating appropriate alternative text; means for collecting conversation content using speech recognition means that convert speech to text in real time; means for determining the risk of customer harassment using a device that performs sentiment analysis; means for determining the risk level based on the determination result and presenting an appropriate alert; means for displaying feedback on a smart device worn by an employee; and natural language processing means for generating appropriate alternative text. This enables staff to detect the risk of customer harassment in real time and take appropriate action quickly.

[0688] "Conversation content" refers to voice-based communication between customer support or call center staff and customers.

[0689] "Email" refers to messages sent and received electronically via the internet.

[0690] "Chat" refers to real-time text-based communication.

[0691] "Means of real-time collection" refers to methods or devices for instantly acquiring this data as conversations, emails, or chats occur.

[0692] "Means of analyzing collected data" refers to methods or devices for analyzing collected conversations, emails, or chats to assess the risk of customer harassment.

[0693] "Means for determining the risk of customer harassment" refers to a method or device that detects inappropriate words or actions from customers and determines their risk level based on collected and analyzed data.

[0694] "Means of issuing warnings" refers to methods or devices for sending appropriate notifications or warning messages to users when a risk of customer harassment is identified.

[0695] "Feedback" or "means for generating appropriate alternative text" refers to a method or device for generating and proposing appropriate countermeasures or alternative expressions to the user in response to the risk of customer harassment.

[0696] "Speech recognition means" refers to technologies and devices for converting speech data into text data.

[0697] "An emotional analysis device" refers to a method or device for analyzing the emotional aspects of text data and detecting specific emotions or tones.

[0698] "Means for determining risk levels" refers to methods or devices for evaluating the degree of customer harassment risk and quantifying that level based on the results of sentiment analysis.

[0699] "Means of providing appropriate alerts" refers to methods or devices for providing warnings or notifications to employees based on the determined risk level.

[0700] "Means of displaying feedback on smart devices" refers to methods or devices for displaying risk levels or feedback messages on wearable devices or other smart devices worn by users.

[0701] "Natural language processing means" refers to language processing technologies and devices that understand text data and generate appropriate alternative text.

[0702] This invention provides a system for detecting the risk of customer harassment in real time in customer support and call centers, and for taking appropriate action quickly based on that risk. The system includes user terminals, servers, and smart devices worn by employees.

[0703] System Configuration

[0704] 1. User's terminal

[0705] It has the ability to collect data from conversations, emails, and chats.

[0706] It is equipped with speech recognition means for converting audio data into text data.

[0707] It has the function of communicating with a server and sending data.

[0708] 2. Server

[0709] It has the ability to receive data sent from a device and analyze it in real time.

[0710] This includes an emotion analysis device for analyzing data and determining the risk of customer harassment, as well as an artificial intelligence model for detecting specific keywords.

[0711] Based on the assessment results, the risk level is determined, and warning and feedback messages are generated.

[0712] The generated warnings and feedback are sent to the user's device and to smart devices that display the feedback.

[0713] 3. Smart devices

[0714] Employees wear the device, which displays warning and feedback messages sent from the server.

[0715] Hardware and software to be used

[0716] Speech recognition method: The "speech_recognition" library and Google's speech recognition API are used.

[0717] Sentiment analysis device: Uses a pre-trained sentiment classification model (such as BERT) using the "transformers" library.

[0718] Smart devices: Employees can wear smart glasses or use smartphones.

[0719] Servers: High-performance servers are used for processing and communicating this data.

[0720] Processing flow

[0721] 1. Data collection:

[0722] When a user initiates a conversation or chat with a customer, the device collects conversation and chat data in real time, and the voice data is converted into text data using speech recognition technology.

[0723] 2. Real-time analysis:

[0724] The collected text data is sent to a server, which uses sentiment analysis devices and artificial intelligence models to analyze the risk of customer harassment.

[0725] 3. Generating warnings and feedback:

[0726] If the risk level of customer harassment is high, the server will generate a warning and specific feedback message, or appropriate alternative text, depending on the risk level.

[0727] 4. Notifications and Feedback:

[0728] The generated warning and feedback messages are sent to and displayed on the user's device and the employee's smart device, allowing employees to take appropriate action immediately.

[0729] Specific example

[0730] For example, if a support staff member is talking to a customer and the customer makes an aggressive statement such as, "Your response is too slow! What's going on?", this system collects the statement in real time and converts it to text using speech recognition. Then, an emotion analyzer analyzes this text, and if it is determined to be high risk, the server generates a warning message and feedback such as, "Specifically, what is the problem? We'd be happy to help," which is displayed on the staff member's smart glasses.

[0731] Prompts for Generative AI Models

[0732] Design a custom-made smart glasses application to detect customer harassment during customer support conversations and provide appropriate alerts and feedback in real time. Generate Python pseudocode that meets the following requirements.

[0733] 1. Collection of audio data

[0734] 2. Text conversion using speech recognition

[0735] 3. Risk assessment using sentiment analysis

[0736] 4. Generate feedback based on the judgment result.

[0737] The above describes the embodiments for carrying out the present invention. By introducing this system, it becomes possible to detect customer harassment in real time at customer support sites and take prompt and appropriate action, thereby reducing the psychological burden on employees.

[0738] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0739] Step 1:

[0740] The user initiates a conversation with a customer. The input is the voice conversation with the customer. The device collects this voice data. The device captures the voice data in real time using its built-in microphone.

[0741] Step 2:

[0742] The device converts the collected audio data into text data. The input is the audio data collected in step 1. The device uses its speech recognition capabilities (using the speech_recognition library and Google's speech recognition API) to convert the audio data into text data. The output is text data.

[0743] Step 3:

[0744] The text data is sent to the server. The input is the text data generated in step 2. The terminal communicates with the server and sends the text data to the server. The output is that the text data is sent to the server.

[0745] Step 4:

[0746] The server analyzes the received text data. The input is the text data sent to the server in step 3. The server uses a sentiment analyzer (using the transformers library) and an artificial intelligence model that detects specific keywords to perform sentiment analysis and keyword detection on the text data. The output is the risk assessment result.

[0747] Step 5:

[0748] The server determines the risk level based on the risk assessment results and generates warnings and feedback. The input is the risk assessment results obtained in step 4. The server determines the risk level (high, medium, low) according to the assessment results and generates appropriate warning messages and feedback messages, as well as appropriate alternative text using natural language processing. The output is the generated warning messages and feedback messages.

[0749] Step 6:

[0750] The server sends the generated warning and feedback messages to the user's terminal and smart device. The input is the warning and feedback messages generated in step 5. The server sends these messages to the terminal and smart device via communication. The output is the warning and feedback displayed on the user's terminal and smart device.

[0751] Step 7:

[0752] The user's terminal and smart device display warnings and feedback. The input consists of the warning and feedback messages sent from the server in step 6. The terminal and smart device use their display functions to present the risk level, warnings, feedback, and alternative text to the user. The output is the visually displayed message.

[0753] The above outlines the processing steps of a system that detects customer harassment risks in real time and provides appropriate feedback.

[0754] 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.

[0755] overview

[0756] This invention relates to a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it provides more accurate risk assessment and feedback. This system analyzes user emotions in real time and proposes appropriate measures to reduce the risk of customer harassment.

[0757] System Configuration

[0758] 1. User's terminal

[0759] It collects data from conversations, emails, and chats.

[0760] Includes speech recognition functionality to convert audio data into text data.

[0761] It communicates with the server and sends data.

[0762] 2. Server

[0763] Receive data sent from the device.

[0764] The system analyzes data in real time to determine the risk of customer harassment.

[0765] We use an emotion engine that recognizes user emotions to perform emotional analysis.

[0766] Based on the judgment result, a warning message and feedback are generated.

[0767] Generate appropriate alternative text and send it to the user's device.

[0768] Program processing

[0769] 1. Data Collection

[0770] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[0771] 2. Real-time analysis

[0772] The server inputs the received data into an analysis tool. The analysis tool includes sentiment analysis and specific keyword detection algorithms, and in addition, it operates an emotion engine that recognizes the user's emotions.

[0773] 3. Emotion recognition

[0774] The emotion engine analyzes the user's emotions from incoming data. For audio data, it analyzes tone and speed of voice; for text data, it analyzes context and word choice to determine whether the user is emotional.

[0775] 4. Risk Assessment

[0776] The server determines the risk level of customer harassment based on analysis results, including those from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[0777] 5. Generating warnings and feedback

[0778] If the risk level is determined to be high or medium, the server will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[0779] Specific example

[0780] Example: Detecting customer harassment via email

[0781] 1. The user composes an emotional email and clicks the send button.

[0782] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0783] 2. The device sends the email body to the server.

[0784] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[0785] 3. The server analyzes the email content, and the emotion engine recognizes the user's emotions.

[0786] Emotional analysis results: High level of anger.

[0787] 4. The server determines the risk of customer harassment based on the overall analysis results.

[0788] Judgment result: High risk

[0789] 5. The server sends feedback to the user's terminal along with a warning message.

[0790] Warning message: "Suspected customer harassment. Risk level: High"

[0791] Feedback: "This kind of expression should be avoided."

[0792] 6. The server generates an appropriate alternative sentence and provides it to the user.

[0793] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[0794] Example: Detecting customer harassment in chat

[0795] 1. The user types an aggressive message in the chat and clicks the send button.

[0796] Example: "Your response is too slow. What are you doing?"

[0797] 2. The device sends the chat content to the server.

[0798] Sent text: "Your response is too slow. What are you doing?"

[0799] 3. The server analyzes the chat content, and the emotion engine recognizes the user's emotions.

[0800] Emotional analysis results: High level of irritation.

[0801] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[0802] Assessment result: Medium risk

[0803] 5. The server sends a warning and feedback to the user's terminal.

[0804] Warning message: "Caution is advised. Risk level: Medium"

[0805] Feedback: "This expression may be considered rude."

[0806] 6. The server generates an appropriate alternative sentence and provides it to the user.

[0807] Alternative suggestion: "Please respond promptly if any problems arise."

[0808] In this way, by combining an emotion engine, this system accurately grasps the user's emotions, determines the risk of customer harassment in real time, and provides appropriate feedback and alternative expressions. This optimizes communication between companies and customers and reduces the psychological burden on employees.

[0809] The following describes the processing flow.

[0810] Step 1:

[0811] The user initiates an email, chat message, or conversation. The device prepares to collect this data in real time.

[0812] Step 2:

[0813] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[0814] Step 3:

[0815] The device collects email content, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[0816] Step 4:

[0817] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[0818] Step 5:

[0819] The server inputs the received data into the analysis tool. The analysis tool includes a sentiment analysis engine and a specific keyword detection algorithm.

[0820] Step 6:

[0821] The emotion engine analyzes the user's emotions. Specifically, it recognizes emotions based on tone and speed of voice in the case of audio data, and on context and word choice in the case of text data.

[0822] Step 7:

[0823] The server integrates analysis results, including those from the emotion engine, to determine the risk level of customer harassment. The risk level is evaluated in three stages: high, medium, and low.

[0824] Step 8:

[0825] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[0826] Step 9:

[0827] The server generates a warning message along with appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[0828] Step 10:

[0829] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[0830] Step 11:

[0831] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[0832] Step 12:

[0833] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[0834] As described above, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions. By combining it with an emotion engine, it can accurately grasp the user's emotions and enable more accurate risk assessment and countermeasures.

[0835] (Example 2)

[0836] 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".

[0837] There is a problem of increasing psychological burden on employees due to customer harassment. Furthermore, it is difficult to determine the risk of customer harassment in real time, making it difficult to take appropriate action and provide feedback. To solve this problem, a system is needed that analyzes user emotions in real time and proposes quick and appropriate countermeasures.

[0838] 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.

[0839] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for analyzing the user's emotions using an emotion engine; means for evaluating the risk level in three stages: high, medium, and low; and means for inputting the data into an analysis tool. This makes it possible to evaluate the risk of customer harassment in real time and provide rapid and appropriate feedback and alternative text.

[0840] "Conversation content" refers to all communication content in audio and text format, primarily including information in dialogue and chat formats.

[0841] "Email" refers to messages containing data such as text, images, and audio that are sent and received over the internet.

[0842] "Chat" refers to online conversational communication where text messages are exchanged in real time.

[0843] "Real-time data collection" means collecting data simultaneously with user actions and processing it without delay.

[0844] "Data analysis" refers to the techniques used to process and analyze collected information to identify useful insights and risks.

[0845] "Customer harassment" refers to acts that cause employees psychological distress due to excessive demands or intimidating behavior from customers.

[0846] "Risk assessment" refers to the process of evaluating whether a particular action or statement carries a risk of customer harassment through sentiment analysis and keyword detection.

[0847] A "warning" is a message from the system to alert the user, and may include feedback based on the risk level.

[0848] "Feedback" refers to evaluations and suggestions for improvement regarding a user's actions and comments.

[0849] "Alternative text" refers to improved versions of the original statement, suggested to avoid customer harassment.

[0850] An "emotion engine" refers to an algorithm or model used to analyze a user's emotional state, sensing emotions based on voice and text data.

[0851] "Risk level" refers to a three-tiered evaluation (high, medium, low) that indicates how high the risk of customer harassment is.

[0852] "Analysis tools" refer to software and algorithms used for data analysis and sentiment analysis.

[0853] Modes for carrying out the invention

[0854] System Overview

[0855] This invention is a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it can provide more accurate risk assessment and feedback. This system can analyze user emotions in real time and propose appropriate measures to reduce the risk of customer harassment.

[0856] System Configuration

[0857] 1. User's device:

[0858] This includes means of collecting data from conversations, emails, and chats.

[0859] Includes speech recognition means for converting audio data into text data.

[0860] Includes means for communicating with a server and transmitting data.

[0861] 2. Server:

[0862] Includes means for receiving data transmitted from a terminal.

[0863] This includes methods for analyzing data in real time and determining the risk of customer harassment.

[0864] This includes means for performing emotional analysis using an emotion engine that recognizes the user's emotions.

[0865] Includes means for generating warning messages and feedback based on the judgment result.

[0866] Includes means for generating appropriate alternative text and sending it to the user's terminal.

[0867] Specific hardware and software to be used

[0868] We will use "Google Cloud Speech-to-Text API" or "IBM Watson Speech to Text" as speech recognition software.

[0869] We use Google Cloud Natural Language API and IBM Watson Natural Language Understanding for sentiment analysis and text analysis.

[0870] Standard server and cloud computing environments are used for data processing and analysis.

[0871] Specific examples of operation

[0872] Example 1: Detecting customer harassment via email

[0873] 1. The user composes an emotional email and clicks the send button.

[0874] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0875] 2. The device sends the email body to the server.

[0876] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[0877] 3. The server sends the email content to the Google Cloud Natural Language API for sentiment analysis.

[0878] Emotional analysis results: High level of anger.

[0879] 4. The server determines the risk of customer harassment based on the overall analysis results.

[0880] Judgment result: High risk

[0881] 5. The server sends feedback to the user's terminal along with a warning message.

[0882] Warning message: "Suspected customer harassment. Risk level: High"

[0883] Feedback: "This kind of expression should be avoided."

[0884] 6. The server generates appropriate alternative text and provides it to the user.

[0885] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[0886] Example 2: Detection of customer harassment in chat

[0887] 1. The user types an aggressive message in the chat and clicks the send button.

[0888] Example: "Your response is too slow. What are you doing?"

[0889] 2. The device sends the chat content to the server.

[0890] Sent text: "Your response is too slow. What are you doing?"

[0891] 3. The server sends the chat content to "IBM Watson Natural Language Understanding" for sentiment analysis.

[0892] Emotional analysis results: High level of irritation.

[0893] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[0894] Assessment result: Medium risk

[0895] 5. The server sends a warning and feedback to the user's terminal.

[0896] Warning message: "Caution is advised. Risk level: Medium"

[0897] Feedback: "This expression may be considered rude."

[0898] 6. The server generates appropriate alternative text and provides it to the user.

[0899] Alternative suggestion: "Please respond promptly if any problems arise."

[0900] Example of a prompt

[0901] Examples of prompt statements to input into the generative AI model are as follows:

[0902] User-generated email: "You can't even fulfill such a simple request? Do you even have any motivation?"

[0903] Please suggest the feedback and alternative sentences that the system will generate.

[0904] conclusion

[0905] This system optimizes communication between companies and customers, reduces the psychological burden on employees, and effectively manages the potential risk of customer harassment.

[0906] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0907] Step 1: Data Collection

[0908] The user composes an email or chat message and clicks the send button.

[0909] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[0910] The device collects user input data and extracts email content and chat messages. If audio data is present, it is converted into text using speech recognition technology.

[0911] Input: User's email or chat message

[0912] Output: Text data ("You can't even meet such a simple request? Do you even have any motivation?")

[0913] Step 2: Data transmission

[0914] The device sends the collected data to the server.

[0915] Input: Text data

[0916] Output: Sending data to the server

[0917] Step 3: Data Reception

[0918] The server receives data sent from the terminal.

[0919] Input: Text data from the terminal

[0920] Output: Received data

[0921] Step 4: Emotion Analysis

[0922] The server inputs the received data into an analysis tool. The analysis tool uses either "Google Cloud Natural Language API" or "IBM Watson Natural Language Understanding".

[0923] Input: Received data (text data)

[0924] Data processing: An emotion engine analyzes emotions from context and word choice.

[0925] Output: Emotion analysis results (e.g., high level of anger)

[0926] Step 5: Risk Assessment

[0927] The server determines the risk level of customer harassment based on the analysis results from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[0928] Input: Sentiment analysis results

[0929] Data processing: Assess risk based on the intensity of emotions and specific keywords.

[0930] Output: Risk level (high, medium, low)

[0931] Step 6: Generate warning message

[0932] If the server is determined to have a high or medium risk level, it will generate a warning message and feedback.

[0933] Input: Risk level

[0934] Data processing: Generate warning messages and feedback based on risk level.

[0935] Output: Warning messages and feedback

[0936] Step 7: Send a warning message

[0937] The server sends the generated warning message and feedback to the user's terminal.

[0938] Input: Warning messages and feedback

[0939] Output: Sent to the user's terminal

[0940] Step 8: Generate alternative text

[0941] The server generates appropriate alternative text and provides it to the user.

[0942] Input: Risk level and sentiment analysis results

[0943] Data processing: Generate appropriate and gentle alternative text based on sentiment analysis results.

[0944] Output: Alternative text suggestions

[0945] Step 9: Send alternative text

[0946] The server sends the generated alternative text to the user's terminal.

[0947] Input: Alternative text

[0948] Output: Sent to the user's terminal

[0949] Specific example:

[0950] When a user sends a message via chat saying, "Your response is too slow. What are you doing?",

[0951] 1. The device collects messages and sends them to the server.

[0952] 2. The server receives the data, performs sentiment analysis, and determines that the level of irritation is high.

[0953] 3. The server assesses the risk as moderate and generates a warning message and feedback indicating that attention is needed.

[0954] 4. The server sends a warning message ("Caution is advised. Risk level: Medium") and feedback ("This language may be offensive.") to the user.

[0955] 5. The server generates alternative text ("Please respond promptly if a problem occurs.") and provides it to the user.

[0956] In this way, the system assesses the risk of customer harassment at each step and provides appropriate responses and feedback.

[0957] (Application Example 2)

[0958] 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."

[0959] In modern businesses, customer harassment is on the rise during customer service interactions, negatively impacting employee psychological burden and overall business efficiency. To prevent such harassment, a system is needed that monitors customer interactions in real time and takes appropriate action. However, conventional systems struggle to accurately grasp user emotions and provide instant feedback, and they also lack sufficient suggestions for appropriate alternative expressions to mitigate risks. As a result, effective harassment prevention remains difficult, and more sophisticated systems are required.

[0960] 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.

[0961] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for providing feedback on the warnings and generating appropriate alternative text; means for analyzing the user's emotions and recognizing their emotional state; and means for sending improvement suggestions when the risk of customer harassment exceeds a certain level. This enables real-time determination of the risk of customer harassment, prompt warnings to employees, appropriate feedback, and suggestions for alternative expressions.

[0962] "Conversation content" refers to the words and expressions used when exchanging information in audio or text format.

[0963] "Email" refers to messages that are sent and received electronically via networks such as the internet.

[0964] "Chat" refers to real-time, text-based communication.

[0965] "Means of real-time collection" refers to devices or programs that have the function of collecting conversation content and text messages and sending them to a server.

[0966] "Means for analyzing collected data" refers to a system that has the function of analyzing collected information and detecting specific patterns or keywords.

[0967] "Means for determining the risk of customer harassment" refers to a system that has processes and technologies for evaluating the possibility of harassment in interactions with customers.

[0968] "Means of issuing warnings" refers to a system that has the function of sending alerts and warnings to users based on the results of risk assessment.

[0969] "Means of generating feedback and appropriate alternative text for warnings" refers to a system that has the function of suggesting directions for improvement or appropriate wording to the user.

[0970] "Means for analyzing user emotions and recognizing emotional states" refers to a system that possesses technologies and processes for performing emotion analysis on collected data and identifying the user's emotional state.

[0971] "Emotional state" refers to the user's psychological state, emotions, and mood.

[0972] "A means of sending improvement suggestions when the risk of customer harassment exceeds a certain level" refers to a system that has the technology or process to notify users of specific countermeasures or alternative expressions when the risk assessment is high.

[0973] System Overview

[0974] This invention provides a system that collects conversation content, emails, or chats in real time, determines the risk of customer harassment, and provides warnings and feedback based on the results. Furthermore, by recognizing the user's emotions, it enables more accurate risk assessment and generation of appropriate alternative text. The aim of this system is to reduce the risk of harassment in everyday business communications such as conversations and chats, and to reduce the psychological burden on companies and employees.

[0975] Hardware and software to be used

[0976] Hardware:

[0977] Smartphone: Used for collecting and sending conversations, chats, and emails.

[0978] Server: Used for data analysis, risk assessment, and feedback generation.

[0979] Robots: They can also be used to collect and transmit communications in security operations.

[0980] software:

[0981] Speech Recognition: Google Cloud Speech-to-Text

[0982] Sentiment Analysis: IBM Watson Natural Language Understanding, Amazon Comprehend Sentiment Analysis

[0983] Cloud services: AWS, Google Cloud Platform

[0984] System operation

[0985] Data collection:

[0986] When users engage in conversations or chats using their smartphones or robots, the data is collected in real time. The voice data is converted into text data using speech recognition technology and sent to a server.

[0987] Real-time analysis:

[0988] The server inputs the collected text data into a sentiment analysis tool, checking for specific keywords and contexts in real time. This helps identify the risk of customer harassment.

[0989] Emotion recognition:

[0990] The emotion engine is used to analyze the user's emotional state. For audio data, it analyzes the tone and speed of the voice; for text data, it analyzes the structure and word choice of the sentences to identify the user's emotional state.

[0991] Risk assessment:

[0992] Based on the analysis results from the emotion engine, the server determines the risk of customer harassment in three stages: high, medium, and low. If the risk level is determined to be medium or high, immediate action is taken.

[0993] Generating warnings and feedback:

[0994] Depending on the risk level, the server sends a warning message to the user's terminal. The warning includes the risk level and specific improvement suggestions, and appropriate alternative text is also generated and provided to the user.

[0995] Specific example

[0996] Preventing customer harassment in call centers

[0997] 1. When a user (employee) is providing chat support at a call center, they receive emotional messages.

[0998] Example: "Why are you taking so long to respond? Hurry up!"

[0999] 2. The employee's device sends the chat content to the server.

[1000] Sent text: "Why are you taking so long to respond? Hurry up!"

[1001] 3. The server analyzes the chat content, and the emotion engine recognizes the recipient's (customer's) emotions.

[1002] Emotional analysis results: High level of anger.

[1003] 4. The server determines the risk of customer harassment.

[1004] Judgment result: High risk

[1005] 5. The server sends a warning message and feedback to the employee's terminal.

[1006] Warning message: "There is a high risk. Please consider how to mitigate customer anger."

[1007] Feedback: "Explain the situation carefully and patiently until the customer is satisfied."

[1008] 6. The server generates an appropriate alternative sentence and provides it to the employee.

[1009] Alternative suggestion: "We apologize for the delay. We are currently working to process your request as quickly as possible, so please wait a little longer."

[1010] Examples of prompts to input into a generative AI model

[1011] "In customer service, I received a message that conveyed anger. What would be an appropriate response? For example, please suggest a response to the following message."

[1012] In this way, this system can significantly reduce the risk of customer harassment by assessing risks in real time and prompting users to take appropriate action immediately.

[1013] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1014] Step 1: Data Collection

[1015] Input: Data from conversations, emails, and chats that users have with smartphones or robots.

[1016] Specific operation: The user initiates communication, and voice and text data are generated.

[1017] Data processing: The audio data is converted to text data using Google Cloud Speech-to-Text. The text data is then sent directly to the server.

[1018] Output: Text-formatted conversation data, chat data, and email data are sent to the server.

[1019] Step 2: Forecast Analysis

[1020] Input: Text data collected in Step 1.

[1021] Specific operation: The server inputs the collected data into IBM Watson Natural Language Understanding or Amazon Comprehend Sentiment Analysis for real-time analysis.

[1022] Data processing: This involves detecting keywords within text data and performing overall contextual analysis. It also uses an emotion engine to analyze the user's emotional state.

[1023] Output: Sentiment analysis results and a list of specific keywords.

[1024] Step 3: Emotion Recognition

[1025] Input: Sentiment analysis results and specific keyword list analyzed in Step 2.

[1026] Specific operation: The server uses the results of the emotion engine to determine the user's emotional state. In particular, it analyzes the intensity and trend of emotions.

[1027] Data processing: Based on the emotion analysis results, the user's emotional state is classified into categories such as "anger," "sadness," and "joy."

[1028] Output: Classified emotion state data.

[1029] Step 4: Risk Assessment

[1030] Input: Emotional state data and specific keyword list obtained in Step 3.

[1031] Specific operation: The server determines the risk of customer harassment based on the collected data. The risk level is evaluated in three stages: "high," "medium," and "low."

[1032] Data calculation: Calculate harassment risk based on the intensity of emotions and the frequency of specific keywords.

[1033] Output: Risk assessment result (high, medium, low).

[1034] Step 5: Generating warnings and feedback

[1035] Input: Risk assessment results (high, medium, low) and emotional state data.

[1036] Specific operation: Based on the determination result, the server sends a warning message to the user's terminal.

[1037] Data processing: Warning messages will include risk levels along with specific improvement suggestions. Additionally, a generative AI model will be used to generate appropriate alternative text.

[1038] Output: Warning message, improvement suggestion, and appropriate alternative text.

[1039] Step 6: Suggesting alternative text

[1040] Input: Improvement suggestions and alternative text.

[1041] Specific operation: The server uses a generative AI model to provide the user with appropriate alternative text.

[1042] Output: The provided alternative text will be displayed on the user's terminal.

[1043] This allows the entire system to assess the risk of customer harassment in real time and promptly prompt users to take appropriate action.

[1044] 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.

[1045] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). An 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.

[1046] 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.

[1047] [Third Embodiment]

[1048] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[1049] 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.

[1050] 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).

[1051] 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.

[1052] 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.

[1053] 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).

[1054] 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.

[1055] 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.

[1056] 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.

[1057] 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.

[1058] 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.

[1059] 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".

[1060] overview

[1061] This invention is a system that collects conversation content, emails, or chats in real time, analyzes the risk of customer harassment, and issues warnings. Furthermore, it is characterized by generating appropriate feedback and alternative text along with the warning and providing them to the user.

[1062] System Configuration

[1063] 1. User's terminal

[1064] It collects data from conversations, emails, and chats.

[1065] Includes speech recognition functionality for converting audio data into text data.

[1066] It communicates with the server and sends data.

[1067] 2. Server

[1068] Receive data sent from the device.

[1069] The system analyzes data in real time to determine the risk of customer harassment.

[1070] Based on the judgment, a warning message and feedback are generated.

[1071] Generate appropriate alternative text and send it to the user's device.

[1072] Program processing

[1073] 1. Data Collection

[1074] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[1075] 2. Real-time analysis

[1076] The server analyzes the received text data and assesses the risk of customer harassment. This assessment includes sentiment analysis and detection of specific keywords. Based on the analysis results, a risk level is determined.

[1077] 3. Generating warnings and feedback

[1078] If the server determines that there is a high risk of customer harassment, it will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[1079] Specific example

[1080] Example: Detecting customer harassment via email

[1081] 1. The user composes an emotional email and clicks the send button.

[1082] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1083] 2. The device sends the email body to the server.

[1084] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[1085] 3. The server analyzes the content of the email and determines the risk of customer harassment.

[1086] Judgment result: High risk

[1087] 4. The server sends feedback to the user's terminal along with a warning message.

[1088] Warning message: "Suspected customer harassment. Risk level: High"

[1089] Feedback: "This kind of expression should be avoided."

[1090] 5. The server generates an appropriate alternative sentence and provides it to the user.

[1091] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[1092] Example: Detecting customer harassment in chat

[1093] 1. The user types an aggressive message in the chat and clicks the send button.

[1094] Example: "Your response is too slow. What are you doing?"

[1095] 2. The device sends the chat content to the server.

[1096] Sent text: "Your response is too slow. What are you doing?"

[1097] 3. The server analyzes the chat content and assesses the risk of customer harassment.

[1098] Assessment result: Medium risk

[1099] 4. The server sends a warning and feedback to the user's terminal.

[1100] Warning message: "Caution is advised. Risk level: Medium"

[1101] Feedback: "This expression may be considered rude."

[1102] 5. The server generates an appropriate alternative sentence and provides it to the user.

[1103] Alternative suggestion: "Please respond promptly if any problems arise."

[1104] In this way, this system can assess the risk of customer harassment in real time and provide appropriate feedback and alternative text, thereby promoting proper communication between companies and customers and reducing the psychological burden on employees.

[1105] The following describes the processing flow.

[1106] Step 1:

[1107] The user begins typing an email, chat message, or voice call. The device prepares to collect this data.

[1108] Step 2:

[1109] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[1110] Step 3:

[1111] The device collects email body text, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[1112] Step 4:

[1113] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[1114] Step 5:

[1115] The server inputs the received data into an analysis tool. This analysis tool includes sentiment analysis and specific keyword detection algorithms.

[1116] Step 6:

[1117] The server uses analysis tools to analyze incoming data in real time. As a result, the risk level of customer harassment is determined. The risk level is evaluated in three stages: high, medium, and low.

[1118] Step 7:

[1119] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[1120] Step 8:

[1121] The server generates appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[1122] Step 9:

[1123] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[1124] Step 10:

[1125] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[1126] Step 11:

[1127] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[1128] In this way, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions.

[1129] (Example 1)

[1130] 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."

[1131] In today's business environment, communication between companies and customers takes place through a variety of means. However, excessive demands and inappropriate remarks from customers, i.e., customer harassment, can increase the psychological burden on employees and lead to a deterioration of the working environment. Therefore, it is necessary to monitor conversation content in real time, detect the risk of customer harassment early, and take appropriate measures. However, current systems have problems such as delays in analyzing collected data and inability to accurately assess risks.

[1132] 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.

[1133] In this invention, the server includes means for collecting conversation content, electronic messages, or instant messages in real time; means for analyzing the collected data and evaluating the risk of customer harassment; and means for issuing warnings based on the evaluation results. This makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[1134] "Conversation content" refers to the content of communication conducted through voice.

[1135] "Electronic messages" refer to messages sent and received in digital format, such as emails and text messages.

[1136] "Instant messaging" refers to real-time chats and instant messages.

[1137] "Methods for collecting information in real time" refers to technologies for collecting conversation content and messages almost instantly.

[1138] "Analysis" refers to the process of evaluating collected data using mechanical and algorithmic methods to extract meaning and emotion.

[1139] "Customer harassment" refers to acts that cause psychological distress to employees through excessive demands or inappropriate remarks from customers.

[1140] "Means of risk assessment" refers to technologies used to determine the risk level of customer harassment through data analysis.

[1141] "Warning mechanisms" refer to technologies used to send warning messages to users when there is a risk.

[1142] "Speech recognition means" refers to technology for converting speech data into text data.

[1143] A "machine intelligence model" refers to a technology that uses artificial intelligence to analyze data and recognize specific patterns or emotions.

[1144] "Alternative text" refers to appropriate wording that is recommended to be used instead of the original message.

[1145] "Feedback" refers to suggestions, proposals, or specific advice regarding a warning.

[1146] The system in this invention is built to collect conversation content, electronic messages, or instant messages in real time, determine the risk of customer harassment, and issue warnings. This system includes a user terminal, a server, and a generative AI model.

[1147] 1. User terminal

[1148] The user terminal is a device that collects data from conversations, electronic messages, or instant messages. It also has a speech recognition system that converts voice data into text data. Speech recognition software such as "Google Cloud Speech-to-Text" is used for speech recognition. It also has the function to collect text data entered by the user and send it to the server.

[1149] Specific example: A user types the message "You can't even fulfill such a simple request? Do you even have any motivation?" into the chat window and clicks the send button.

[1150] 2. Server

[1151] The server is a central processing unit that analyzes text data received from user terminals. It analyzes the received data in real time and assesses the risk of customer harassment. The analysis uses natural language processing (NLP) technologies such as "SpaCy" and "IBM Watson" for sentiment analysis and detection of specific keywords. Based on the analysis results, it determines the risk level and generates warning messages and feedback as needed. Furthermore, it uses generative AI models (e.g., OpenAI's GPT-3) to generate appropriate alternative text.

[1152] Specific example: The server receives a message saying, "You can't even fulfill such a simple request? Are you even motivated?" and the analysis results indicate it's "high risk." Subsequently, it generates a warning message, "Suspected customer harassment. Risk level: high," and appropriate feedback, "Such language should be avoided," and creates alternative text, "Is there anything you didn't understand in this section? I'd be happy to explain in more detail."

[1153] Example of a prompt

[1154] A user is attempting to send the following message. Please check if this message poses a risk of customer harassment and generate a warning message and appropriate alternative text. Original message: "You can't even fulfill such a simple request? Are you even motivated?"

[1155] The above describes the embodiments for carrying out the present invention. This system makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[1156] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1157] Step 1:

[1158] The user creates a conversation, e-message, or instant message and clicks the send button. The user's device then collects the created message.

[1159] Input: User-entered conversation, e-message, or instant message (e.g., "You can't even fulfill such a simple request? Are you even motivated?")

[1160] Output: Message data stored on the user's terminal

[1161] Specific action: The user types a message in the chat window and clicks the send button.

[1162] Step 2:

[1163] The user terminal prepares to send the collected message data to the server in real time. If voice data is included, speech recognition is used to convert the voice to text.

[1164] Input: Voice data or text data

[1165] Output: Text data

[1166] Specific operation: A speech recognition engine (e.g., Google Cloud Speech-to-Text) converts speech to text. If the data is text, it is used as is.

[1167] Step 3:

[1168] The user terminal sends text data to the server.

[1169] Input: Text data (Example: "You can't even meet such a simple request? Do you even have any motivation?")

[1170] Output: Text data sent to the server

[1171] Specific operation: The user terminal uses its communication function to send the converted text data to the server.

[1172] Step 4:

[1173] The server analyzes the received text data. The server uses natural language processing technologies (e.g., SpaCy, IBM Watson) to perform sentiment analysis and detect specific keywords.

[1174] Input: Text data

[1175] Output: Analysis results (sentiment score, keyword detection results)

[1176] Specific operation: The server uses SpaCy or IBM Watson to analyze the sentiment of incoming messages and detect specific keywords.

[1177] Step 5:

[1178] The server assesses the risk of customer harassment based on the analysis results. This assessment includes sentiment scores and keyword detection results.

[1179] Input: Analysis results (sentiment score, keyword detection results)

[1180] Output: Risk assessment results (high risk, medium risk, low risk, etc.)

[1181] Specific operation: The server evaluates the risk level of customer harassment based on sentiment scores and keyword detection results (e.g., determines it as "high risk").

[1182] Step 6:

[1183] The server generates a warning message based on the risk assessment results. If the risk is high, it also generates specific improvement suggestions and alternative text. A generative AI model is used to create appropriate feedback and alternative text.

[1184] Input: Risk assessment results

[1185] Output: Warning message, feedback, alternative text

[1186] Specific actions: The server generates a warning message "Suspected customer harassment. Risk level: High," feedback "Such language should be avoided," and alternative text "Is there anything you don't understand about this part? We'd be happy to explain in more detail."

[1187] Step 7:

[1188] The server sends generated warning messages, feedback, and alternative text to the user's terminal.

[1189] Input: Warning message, feedback, alternative text

[1190] Output: Warning messages, feedback, and alternative text sent to the user's terminal.

[1191] Specific operation: The server uses its communication function to send warning messages, feedback, and alternative text to the user's terminal.

[1192] Through the above processing steps, this system can assess the risk of customer harassment in real time and immediately provide necessary countermeasures.

[1193] (Application Example 1)

[1194] 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."

[1195] Customer support and call center staff are required to detect the risk of customer harassment in real time during customer conversations and to take appropriate action quickly based on that risk. Traditional methods require staff to assess the risk and devise countermeasures themselves, which places a significant psychological burden on them and carries the risk of incorrect responses. This challenge needs to be addressed.

[1196] 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.

[1197] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing a warning based on the determination result; means for providing feedback on the warning and generating appropriate alternative text; means for collecting conversation content using speech recognition means that convert speech to text in real time; means for determining the risk of customer harassment using a device that performs sentiment analysis; means for determining the risk level based on the determination result and presenting an appropriate alert; means for displaying feedback on a smart device worn by an employee; and natural language processing means for generating appropriate alternative text. This enables staff to detect the risk of customer harassment in real time and take appropriate action quickly.

[1198] "Conversation content" refers to voice-based communication between customer support or call center staff and customers.

[1199] "Email" refers to messages sent and received electronically via the internet.

[1200] "Chat" refers to real-time text-based communication.

[1201] "Means of real-time collection" refers to methods or devices for instantly acquiring this data as conversations, emails, or chats occur.

[1202] "Means of analyzing collected data" refers to methods or devices for analyzing collected conversations, emails, or chats to assess the risk of customer harassment.

[1203] "Means for determining the risk of customer harassment" refers to a method or device that detects inappropriate words or actions from customers and determines their risk level based on collected and analyzed data.

[1204] "Means of issuing warnings" refers to methods or devices for sending appropriate notifications or warning messages to users when a risk of customer harassment is identified.

[1205] "Feedback" or "means for generating appropriate alternative text" refers to a method or device for generating and proposing appropriate countermeasures or alternative expressions to the user in response to the risk of customer harassment.

[1206] "Speech recognition means" refers to technologies and devices for converting speech data into text data.

[1207] "An emotional analysis device" refers to a method or device for analyzing the emotional aspects of text data and detecting specific emotions or tones.

[1208] "Means for determining risk levels" refers to methods or devices for evaluating the degree of customer harassment risk and quantifying that level based on the results of sentiment analysis.

[1209] "Means of providing appropriate alerts" refers to methods or devices for providing warnings or notifications to employees based on the determined risk level.

[1210] "Means of displaying feedback on smart devices" refers to methods or devices for displaying risk levels or feedback messages on wearable devices or other smart devices worn by users.

[1211] "Natural language processing means" refers to language processing technologies and devices that understand text data and generate appropriate alternative text.

[1212] This invention provides a system for detecting the risk of customer harassment in real time in customer support and call centers, and for taking appropriate action quickly based on that risk. The system includes user terminals, servers, and smart devices worn by employees.

[1213] System Configuration

[1214] 1. User's terminal

[1215] It has the ability to collect data from conversations, emails, and chats.

[1216] It is equipped with speech recognition means for converting audio data into text data.

[1217] It has the function of communicating with a server and sending data.

[1218] 2. Server

[1219] It has the ability to receive data sent from a device and analyze it in real time.

[1220] This includes an emotion analysis device for analyzing data and determining the risk of customer harassment, as well as an artificial intelligence model for detecting specific keywords.

[1221] Based on the assessment results, the risk level is determined, and warning and feedback messages are generated.

[1222] The generated warnings and feedback are sent to the user's device and to smart devices that display the feedback.

[1223] 3. Smart devices

[1224] Employees wear the device, which displays warning and feedback messages sent from the server.

[1225] Hardware and software to be used

[1226] Speech recognition method: The "speech_recognition" library and Google's speech recognition API are used.

[1227] Sentiment analysis device: Uses a pre-trained sentiment classification model (such as BERT) using the "transformers" library.

[1228] Smart devices: Employees can wear smart glasses or use smartphones.

[1229] Servers: High-performance servers are used for processing and communicating this data.

[1230] Processing flow

[1231] 1. Data collection:

[1232] When a user initiates a conversation or chat with a customer, the device collects conversation and chat data in real time, and the voice data is converted into text data using speech recognition technology.

[1233] 2. Real-time analysis:

[1234] The collected text data is sent to a server, which uses sentiment analysis devices and artificial intelligence models to analyze the risk of customer harassment.

[1235] 3. Generating warnings and feedback:

[1236] If the risk level of customer harassment is high, the server will generate a warning and specific feedback message, or appropriate alternative text, depending on the risk level.

[1237] 4. Notifications and Feedback:

[1238] The generated warning and feedback messages are sent to and displayed on the user's device and the employee's smart device, allowing employees to take appropriate action immediately.

[1239] Specific example

[1240] For example, if a support staff member is talking to a customer and the customer makes an aggressive statement such as, "Your response is too slow! What's going on?", this system collects the statement in real time and converts it to text using speech recognition. Then, an emotion analyzer analyzes this text, and if it is determined to be high risk, the server generates a warning message and feedback such as, "Specifically, what is the problem? We'd be happy to help," which is displayed on the staff member's smart glasses.

[1241] Prompts for Generative AI Models

[1242] Design a custom-made smart glasses application to detect customer harassment during customer support conversations and provide appropriate alerts and feedback in real time. Generate Python pseudocode that meets the following requirements.

[1243] 1. Collection of audio data

[1244] 2. Text conversion using speech recognition

[1245] 3. Risk assessment using sentiment analysis

[1246] 4. Generate feedback based on the judgment result.

[1247] The above describes the embodiments for carrying out the present invention. By introducing this system, it becomes possible to detect customer harassment in real time at customer support sites and take prompt and appropriate action, thereby reducing the psychological burden on employees.

[1248] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1249] Step 1:

[1250] The user initiates a conversation with a customer. The input is the voice conversation with the customer. The device collects this voice data. The device captures the voice data in real time using its built-in microphone.

[1251] Step 2:

[1252] The device converts the collected audio data into text data. The input is the audio data collected in step 1. The device uses its speech recognition capabilities (using the speech_recognition library and Google's speech recognition API) to convert the audio data into text data. The output is text data.

[1253] Step 3:

[1254] The text data is sent to the server. The input is the text data generated in step 2. The terminal communicates with the server and sends the text data to the server. The output is that the text data is sent to the server.

[1255] Step 4:

[1256] The server analyzes the received text data. The input is the text data sent to the server in step 3. The server uses a sentiment analyzer (using the transformers library) and an artificial intelligence model that detects specific keywords to perform sentiment analysis and keyword detection on the text data. The output is the risk assessment result.

[1257] Step 5:

[1258] The server determines the risk level based on the risk assessment results and generates warnings and feedback. The input is the risk assessment results obtained in step 4. The server determines the risk level (high, medium, low) according to the assessment results and generates appropriate warning messages and feedback messages, as well as appropriate alternative text using natural language processing. The output is the generated warning messages and feedback messages.

[1259] Step 6:

[1260] The server sends the generated warning and feedback messages to the user's terminal and smart device. The input is the warning and feedback messages generated in step 5. The server sends these messages to the terminal and smart device via communication. The output is the warning and feedback displayed on the user's terminal and smart device.

[1261] Step 7:

[1262] The user's terminal and smart device display warnings and feedback. The input consists of the warning and feedback messages sent from the server in step 6. The terminal and smart device use their display functions to present the risk level, warnings, feedback, and alternative text to the user. The output is the visually displayed message.

[1263] The above outlines the processing steps of a system that detects customer harassment risks in real time and provides appropriate feedback.

[1264] 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.

[1265] overview

[1266] This invention relates to a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it provides more accurate risk assessment and feedback. This system analyzes user emotions in real time and proposes appropriate measures to reduce the risk of customer harassment.

[1267] System Configuration

[1268] 1. User's terminal

[1269] It collects data from conversations, emails, and chats.

[1270] Includes speech recognition functionality to convert audio data into text data.

[1271] It communicates with the server and sends data.

[1272] 2. Server

[1273] Receive data sent from the device.

[1274] The system analyzes data in real time to determine the risk of customer harassment.

[1275] We use an emotion engine that recognizes user emotions to perform emotional analysis.

[1276] Based on the judgment result, a warning message and feedback are generated.

[1277] Generate appropriate alternative text and send it to the user's device.

[1278] Program processing

[1279] 1. Data Collection

[1280] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[1281] 2. Real-time analysis

[1282] The server inputs the received data into an analysis tool. The analysis tool includes sentiment analysis and specific keyword detection algorithms, and in addition, it operates an emotion engine that recognizes the user's emotions.

[1283] 3. Emotion recognition

[1284] The emotion engine analyzes the user's emotions from incoming data. For audio data, it analyzes tone and speed of voice; for text data, it analyzes context and word choice to determine whether the user is emotional.

[1285] 4. Risk Assessment

[1286] The server determines the risk level of customer harassment based on analysis results, including those from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[1287] 5. Generating warnings and feedback

[1288] If the risk level is determined to be high or medium, the server will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[1289] Specific example

[1290] Example: Detecting customer harassment via email

[1291] 1. The user composes an emotional email and clicks the send button.

[1292] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1293] 2. The device sends the email body to the server.

[1294] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[1295] 3. The server analyzes the email content, and the emotion engine recognizes the user's emotions.

[1296] Emotional analysis results: High level of anger.

[1297] 4. The server determines the risk of customer harassment based on the overall analysis results.

[1298] Judgment result: High risk

[1299] 5. The server sends feedback to the user's terminal along with a warning message.

[1300] Warning message: "Suspected customer harassment. Risk level: High"

[1301] Feedback: "This kind of expression should be avoided."

[1302] 6. The server generates an appropriate alternative sentence and provides it to the user.

[1303] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[1304] Example: Detecting customer harassment in chat

[1305] 1. The user types an aggressive message in the chat and clicks the send button.

[1306] Example: "Your response is too slow. What are you doing?"

[1307] 2. The device sends the chat content to the server.

[1308] Sent text: "Your response is too slow. What are you doing?"

[1309] 3. The server analyzes the chat content, and the emotion engine recognizes the user's emotions.

[1310] Emotional analysis results: High level of irritation.

[1311] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[1312] Assessment result: Medium risk

[1313] 5. The server sends a warning and feedback to the user's terminal.

[1314] Warning message: "Caution is advised. Risk level: Medium"

[1315] Feedback: "This expression may be considered rude."

[1316] 6. The server generates an appropriate alternative sentence and provides it to the user.

[1317] Alternative suggestion: "Please respond promptly if any problems arise."

[1318] In this way, by combining an emotion engine, this system accurately grasps the user's emotions, determines the risk of customer harassment in real time, and provides appropriate feedback and alternative expressions. This optimizes communication between companies and customers and reduces the psychological burden on employees.

[1319] The following describes the processing flow.

[1320] Step 1:

[1321] The user initiates an email, chat message, or conversation. The device prepares to collect this data in real time.

[1322] Step 2:

[1323] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[1324] Step 3:

[1325] The device collects email content, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[1326] Step 4:

[1327] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[1328] Step 5:

[1329] The server inputs the received data into the analysis tool. The analysis tool includes a sentiment analysis engine and a specific keyword detection algorithm.

[1330] Step 6:

[1331] The emotion engine analyzes the user's emotions. Specifically, it recognizes emotions based on tone and speed of voice in the case of audio data, and on context and word choice in the case of text data.

[1332] Step 7:

[1333] The server integrates analysis results, including those from the emotion engine, to determine the risk level of customer harassment. The risk level is evaluated in three stages: high, medium, and low.

[1334] Step 8:

[1335] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[1336] Step 9:

[1337] The server generates a warning message along with appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[1338] Step 10:

[1339] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[1340] Step 11:

[1341] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[1342] Step 12:

[1343] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[1344] As described above, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions. By combining it with an emotion engine, it can accurately grasp the user's emotions and enable more accurate risk assessment and countermeasures.

[1345] (Example 2)

[1346] 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."

[1347] There is a problem of increasing psychological burden on employees due to customer harassment. Furthermore, it is difficult to determine the risk of customer harassment in real time, making it difficult to take appropriate action and provide feedback. To solve this problem, a system is needed that analyzes user emotions in real time and proposes quick and appropriate countermeasures.

[1348] 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.

[1349] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for analyzing the user's emotions using an emotion engine; means for evaluating the risk level in three stages: high, medium, and low; and means for inputting the data into an analysis tool. This makes it possible to evaluate the risk of customer harassment in real time and provide rapid and appropriate feedback and alternative text.

[1350] "Conversation content" refers to all communication content in audio and text format, primarily including information in dialogue and chat formats.

[1351] "Email" refers to messages containing data such as text, images, and audio that are sent and received over the internet.

[1352] "Chat" refers to online conversational communication where text messages are exchanged in real time.

[1353] "Real-time data collection" means collecting data simultaneously with user actions and processing it without delay.

[1354] "Data analysis" refers to the techniques used to process and analyze collected information to identify useful insights and risks.

[1355] "Customer harassment" refers to acts that cause employees psychological distress due to excessive demands or intimidating behavior from customers.

[1356] "Risk assessment" refers to the process of evaluating whether a particular action or statement carries a risk of customer harassment through sentiment analysis and keyword detection.

[1357] A "warning" is a message from the system to alert the user, and may include feedback based on the risk level.

[1358] "Feedback" refers to evaluations and suggestions for improvement regarding a user's actions and comments.

[1359] "Alternative text" refers to improved versions of the original statement, suggested to avoid customer harassment.

[1360] An "emotion engine" refers to an algorithm or model used to analyze a user's emotional state, sensing emotions based on voice and text data.

[1361] "Risk level" refers to a three-tiered evaluation (high, medium, low) that indicates how high the risk of customer harassment is.

[1362] "Analysis tools" refer to software and algorithms used for data analysis and sentiment analysis.

[1363] Modes for carrying out the invention

[1364] System Overview

[1365] This invention is a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it can provide more accurate risk assessment and feedback. This system can analyze user emotions in real time and propose appropriate measures to reduce the risk of customer harassment.

[1366] System Configuration

[1367] 1. User's device:

[1368] This includes means of collecting data from conversations, emails, and chats.

[1369] Includes speech recognition means for converting audio data into text data.

[1370] Includes means for communicating with a server and transmitting data.

[1371] 2. Server:

[1372] Includes means for receiving data transmitted from a terminal.

[1373] This includes methods for analyzing data in real time and determining the risk of customer harassment.

[1374] This includes means for performing emotional analysis using an emotion engine that recognizes the user's emotions.

[1375] Includes means for generating warning messages and feedback based on the judgment result.

[1376] Includes means for generating appropriate alternative text and sending it to the user's terminal.

[1377] Specific hardware and software to be used

[1378] We will use "Google Cloud Speech-to-Text API" or "IBM Watson Speech to Text" as speech recognition software.

[1379] We use Google Cloud Natural Language API and IBM Watson Natural Language Understanding for sentiment analysis and text analysis.

[1380] Standard server and cloud computing environments are used for data processing and analysis.

[1381] Specific examples of operation

[1382] Example 1: Detecting customer harassment via email

[1383] 1. The user composes an emotional email and clicks the send button.

[1384] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1385] 2. The device sends the email body to the server.

[1386] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[1387] 3. The server sends the email content to the Google Cloud Natural Language API for sentiment analysis.

[1388] Emotional analysis results: High level of anger.

[1389] 4. The server determines the risk of customer harassment based on the overall analysis results.

[1390] Judgment result: High risk

[1391] 5. The server sends feedback to the user's terminal along with a warning message.

[1392] Warning message: "Suspected customer harassment. Risk level: High"

[1393] Feedback: "This kind of expression should be avoided."

[1394] 6. The server generates appropriate alternative text and provides it to the user.

[1395] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[1396] Example 2: Detection of customer harassment in chat

[1397] 1. The user types an aggressive message in the chat and clicks the send button.

[1398] Example: "Your response is too slow. What are you doing?"

[1399] 2. The device sends the chat content to the server.

[1400] Sent text: "Your response is too slow. What are you doing?"

[1401] 3. The server sends the chat content to "IBM Watson Natural Language Understanding" for sentiment analysis.

[1402] Emotional analysis results: High level of irritation.

[1403] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[1404] Assessment result: Medium risk

[1405] 5. The server sends a warning and feedback to the user's terminal.

[1406] Warning message: "Caution is advised. Risk level: Medium"

[1407] Feedback: "This expression may be considered rude."

[1408] 6. The server generates appropriate alternative text and provides it to the user.

[1409] Alternative suggestion: "Please respond promptly if any problems arise."

[1410] Example of a prompt

[1411] Examples of prompt statements to input into the generative AI model are as follows:

[1412] User-generated email: "You can't even fulfill such a simple request? Do you even have any motivation?"

[1413] Please suggest the feedback and alternative sentences that the system will generate.

[1414] conclusion

[1415] This system optimizes communication between companies and customers, reduces the psychological burden on employees, and effectively manages the potential risk of customer harassment.

[1416] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1417] Step 1: Data Collection

[1418] The user composes an email or chat message and clicks the send button.

[1419] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1420] The device collects user input data and extracts email content and chat messages. If audio data is present, it is converted into text using speech recognition technology.

[1421] Input: User's email or chat message

[1422] Output: Text data ("You can't even meet such a simple request? Do you even have any motivation?")

[1423] Step 2: Data transmission

[1424] The device sends the collected data to the server.

[1425] Input: Text data

[1426] Output: Sending data to the server

[1427] Step 3: Data Reception

[1428] The server receives data sent from the terminal.

[1429] Input: Text data from the terminal

[1430] Output: Received data

[1431] Step 4: Emotion Analysis

[1432] The server inputs the received data into an analysis tool. The analysis tool uses either "Google Cloud Natural Language API" or "IBM Watson Natural Language Understanding".

[1433] Input: Received data (text data)

[1434] Data processing: An emotion engine analyzes emotions from context and word choice.

[1435] Output: Emotion analysis results (e.g., high level of anger)

[1436] Step 5: Risk Assessment

[1437] The server determines the risk level of customer harassment based on the analysis results from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[1438] Input: Sentiment analysis results

[1439] Data processing: Assess risk based on the intensity of emotions and specific keywords.

[1440] Output: Risk level (high, medium, low)

[1441] Step 6: Generate warning message

[1442] If the server is determined to have a high or medium risk level, it will generate a warning message and feedback.

[1443] Input: Risk level

[1444] Data processing: Generate warning messages and feedback based on risk level.

[1445] Output: Warning messages and feedback

[1446] Step 7: Send a warning message

[1447] The server sends the generated warning message and feedback to the user's terminal.

[1448] Input: Warning messages and feedback

[1449] Output: Sent to the user's terminal

[1450] Step 8: Generate alternative text

[1451] The server generates appropriate alternative text and provides it to the user.

[1452] Input: Risk level and sentiment analysis results

[1453] Data processing: Generate appropriate and gentle alternative text based on sentiment analysis results.

[1454] Output: Alternative text suggestions

[1455] Step 9: Send alternative text

[1456] The server sends the generated alternative text to the user's terminal.

[1457] Input: Alternative text

[1458] Output: Sent to the user's terminal

[1459] Specific example:

[1460] When a user sends a message via chat saying, "Your response is too slow. What are you doing?",

[1461] 1. The device collects messages and sends them to the server.

[1462] 2. The server receives the data, performs sentiment analysis, and determines that the level of irritation is high.

[1463] 3. The server assesses the risk as moderate and generates a warning message and feedback indicating that attention is needed.

[1464] 4. The server sends a warning message ("Caution is advised. Risk level: Medium") and feedback ("This language may be offensive.") to the user.

[1465] 5. The server generates alternative text ("Please respond promptly if a problem occurs.") and provides it to the user.

[1466] In this way, the system assesses the risk of customer harassment at each step and provides appropriate responses and feedback.

[1467] (Application Example 2)

[1468] 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."

[1469] In modern businesses, customer harassment is on the rise during customer service interactions, negatively impacting employee psychological burden and overall business efficiency. To prevent such harassment, a system is needed that monitors customer interactions in real time and takes appropriate action. However, conventional systems struggle to accurately grasp user emotions and provide instant feedback, and they also lack sufficient suggestions for appropriate alternative expressions to mitigate risks. As a result, effective harassment prevention remains difficult, and more sophisticated systems are required.

[1470] 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.

[1471] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for providing feedback on the warnings and generating appropriate alternative text; means for analyzing the user's emotions and recognizing their emotional state; and means for sending improvement suggestions when the risk of customer harassment exceeds a certain level. This enables real-time determination of the risk of customer harassment, prompt warnings to employees, appropriate feedback, and suggestions for alternative expressions.

[1472] "Conversation content" refers to the words and expressions used when exchanging information in audio or text format.

[1473] "Email" refers to messages that are sent and received electronically via networks such as the internet.

[1474] "Chat" refers to real-time, text-based communication.

[1475] "Means of real-time collection" refers to devices or programs that have the function of collecting conversation content and text messages and sending them to a server.

[1476] "Means for analyzing collected data" refers to a system that has the function of analyzing collected information and detecting specific patterns or keywords.

[1477] "Means for determining the risk of customer harassment" refers to a system that has processes and technologies for evaluating the possibility of harassment in interactions with customers.

[1478] "Means of issuing warnings" refers to a system that has the function of sending alerts and warnings to users based on the results of risk assessment.

[1479] "Means of generating feedback and appropriate alternative text for warnings" refers to a system that has the function of suggesting directions for improvement or appropriate wording to the user.

[1480] "Means for analyzing user emotions and recognizing emotional states" refers to a system that possesses technologies and processes for performing emotion analysis on collected data and identifying the user's emotional state.

[1481] "Emotional state" refers to the user's psychological state, emotions, and mood.

[1482] "A means of sending improvement suggestions when the risk of customer harassment exceeds a certain level" refers to a system that has the technology or process to notify users of specific countermeasures or alternative expressions when the risk assessment is high.

[1483] System Overview

[1484] This invention provides a system that collects conversation content, emails, or chats in real time, determines the risk of customer harassment, and provides warnings and feedback based on the results. Furthermore, by recognizing the user's emotions, it enables more accurate risk assessment and generation of appropriate alternative text. The aim of this system is to reduce the risk of harassment in everyday business communications such as conversations and chats, and to reduce the psychological burden on companies and employees.

[1485] Hardware and software to be used

[1486] Hardware:

[1487] Smartphone: Used for collecting and sending conversations, chats, and emails.

[1488] Server: Used for data analysis, risk assessment, and feedback generation.

[1489] Robots: They can also be used to collect and transmit communications in security operations.

[1490] software:

[1491] Speech Recognition: Google Cloud Speech-to-Text

[1492] Sentiment Analysis: IBM Watson Natural Language Understanding, Amazon Comprehend Sentiment Analysis

[1493] Cloud services: AWS, Google Cloud Platform

[1494] System operation

[1495] Data collection:

[1496] When users engage in conversations or chats using their smartphones or robots, the data is collected in real time. The voice data is converted into text data using speech recognition technology and sent to a server.

[1497] Real-time analysis:

[1498] The server inputs the collected text data into a sentiment analysis tool, checking for specific keywords and contexts in real time. This helps identify the risk of customer harassment.

[1499] Emotion recognition:

[1500] The emotion engine is used to analyze the user's emotional state. For audio data, it analyzes the tone and speed of the voice; for text data, it analyzes the structure and word choice of the sentences to identify the user's emotional state.

[1501] Risk assessment:

[1502] Based on the analysis results from the emotion engine, the server determines the risk of customer harassment in three stages: high, medium, and low. If the risk level is determined to be medium or high, immediate action is taken.

[1503] Generating warnings and feedback:

[1504] Depending on the risk level, the server sends a warning message to the user's terminal. The warning includes the risk level and specific improvement suggestions, and appropriate alternative text is also generated and provided to the user.

[1505] Specific example

[1506] Preventing customer harassment in call centers

[1507] 1. When a user (employee) is providing chat support at a call center, they receive emotional messages.

[1508] Example: "Why are you taking so long to respond? Hurry up!"

[1509] 2. The employee's device sends the chat content to the server.

[1510] Sent text: "Why are you taking so long to respond? Hurry up!"

[1511] 3. The server analyzes the chat content, and the emotion engine recognizes the recipient's (customer's) emotions.

[1512] Emotional analysis results: High level of anger.

[1513] 4. The server determines the risk of customer harassment.

[1514] Judgment result: High risk

[1515] 5. The server sends a warning message and feedback to the employee's terminal.

[1516] Warning message: "There is a high risk. Please consider how to mitigate customer anger."

[1517] Feedback: "Explain the situation carefully and patiently until the customer is satisfied."

[1518] 6. The server generates an appropriate alternative sentence and provides it to the employee.

[1519] Alternative suggestion: "We apologize for the delay. We are currently working to process your request as quickly as possible, so please wait a little longer."

[1520] Examples of prompts to input into a generative AI model

[1521] "In customer service, I received a message that conveyed anger. What would be an appropriate response? For example, please suggest a response to the following message."

[1522] In this way, this system can significantly reduce the risk of customer harassment by assessing risks in real time and prompting users to take appropriate action immediately.

[1523] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1524] Step 1: Data Collection

[1525] Input: Data from conversations, emails, and chats that users have with smartphones or robots.

[1526] Specific operation: The user initiates communication, and voice and text data are generated.

[1527] Data processing: The audio data is converted to text data using Google Cloud Speech-to-Text. The text data is then sent directly to the server.

[1528] Output: Text-formatted conversation data, chat data, and email data are sent to the server.

[1529] Step 2: Forecast Analysis

[1530] Input: Text data collected in Step 1.

[1531] Specific operation: The server inputs the collected data into IBM Watson Natural Language Understanding or Amazon Comprehend Sentiment Analysis for real-time analysis.

[1532] Data processing: This involves detecting keywords within text data and performing overall contextual analysis. It also uses an emotion engine to analyze the user's emotional state.

[1533] Output: Sentiment analysis results and a list of specific keywords.

[1534] Step 3: Emotion Recognition

[1535] Input: Sentiment analysis results and specific keyword list analyzed in Step 2.

[1536] Specific operation: The server uses the results of the emotion engine to determine the user's emotional state. In particular, it analyzes the intensity and trend of emotions.

[1537] Data processing: Based on the emotion analysis results, the user's emotional state is classified into categories such as "anger," "sadness," and "joy."

[1538] Output: Classified emotion state data.

[1539] Step 4: Risk Assessment

[1540] Input: Emotional state data and specific keyword list obtained in Step 3.

[1541] Specific operation: The server determines the risk of customer harassment based on the collected data. The risk level is evaluated in three stages: "high," "medium," and "low."

[1542] Data calculation: Calculate harassment risk based on the intensity of emotions and the frequency of specific keywords.

[1543] Output: Risk assessment result (high, medium, low).

[1544] Step 5: Generating warnings and feedback

[1545] Input: Risk assessment results (high, medium, low) and emotional state data.

[1546] Specific operation: Based on the determination result, the server sends a warning message to the user's terminal.

[1547] Data processing: Warning messages will include risk levels along with specific improvement suggestions. Additionally, a generative AI model will be used to generate appropriate alternative text.

[1548] Output: Warning message, improvement suggestion, and appropriate alternative text.

[1549] Step 6: Suggesting alternative text

[1550] Input: Improvement suggestions and alternative text.

[1551] Specific operation: The server uses a generative AI model to provide the user with appropriate alternative text.

[1552] Output: The provided alternative text will be displayed on the user's terminal.

[1553] This allows the entire system to assess the risk of customer harassment in real time and promptly prompt users to take appropriate action.

[1554] 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.

[1555] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). An 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.

[1556] 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.

[1557] [Fourth Embodiment]

[1558] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1559] 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.

[1560] 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).

[1561] 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.

[1562] 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.

[1563] 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).

[1564] 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.

[1565] 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.

[1566] 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.

[1567] 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.

[1568] 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.

[1569] 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.

[1570] 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".

[1571] overview

[1572] This invention is a system that collects conversation content, emails, or chats in real time, analyzes the risk of customer harassment, and issues warnings. Furthermore, it is characterized by generating appropriate feedback and alternative text along with the warning and providing them to the user.

[1573] System Configuration

[1574] 1. User's terminal

[1575] It collects data from conversations, emails, and chats.

[1576] Includes speech recognition functionality for converting audio data into text data.

[1577] It communicates with the server and sends data.

[1578] 2. Server

[1579] Receive data sent from the device.

[1580] The system analyzes data in real time to determine the risk of customer harassment.

[1581] Based on the judgment, a warning message and feedback are generated.

[1582] Generate appropriate alternative text and send it to the user's device.

[1583] Program processing

[1584] 1. Data Collection

[1585] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[1586] 2. Real-time analysis

[1587] The server analyzes the received text data and assesses the risk of customer harassment. This assessment includes sentiment analysis and detection of specific keywords. Based on the analysis results, a risk level is determined.

[1588] 3. Generating warnings and feedback

[1589] If the server determines that there is a high risk of customer harassment, it will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[1590] Specific example

[1591] Example: Detecting customer harassment via email

[1592] 1. The user composes an emotional email and clicks the send button.

[1593] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1594] 2. The device sends the email body to the server.

[1595] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[1596] 3. The server analyzes the content of the email and determines the risk of customer harassment.

[1597] Judgment result: High risk

[1598] 4. The server sends feedback to the user's terminal along with a warning message.

[1599] Warning message: "Suspected customer harassment. Risk level: High"

[1600] Feedback: "This kind of expression should be avoided."

[1601] 5. The server generates an appropriate alternative sentence and provides it to the user.

[1602] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[1603] Example: Detecting customer harassment in chat

[1604] 1. The user types an aggressive message in the chat and clicks the send button.

[1605] Example: "Your response is too slow. What are you doing?"

[1606] 2. The device sends the chat content to the server.

[1607] Sent text: "Your response is too slow. What are you doing?"

[1608] 3. The server analyzes the chat content and assesses the risk of customer harassment.

[1609] Assessment result: Medium risk

[1610] 4. The server sends a warning and feedback to the user's terminal.

[1611] Warning message: "Caution is advised. Risk level: Medium"

[1612] Feedback: "This expression may be considered rude."

[1613] 5. The server generates an appropriate alternative sentence and provides it to the user.

[1614] Alternative suggestion: "Please respond promptly if any problems arise."

[1615] In this way, this system can assess the risk of customer harassment in real time and provide appropriate feedback and alternative text, thereby promoting proper communication between companies and customers and reducing the psychological burden on employees.

[1616] The following describes the processing flow.

[1617] Step 1:

[1618] The user begins typing an email, chat message, or voice call. The device prepares to collect this data.

[1619] Step 2:

[1620] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[1621] Step 3:

[1622] The device collects email body text, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[1623] Step 4:

[1624] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[1625] Step 5:

[1626] The server inputs the received data into an analysis tool. This analysis tool includes sentiment analysis and specific keyword detection algorithms.

[1627] Step 6:

[1628] The server uses analysis tools to analyze incoming data in real time. As a result, the risk level of customer harassment is determined. The risk level is evaluated in three stages: high, medium, and low.

[1629] Step 7:

[1630] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[1631] Step 8:

[1632] The server generates appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[1633] Step 9:

[1634] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[1635] Step 10:

[1636] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[1637] Step 11:

[1638] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[1639] In this way, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions.

[1640] (Example 1)

[1641] 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".

[1642] In today's business environment, communication between companies and customers takes place through a variety of means. However, excessive demands and inappropriate remarks from customers, i.e., customer harassment, can increase the psychological burden on employees and lead to a deterioration of the working environment. Therefore, it is necessary to monitor conversation content in real time, detect the risk of customer harassment early, and take appropriate measures. However, current systems have problems such as delays in analyzing collected data and inability to accurately assess risks.

[1643] 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.

[1644] In this invention, the server includes means for collecting conversation content, electronic messages, or instant messages in real time; means for analyzing the collected data and evaluating the risk of customer harassment; and means for issuing warnings based on the evaluation results. This makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[1645] "Conversation content" refers to the content of communication conducted through voice.

[1646] "Electronic messages" refer to messages sent and received in digital format, such as emails and text messages.

[1647] "Instant messaging" refers to real-time chats and instant messages.

[1648] "Methods for collecting information in real time" refers to technologies for collecting conversation content and messages almost instantly.

[1649] "Analysis" refers to the process of evaluating collected data using mechanical and algorithmic methods to extract meaning and emotion.

[1650] "Customer harassment" refers to acts that cause psychological distress to employees through excessive demands or inappropriate remarks from customers.

[1651] "Means of risk assessment" refers to technologies used to determine the risk level of customer harassment through data analysis.

[1652] "Warning mechanisms" refer to technologies used to send warning messages to users when there is a risk.

[1653] "Speech recognition means" refers to technology for converting speech data into text data.

[1654] A "machine intelligence model" refers to a technology that uses artificial intelligence to analyze data and recognize specific patterns or emotions.

[1655] "Alternative text" refers to appropriate wording that is recommended to be used instead of the original message.

[1656] "Feedback" refers to suggestions, proposals, or specific advice regarding a warning.

[1657] The system in this invention is built to collect conversation content, electronic messages, or instant messages in real time, determine the risk of customer harassment, and issue warnings. This system includes a user terminal, a server, and a generative AI model.

[1658] 1. User terminal

[1659] The user terminal is a device that collects data from conversations, electronic messages, or instant messages. It also has a speech recognition system that converts voice data into text data. Speech recognition software such as "Google Cloud Speech-to-Text" is used for speech recognition. It also has the function to collect text data entered by the user and send it to the server.

[1660] Specific example: A user types the message "You can't even fulfill such a simple request? Do you even have any motivation?" into the chat window and clicks the send button.

[1661] 2. Server

[1662] The server is a central processing unit that analyzes text data received from user terminals. It analyzes the received data in real time and assesses the risk of customer harassment. The analysis uses natural language processing (NLP) technologies such as "SpaCy" and "IBM Watson" for sentiment analysis and detection of specific keywords. Based on the analysis results, it determines the risk level and generates warning messages and feedback as needed. Furthermore, it uses generative AI models (e.g., OpenAI's GPT-3) to generate appropriate alternative text.

[1663] Specific example: The server receives a message saying, "You can't even fulfill such a simple request? Are you even motivated?" and the analysis results indicate it's "high risk." Subsequently, it generates a warning message, "Suspected customer harassment. Risk level: high," and appropriate feedback, "Such language should be avoided," and creates alternative text, "Is there anything you didn't understand in this section? I'd be happy to explain in more detail."

[1664] Example of a prompt

[1665] A user is attempting to send the following message. Please check if this message poses a risk of customer harassment and generate a warning message and appropriate alternative text. Original message: "You can't even fulfill such a simple request? Are you even motivated?"

[1666] The above describes the embodiments for carrying out the present invention. This system makes it possible to detect the risk of customer harassment in real time and take immediate countermeasures.

[1667] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1668] Step 1:

[1669] The user creates a conversation, e-message, or instant message and clicks the send button. The user's device then collects the created message.

[1670] Input: User-entered conversation, e-message, or instant message (e.g., "You can't even fulfill such a simple request? Are you even motivated?")

[1671] Output: Message data stored on the user's terminal

[1672] Specific action: The user types a message in the chat window and clicks the send button.

[1673] Step 2:

[1674] The user terminal prepares to send the collected message data to the server in real time. If voice data is included, speech recognition is used to convert the voice to text.

[1675] Input: Voice data or text data

[1676] Output: Text data

[1677] Specific operation: A speech recognition engine (e.g., Google Cloud Speech-to-Text) converts speech to text. If the data is text, it is used as is.

[1678] Step 3:

[1679] The user terminal sends text data to the server.

[1680] Input: Text data (Example: "You can't even meet such a simple request? Do you even have any motivation?")

[1681] Output: Text data sent to the server

[1682] Specific operation: The user terminal uses its communication function to send the converted text data to the server.

[1683] Step 4:

[1684] The server analyzes the received text data. The server uses natural language processing technologies (e.g., SpaCy, IBM Watson) to perform sentiment analysis and detect specific keywords.

[1685] Input: Text data

[1686] Output: Analysis results (sentiment score, keyword detection results)

[1687] Specific operation: The server uses SpaCy or IBM Watson to analyze the sentiment of incoming messages and detect specific keywords.

[1688] Step 5:

[1689] The server assesses the risk of customer harassment based on the analysis results. This assessment includes sentiment scores and keyword detection results.

[1690] Input: Analysis results (sentiment score, keyword detection results)

[1691] Output: Risk assessment results (high risk, medium risk, low risk, etc.)

[1692] Specific operation: The server evaluates the risk level of customer harassment based on sentiment scores and keyword detection results (e.g., determines it as "high risk").

[1693] Step 6:

[1694] The server generates a warning message based on the risk assessment results. If the risk is high, it also generates specific improvement suggestions and alternative text. A generative AI model is used to create appropriate feedback and alternative text.

[1695] Input: Risk assessment results

[1696] Output: Warning message, feedback, alternative text

[1697] Specific actions: The server generates a warning message "Suspected customer harassment. Risk level: High," feedback "Such language should be avoided," and alternative text "Is there anything you don't understand about this part? We'd be happy to explain in more detail."

[1698] Step 7:

[1699] The server sends generated warning messages, feedback, and alternative text to the user's terminal.

[1700] Input: Warning message, feedback, alternative text

[1701] Output: Warning messages, feedback, and alternative text sent to the user's terminal.

[1702] Specific operation: The server uses its communication function to send warning messages, feedback, and alternative text to the user's terminal.

[1703] Through the above processing steps, this system can assess the risk of customer harassment in real time and immediately provide necessary countermeasures.

[1704] (Application Example 1)

[1705] 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".

[1706] Customer support and call center staff are required to detect the risk of customer harassment in real time during customer conversations and to take appropriate action quickly based on that risk. Traditional methods require staff to assess the risk and devise countermeasures themselves, which places a significant psychological burden on them and carries the risk of incorrect responses. This challenge needs to be addressed.

[1707] 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.

[1708] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing a warning based on the determination result; means for providing feedback on the warning and generating appropriate alternative text; means for collecting conversation content using speech recognition means that convert speech to text in real time; means for determining the risk of customer harassment using a device that performs sentiment analysis; means for determining the risk level based on the determination result and presenting an appropriate alert; means for displaying feedback on a smart device worn by an employee; and natural language processing means for generating appropriate alternative text. This enables staff to detect the risk of customer harassment in real time and take appropriate action quickly.

[1709] "Conversation content" refers to voice-based communication between customer support or call center staff and customers.

[1710] "Email" refers to messages sent and received electronically via the internet.

[1711] "Chat" refers to real-time text-based communication.

[1712] "Means of real-time collection" refers to methods or devices for instantly acquiring this data as conversations, emails, or chats occur.

[1713] "Means of analyzing collected data" refers to methods or devices for analyzing collected conversations, emails, or chats to assess the risk of customer harassment.

[1714] "Means for determining the risk of customer harassment" refers to a method or device that detects inappropriate words or actions from customers and determines their risk level based on collected and analyzed data.

[1715] "Means of issuing warnings" refers to methods or devices for sending appropriate notifications or warning messages to users when a risk of customer harassment is identified.

[1716] "Feedback" or "means for generating appropriate alternative text" refers to a method or device for generating and proposing appropriate countermeasures or alternative expressions to the user in response to the risk of customer harassment.

[1717] "Speech recognition means" refers to technologies and devices for converting speech data into text data.

[1718] "An emotional analysis device" refers to a method or device for analyzing the emotional aspects of text data and detecting specific emotions or tones.

[1719] "Means for determining risk levels" refers to methods or devices for evaluating the degree of customer harassment risk and quantifying that level based on the results of sentiment analysis.

[1720] "Means of providing appropriate alerts" refers to methods or devices for providing warnings or notifications to employees based on the determined risk level.

[1721] "Means of displaying feedback on smart devices" refers to methods or devices for displaying risk levels or feedback messages on wearable devices or other smart devices worn by users.

[1722] "Natural language processing means" refers to language processing technologies and devices that understand text data and generate appropriate alternative text.

[1723] This invention provides a system for detecting the risk of customer harassment in real time in customer support and call centers, and for taking appropriate action quickly based on that risk. The system includes user terminals, servers, and smart devices worn by employees.

[1724] System Configuration

[1725] 1. User's terminal

[1726] It has the ability to collect data from conversations, emails, and chats.

[1727] It is equipped with speech recognition means for converting audio data into text data.

[1728] It has the function of communicating with a server and sending data.

[1729] 2. Server

[1730] It has the ability to receive data sent from a device and analyze it in real time.

[1731] This includes an emotion analysis device for analyzing data and determining the risk of customer harassment, as well as an artificial intelligence model for detecting specific keywords.

[1732] Based on the assessment results, the risk level is determined, and warning and feedback messages are generated.

[1733] The generated warnings and feedback are sent to the user's device and to smart devices that display the feedback.

[1734] 3. Smart devices

[1735] Employees wear the device, which displays warning and feedback messages sent from the server.

[1736] Hardware and software to be used

[1737] Speech recognition method: The "speech_recognition" library and Google's speech recognition API are used.

[1738] Sentiment analysis device: Uses a pre-trained sentiment classification model (such as BERT) using the "transformers" library.

[1739] Smart devices: Employees can wear smart glasses or use smartphones.

[1740] Servers: High-performance servers are used for processing and communicating this data.

[1741] Processing flow

[1742] 1. Data collection:

[1743] When a user initiates a conversation or chat with a customer, the device collects conversation and chat data in real time, and the voice data is converted into text data using speech recognition technology.

[1744] 2. Real-time analysis:

[1745] The collected text data is sent to a server, which uses sentiment analysis devices and artificial intelligence models to analyze the risk of customer harassment.

[1746] 3. Generating warnings and feedback:

[1747] If the risk level of customer harassment is high, the server will generate a warning and specific feedback message, or appropriate alternative text, depending on the risk level.

[1748] 4. Notifications and Feedback:

[1749] The generated warning and feedback messages are sent to and displayed on the user's device and the employee's smart device, allowing employees to take appropriate action immediately.

[1750] Specific example

[1751] For example, if a support staff member is talking to a customer and the customer makes an aggressive statement such as, "Your response is too slow! What's going on?", this system collects the statement in real time and converts it to text using speech recognition. Then, an emotion analyzer analyzes this text, and if it is determined to be high risk, the server generates a warning message and feedback such as, "Specifically, what is the problem? We'd be happy to help," which is displayed on the staff member's smart glasses.

[1752] Prompts for Generative AI Models

[1753] Design a custom-made smart glasses application to detect customer harassment during customer support conversations and provide appropriate alerts and feedback in real time. Generate Python pseudocode that meets the following requirements.

[1754] 1. Collection of audio data

[1755] 2. Text conversion using speech recognition

[1756] 3. Risk assessment using sentiment analysis

[1757] 4. Generate feedback based on the judgment result.

[1758] The above describes the embodiments for carrying out the present invention. By introducing this system, it becomes possible to detect customer harassment in real time at customer support sites and take prompt and appropriate action, thereby reducing the psychological burden on employees.

[1759] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1760] Step 1:

[1761] The user initiates a conversation with a customer. The input is the voice conversation with the customer. The device collects this voice data. The device captures the voice data in real time using its built-in microphone.

[1762] Step 2:

[1763] The device converts the collected audio data into text data. The input is the audio data collected in step 1. The device uses its speech recognition capabilities (using the speech_recognition library and Google's speech recognition API) to convert the audio data into text data. The output is text data.

[1764] Step 3:

[1765] The text data is sent to the server. The input is the text data generated in step 2. The terminal communicates with the server and sends the text data to the server. The output is that the text data is sent to the server.

[1766] Step 4:

[1767] The server analyzes the received text data. The input is the text data sent to the server in step 3. The server uses a sentiment analyzer (using the transformers library) and an artificial intelligence model that detects specific keywords to perform sentiment analysis and keyword detection on the text data. The output is the risk assessment result.

[1768] Step 5:

[1769] The server determines the risk level based on the risk assessment results and generates warnings and feedback. The input is the risk assessment results obtained in step 4. The server determines the risk level (high, medium, low) according to the assessment results and generates appropriate warning messages and feedback messages, as well as appropriate alternative text using natural language processing. The output is the generated warning messages and feedback messages.

[1770] Step 6:

[1771] The server sends the generated warning and feedback messages to the user's terminal and smart device. The input is the warning and feedback messages generated in step 5. The server sends these messages to the terminal and smart device via communication. The output is the warning and feedback displayed on the user's terminal and smart device.

[1772] Step 7:

[1773] The user's terminal and smart device display warnings and feedback. The input consists of the warning and feedback messages sent from the server in step 6. The terminal and smart device use their display functions to present the risk level, warnings, feedback, and alternative text to the user. The output is the visually displayed message.

[1774] The above outlines the processing steps of a system that detects customer harassment risks in real time and provides appropriate feedback.

[1775] 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.

[1776] overview

[1777] This invention relates to a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it provides more accurate risk assessment and feedback. This system analyzes user emotions in real time and proposes appropriate measures to reduce the risk of customer harassment.

[1778] System Configuration

[1779] 1. User's terminal

[1780] It collects data from conversations, emails, and chats.

[1781] Includes speech recognition functionality to convert audio data into text data.

[1782] It communicates with the server and sends data.

[1783] 2. Server

[1784] Receive data sent from the device.

[1785] The system analyzes data in real time to determine the risk of customer harassment.

[1786] We use an emotion engine that recognizes user emotions to perform emotional analysis.

[1787] Based on the judgment result, a warning message and feedback are generated.

[1788] Generate appropriate alternative text and send it to the user's device.

[1789] Program processing

[1790] 1. Data Collection

[1791] When a user composes an email and clicks the send button, the user's device extracts the email body and sends it to the server. Similarly, chat messages and audio data from conversations are collected, converted to text, and then sent to the server.

[1792] 2. Real-time analysis

[1793] The server inputs the received data into an analysis tool. The analysis tool includes sentiment analysis and specific keyword detection algorithms, and in addition, it operates an emotion engine that recognizes the user's emotions.

[1794] 3. Emotion recognition

[1795] The emotion engine analyzes the user's emotions from incoming data. For audio data, it analyzes tone and speed of voice; for text data, it analyzes context and word choice to determine whether the user is emotional.

[1796] 4. Risk Assessment

[1797] The server determines the risk level of customer harassment based on analysis results, including those from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[1798] 5. Generating warnings and feedback

[1799] If the risk level is determined to be high or medium, the server will send a warning message to the user's terminal. This message will include the risk level and specific suggestions for improvement. Appropriate alternative text will also be generated and provided to the user.

[1800] Specific example

[1801] Example: Detecting customer harassment via email

[1802] 1. The user composes an emotional email and clicks the send button.

[1803] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1804] 2. The device sends the email body to the server.

[1805] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[1806] 3. The server analyzes the email content, and the emotion engine recognizes the user's emotions.

[1807] Emotional analysis results: High level of anger.

[1808] 4. The server determines the risk of customer harassment based on the overall analysis results.

[1809] Judgment result: High risk

[1810] 5. The server sends feedback to the user's terminal along with a warning message.

[1811] Warning message: "Suspected customer harassment. Risk level: High"

[1812] Feedback: "This kind of expression should be avoided."

[1813] 6. The server generates an appropriate alternative sentence and provides it to the user.

[1814] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[1815] Example: Detecting customer harassment in chat

[1816] 1. The user types an aggressive message in the chat and clicks the send button.

[1817] Example: "Your response is too slow. What are you doing?"

[1818] 2. The device sends the chat content to the server.

[1819] Sent text: "Your response is too slow. What are you doing?"

[1820] 3. The server analyzes the chat content, and the emotion engine recognizes the user's emotions.

[1821] Emotional analysis results: High level of irritation.

[1822] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[1823] Assessment result: Medium risk

[1824] 5. The server sends a warning and feedback to the user's terminal.

[1825] Warning message: "Caution is advised. Risk level: Medium"

[1826] Feedback: "This expression may be considered rude."

[1827] 6. The server generates an appropriate alternative sentence and provides it to the user.

[1828] Alternative suggestion: "Please respond promptly if any problems arise."

[1829] In this way, by combining an emotion engine, this system accurately grasps the user's emotions, determines the risk of customer harassment in real time, and provides appropriate feedback and alternative expressions. This optimizes communication between companies and customers and reduces the psychological burden on employees.

[1830] The following describes the processing flow.

[1831] Step 1:

[1832] The user initiates an email, chat message, or conversation. The device prepares to collect this data in real time.

[1833] Step 2:

[1834] The user composes an email and clicks the send button. Alternatively, the user types a chat message and clicks the send button. In the case of a voice call, the conversation begins.

[1835] Step 3:

[1836] The device collects email content, chat messages, or audio data. Audio data is converted to text data using speech recognition technology.

[1837] Step 4:

[1838] The collected data (text and converted speech-to-text) is sent from the terminal to the server. The server receives the data.

[1839] Step 5:

[1840] The server inputs the received data into the analysis tool. The analysis tool includes a sentiment analysis engine and a specific keyword detection algorithm.

[1841] Step 6:

[1842] The emotion engine analyzes the user's emotions. Specifically, it recognizes emotions based on tone and speed of voice in the case of audio data, and on context and word choice in the case of text data.

[1843] Step 7:

[1844] The server integrates analysis results, including those from the emotion engine, to determine the risk level of customer harassment. The risk level is evaluated in three stages: high, medium, and low.

[1845] Step 8:

[1846] If the risk level is determined to be high or medium, the server will generate a warning message and feedback. The feedback will include specific suggestions for improvement and points to note.

[1847] Step 9:

[1848] The server generates a warning message along with appropriate alternative text to reduce the risk of customer harassment. This alternative text replaces the original expression with a gentler and more appropriate expression.

[1849] Step 10:

[1850] The generated warning message, feedback, and alternative text are sent from the server to the user's terminal.

[1851] Step 11:

[1852] The system displays warning messages, feedback, and alternative text received by the user's device. The user can review this and make corrections as needed.

[1853] Step 12:

[1854] The user resends the corrected email or chat message. In the case of a voice call, the user incorporates appropriate feedback and resumes the conversation.

[1855] As described above, this system detects the risk of customer harassment in real time through multiple steps and provides appropriate feedback and alternative expressions. By combining it with an emotion engine, it can accurately grasp the user's emotions and enable more accurate risk assessment and countermeasures.

[1856] (Example 2)

[1857] 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".

[1858] There is a problem of increasing psychological burden on employees due to customer harassment. Furthermore, it is difficult to determine the risk of customer harassment in real time, making it difficult to take appropriate action and provide feedback. To solve this problem, a system is needed that analyzes user emotions in real time and proposes quick and appropriate countermeasures.

[1859] 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.

[1860] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for analyzing the user's emotions using an emotion engine; means for evaluating the risk level in three stages: high, medium, and low; and means for inputting the data into an analysis tool. This makes it possible to evaluate the risk of customer harassment in real time and provide rapid and appropriate feedback and alternative text.

[1861] "Conversation content" refers to all communication content in audio and text format, primarily including information in dialogue and chat formats.

[1862] "Email" refers to messages containing data such as text, images, and audio that are sent and received over the internet.

[1863] "Chat" refers to online conversational communication where text messages are exchanged in real time.

[1864] "Real-time data collection" means collecting data simultaneously with user actions and processing it without delay.

[1865] "Data analysis" refers to the techniques used to process and analyze collected information to identify useful insights and risks.

[1866] "Customer harassment" refers to acts that cause employees psychological distress due to excessive demands or intimidating behavior from customers.

[1867] "Risk assessment" refers to the process of evaluating whether a particular action or statement carries a risk of customer harassment through sentiment analysis and keyword detection.

[1868] A "warning" is a message from the system to alert the user, and may include feedback based on the risk level.

[1869] "Feedback" refers to evaluations and suggestions for improvement regarding a user's actions and comments.

[1870] "Alternative text" refers to improved versions of the original statement, suggested to avoid customer harassment.

[1871] An "emotion engine" refers to an algorithm or model used to analyze a user's emotional state, sensing emotions based on voice and text data.

[1872] "Risk level" refers to a three-tiered evaluation (high, medium, low) that indicates how high the risk of customer harassment is.

[1873] "Analysis tools" refer to software and algorithms used for data analysis and sentiment analysis.

[1874] Modes for carrying out the invention

[1875] System Overview

[1876] This invention is a system that collects conversation content, emails, or chats in real time and analyzes the risk of customer harassment. Furthermore, by combining it with an emotion engine that recognizes user emotions, it can provide more accurate risk assessment and feedback. This system can analyze user emotions in real time and propose appropriate measures to reduce the risk of customer harassment.

[1877] System Configuration

[1878] 1. User's device:

[1879] This includes means of collecting data from conversations, emails, and chats.

[1880] Includes speech recognition means for converting audio data into text data.

[1881] Includes means for communicating with a server and transmitting data.

[1882] 2. Server:

[1883] Includes means for receiving data transmitted from a terminal.

[1884] This includes methods for analyzing data in real time and determining the risk of customer harassment.

[1885] This includes means for performing emotional analysis using an emotion engine that recognizes the user's emotions.

[1886] Includes means for generating warning messages and feedback based on the judgment result.

[1887] Includes means for generating appropriate alternative text and sending it to the user's terminal.

[1888] Specific hardware and software to be used

[1889] We will use "Google Cloud Speech-to-Text API" or "IBM Watson Speech to Text" as speech recognition software.

[1890] We use Google Cloud Natural Language API and IBM Watson Natural Language Understanding for sentiment analysis and text analysis.

[1891] Standard server and cloud computing environments are used for data processing and analysis.

[1892] Specific examples of operation

[1893] Example 1: Detecting customer harassment via email

[1894] 1. The user composes an emotional email and clicks the send button.

[1895] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1896] 2. The device sends the email body to the server.

[1897] Sent text: "You can't even meet such a simple request? Do you even have any motivation?"

[1898] 3. The server sends the email content to the Google Cloud Natural Language API for sentiment analysis.

[1899] Emotional analysis results: High level of anger.

[1900] 4. The server determines the risk of customer harassment based on the overall analysis results.

[1901] Judgment result: High risk

[1902] 5. The server sends feedback to the user's terminal along with a warning message.

[1903] Warning message: "Suspected customer harassment. Risk level: High"

[1904] Feedback: "This kind of expression should be avoided."

[1905] 6. The server generates appropriate alternative text and provides it to the user.

[1906] Alternative suggestion: "Is there anything you don't understand about this part? I'd be happy to explain in more detail."

[1907] Example 2: Detection of customer harassment in chat

[1908] 1. The user types an aggressive message in the chat and clicks the send button.

[1909] Example: "Your response is too slow. What are you doing?"

[1910] 2. The device sends the chat content to the server.

[1911] Sent text: "Your response is too slow. What are you doing?"

[1912] 3. The server sends the chat content to "IBM Watson Natural Language Understanding" for sentiment analysis.

[1913] Emotional analysis results: High level of irritation.

[1914] 4. The server assesses the risk of customer harassment based on the overall analysis results.

[1915] Assessment result: Medium risk

[1916] 5. The server sends a warning and feedback to the user's terminal.

[1917] Warning message: "Caution is advised. Risk level: Medium"

[1918] Feedback: "This expression may be considered rude."

[1919] 6. The server generates appropriate alternative text and provides it to the user.

[1920] Alternative suggestion: "Please respond promptly if any problems arise."

[1921] Example of a prompt

[1922] Examples of prompt statements to input into the generative AI model are as follows:

[1923] User-generated email: "You can't even fulfill such a simple request? Do you even have any motivation?"

[1924] Please suggest the feedback and alternative sentences that the system will generate.

[1925] conclusion

[1926] This system optimizes communication between companies and customers, reduces the psychological burden on employees, and effectively manages the potential risk of customer harassment.

[1927] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1928] Step 1: Data Collection

[1929] The user composes an email or chat message and clicks the send button.

[1930] Example: "You can't even meet such a simple request? Do you even have any motivation?"

[1931] The device collects user input data and extracts email content and chat messages. If audio data is present, it is converted into text using speech recognition technology.

[1932] Input: User's email or chat message

[1933] Output: Text data ("You can't even meet such a simple request? Do you even have any motivation?")

[1934] Step 2: Data transmission

[1935] The device sends the collected data to the server.

[1936] Input: Text data

[1937] Output: Sending data to the server

[1938] Step 3: Data Reception

[1939] The server receives data sent from the terminal.

[1940] Input: Text data from the terminal

[1941] Output: Received data

[1942] Step 4: Emotion Analysis

[1943] The server inputs the received data into an analysis tool. The analysis tool uses either "Google Cloud Natural Language API" or "IBM Watson Natural Language Understanding".

[1944] Input: Received data (text data)

[1945] Data processing: An emotion engine analyzes emotions from context and word choice.

[1946] Output: Emotion analysis results (e.g., high level of anger)

[1947] Step 5: Risk Assessment

[1948] The server determines the risk level of customer harassment based on the analysis results from the emotion engine. The risk level is evaluated in three stages: high, medium, and low.

[1949] Input: Sentiment analysis results

[1950] Data processing: Assess risk based on the intensity of emotions and specific keywords.

[1951] Output: Risk level (high, medium, low)

[1952] Step 6: Generate warning message

[1953] If the server is determined to have a high or medium risk level, it will generate a warning message and feedback.

[1954] Input: Risk level

[1955] Data processing: Generate warning messages and feedback based on risk level.

[1956] Output: Warning messages and feedback

[1957] Step 7: Send a warning message

[1958] The server sends the generated warning message and feedback to the user's terminal.

[1959] Input: Warning messages and feedback

[1960] Output: Sent to the user's terminal

[1961] Step 8: Generate alternative text

[1962] The server generates appropriate alternative text and provides it to the user.

[1963] Input: Risk level and sentiment analysis results

[1964] Data processing: Generate appropriate and gentle alternative text based on sentiment analysis results.

[1965] Output: Alternative text suggestions

[1966] Step 9: Send alternative text

[1967] The server sends the generated alternative text to the user's terminal.

[1968] Input: Alternative text

[1969] Output: Sent to the user's terminal

[1970] Specific example:

[1971] When a user sends a message via chat saying, "Your response is too slow. What are you doing?",

[1972] 1. The device collects messages and sends them to the server.

[1973] 2. The server receives the data, performs sentiment analysis, and determines that the level of irritation is high.

[1974] 3. The server assesses the risk as moderate and generates a warning message and feedback indicating that attention is needed.

[1975] 4. The server sends a warning message ("Caution is advised. Risk level: Medium") and feedback ("This language may be offensive.") to the user.

[1976] 5. The server generates alternative text ("Please respond promptly if a problem occurs.") and provides it to the user.

[1977] In this way, the system assesses the risk of customer harassment at each step and provides appropriate responses and feedback.

[1978] (Application Example 2)

[1979] 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".

[1980] In modern businesses, customer harassment is on the rise during customer service interactions, negatively impacting employee psychological burden and overall business efficiency. To prevent such harassment, a system is needed that monitors customer interactions in real time and takes appropriate action. However, conventional systems struggle to accurately grasp user emotions and provide instant feedback, and they also lack sufficient suggestions for appropriate alternative expressions to mitigate risks. As a result, effective harassment prevention remains difficult, and more sophisticated systems are required.

[1981] 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.

[1982] In this invention, the server includes means for collecting conversation content, emails, or chats in real time; means for analyzing the collected data and determining the risk of customer harassment; means for issuing warnings based on the determination results; means for providing feedback on the warnings and generating appropriate alternative text; means for analyzing the user's emotions and recognizing their emotional state; and means for sending improvement suggestions when the risk of customer harassment exceeds a certain level. This enables real-time determination of the risk of customer harassment, prompt warnings to employees, appropriate feedback, and suggestions for alternative expressions.

[1983] "Conversation content" refers to the words and expressions used when exchanging information in audio or text format.

[1984] "Email" refers to messages that are sent and received electronically via networks such as the internet.

[1985] "Chat" refers to real-time, text-based communication.

[1986] "Means of real-time collection" refers to devices or programs that have the function of collecting conversation content and text messages and sending them to a server.

[1987] "Means for analyzing collected data" refers to a system that has the function of analyzing collected information and detecting specific patterns or keywords.

[1988] "Means for determining the risk of customer harassment" refers to a system that has processes and technologies for evaluating the possibility of harassment in interactions with customers.

[1989] "Means of issuing warnings" refers to a system that has the function of sending alerts and warnings to users based on the results of risk assessment.

[1990] "Means of generating feedback and appropriate alternative text for warnings" refers to a system that has the function of suggesting directions for improvement or appropriate wording to the user.

[1991] "Means for analyzing user emotions and recognizing emotional states" refers to a system that possesses technologies and processes for performing emotion analysis on collected data and identifying the user's emotional state.

[1992] "Emotional state" refers to the user's psychological state, emotions, and mood.

[1993] "A means of sending improvement suggestions when the risk of customer harassment exceeds a certain level" refers to a system that has the technology or process to notify users of specific countermeasures or alternative expressions when the risk assessment is high.

[1994] System Overview

[1995] This invention provides a system that collects conversation content, emails, or chats in real time, determines the risk of customer harassment, and provides warnings and feedback based on the results. Furthermore, by recognizing the user's emotions, it enables more accurate risk assessment and generation of appropriate alternative text. The aim of this system is to reduce the risk of harassment in everyday business communications such as conversations and chats, and to reduce the psychological burden on companies and employees.

[1996] Hardware and software to be used

[1997] Hardware:

[1998] Smartphone: Used for collecting and sending conversations, chats, and emails.

[1999] Server: Used for data analysis, risk assessment, and feedback generation.

[2000] Robots: They can also be used to collect and transmit communications in security operations.

[2001] software:

[2002] Speech Recognition: Google Cloud Speech-to-Text

[2003] Sentiment Analysis: IBM Watson Natural Language Understanding, Amazon Comprehend Sentiment Analysis

[2004] Cloud services: AWS, Google Cloud Platform

[2005] System operation

[2006] Data collection:

[2007] When users engage in conversations or chats using their smartphones or robots, the data is collected in real time. The voice data is converted into text data using speech recognition technology and sent to a server.

[2008] Real-time analysis:

[2009] The server inputs the collected text data into a sentiment analysis tool, checking for specific keywords and contexts in real time. This helps identify the risk of customer harassment.

[2010] Emotion recognition:

[2011] The emotion engine is used to analyze the user's emotional state. For audio data, it analyzes the tone and speed of the voice; for text data, it analyzes the structure and word choice of the sentences to identify the user's emotional state.

[2012] Risk assessment:

[2013] Based on the analysis results from the emotion engine, the server determines the risk of customer harassment in three stages: high, medium, and low. If the risk level is determined to be medium or high, immediate action is taken.

[2014] Generating warnings and feedback:

[2015] Depending on the risk level, the server sends a warning message to the user's terminal. The warning includes the risk level and specific improvement suggestions, and appropriate alternative text is also generated and provided to the user.

[2016] Specific example

[2017] Preventing customer harassment in call centers

[2018] 1. When a user (employee) is providing chat support at a call center, they receive emotional messages.

[2019] Example: "Why are you taking so long to respond? Hurry up!"

[2020] 2. The employee's device sends the chat content to the server.

[2021] Sent text: "Why are you taking so long to respond? Hurry up!"

[2022] 3. The server analyzes the chat content, and the emotion engine recognizes the recipient's (customer's) emotions.

[2023] Emotional analysis results: High level of anger.

[2024] 4. The server determines the risk of customer harassment.

[2025] Judgment result: High risk

[2026] 5. The server sends a warning message and feedback to the employee's terminal.

[2027] Warning message: "There is a high risk. Please consider how to mitigate customer anger."

[2028] Feedback: "Explain the situation carefully and patiently until the customer is satisfied."

[2029] 6. The server generates an appropriate alternative sentence and provides it to the employee.

[2030] Alternative suggestion: "We apologize for the delay. We are currently working to process your request as quickly as possible, so please wait a little longer."

[2031] Examples of prompts to input into a generative AI model

[2032] "In customer service, I received a message that conveyed anger. What would be an appropriate response? For example, please suggest a response to the following message."

[2033] In this way, this system can significantly reduce the risk of customer harassment by assessing risks in real time and prompting users to take appropriate action immediately.

[2034] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[2035] Step 1: Data Collection

[2036] Input: Data from conversations, emails, and chats that users have with smartphones or robots.

[2037] Specific operation: The user initiates communication, and voice and text data are generated.

[2038] Data processing: The audio data is converted to text data using Google Cloud Speech-to-Text. The text data is then sent directly to the server.

[2039] Output: Text-formatted conversation data, chat data, and email data are sent to the server.

[2040] Step 2: Forecast Analysis

[2041] Input: Text data collected in Step 1.

[2042] Specific operation: The server inputs the collected data into IBM Watson Natural Language Understanding or Amazon Comprehend Sentiment Analysis for real-time analysis.

[2043] Data processing: This involves detecting keywords within text data and performing overall contextual analysis. It also uses an emotion engine to analyze the user's emotional state.

[2044] Output: Sentiment analysis results and a list of specific keywords.

[2045] Step 3: Emotion Recognition

[2046] Input: Sentiment analysis results and specific keyword list analyzed in Step 2.

[2047] Specific operation: The server uses the results of the emotion engine to determine the user's emotional state. In particular, it analyzes the intensity and trend of emotions.

[2048] Data processing: Based on the emotion analysis results, the user's emotional state is classified into categories such as "anger," "sadness," and "joy."

[2049] Output: Classified emotion state data.

[2050] Step 4: Risk Assessment

[2051] Input: Emotional state data and specific keyword list obtained in Step 3.

[2052] Specific operation: The server determines the risk of customer harassment based on the collected data. The risk level is evaluated in three stages: "high," "medium," and "low."

[2053] Data calculation: Calculate harassment risk based on the intensity of emotions and the frequency of specific keywords.

[2054] Output: Risk assessment result (high, medium, low).

[2055] Step 5: Generating warnings and feedback

[2056] Input: Risk assessment results (high, medium, low) and emotional state data.

[2057] Specific operation: Based on the determination result, the server sends a warning message to the user's terminal.

[2058] Data processing: Warning messages will include risk levels along with specific improvement suggestions. Additionally, a generative AI model will be used to generate appropriate alternative text.

[2059] Output: Warning message, improvement suggestion, and appropriate alternative text.

[2060] Step 6: Suggesting alternative text

[2061] Input: Improvement suggestions and alternative text.

[2062] Specific operation: The server uses a generative AI model to provide the user with appropriate alternative text.

[2063] Output: The provided alternative text will be displayed on the user's terminal.

[2064] This allows the entire system to assess the risk of customer harassment in real time and promptly prompt users to take appropriate action.

[2065] 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.

[2066] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). An 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.

[2067] 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.

[2068] 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.

[2069] 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. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, 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.

[2070] 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.

[2071] 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.

[2072] 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.

[2073] 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."

[2074] 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.

[2075] 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.

[2076] 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.

[2077] 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.

[2078] 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.

[2079] 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.

[2080] 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.

[2081] 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.

[2082] 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.

[2083] 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.

[2084] 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.

[2085] 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.

[2086] The following is further disclosed regarding the embodiments described above.

[2087] (Claim 1)

[2088] Means for collecting conversation content, emails, or chats in real time,

[2089] A means of analyzing the collected data and determining the risk of customer harassment,

[2090] A means of issuing a warning based on the judgment result,

[2091] A system including feedback on warnings and means for generating appropriate alternative text.

[2092] (Claim 2)

[2093] The system according to claim 1, wherein the means for collecting conversation content includes speech recognition means for converting speech data into text data.

[2094] (Claim 3)

[2095] The system according to claim 1, wherein the means for determining the risk of customer harassment includes an artificial intelligence model that performs sentiment analysis and detects specific keywords.

[2096] "Example 1"

[2097] (Claim 1)

[2098] Means for collecting conversation content, electronic messages, or instant messages in real time,

[2099] A means of analyzing collected data and assessing the risk of customer harassment,

[2100] A means of issuing warnings based on evaluation results,

[2101] A system including feedback on warnings and means for generating appropriate alternative text.

[2102] (Claim 2)

[2103] The system according to claim 1, wherein the means for collecting conversation content includes speech recognition means for converting speech data into text data.

[2104] (Claim 3)

[2105] The system according to claim 1, wherein the means for assessing the risk of customer harassment includes a machine intelligence model that performs sentiment analysis and detects specific keywords.

[2106] "Application Example 1"

[2107] (Claim 1)

[2108] Means for collecting conversation content, emails, or chats in real time,

[2109] A means of analyzing the collected data and determining the risk of customer harassment,

[2110] A means of issuing a warning based on the judgment result,

[2111] A means of providing feedback on warnings and generating appropriate alternative text,

[2112] A means for collecting conversation content using speech recognition means that converts speech to text in real time,

[2113] A means for determining the risk of customer harassment using an emotional analysis device,

[2114] A means of determining the risk level based on the judgment results and issuing appropriate alerts,

[2115] A means of displaying feedback on a smart device worn by an employee,

[2116] A natural language processing means for generating appropriate alternative text,

[2117] A system that includes this.

[2118] (Claim 2)

[2119] The system according to claim 1, wherein the means for collecting conversation content includes speech recognition means for converting speech data into text data.

[2120] (Claim 3)

[2121] The system according to claim 1, wherein the means for determining the risk of customer harassment includes an artificial intelligence model that performs sentiment analysis and detects specific keywords.

[2122] "Example 2 of combining an emotion engine"

[2123] (Claim 1)

[2124] Means for collecting conversation content, emails, or chats in real time,

[2125] A means of analyzing the collected data and determining the risk of customer harassment,

[2126] A means of issuing a warning based on the judgment result,

[2127] A means of providing feedback on warnings and generating appropriate alternative text,

[2128] A means of analyzing user emotions using an emotion engine,

[2129] A method for evaluating risk levels in three stages: high, medium, and low,

[2130] Methods for inputting data into the analysis tool,

[2131] A system that includes this.

[2132] (Claim 2)

[2133] The system according to claim 1, wherein the means for collecting conversation content includes speech recognition means for converting speech data into text data.

[2134] (Claim 3)

[2135] The system according to claim 1, wherein the means for determining the risk of customer harassment includes an artificial intelligence model that performs sentiment analysis and detects specific keywords.

[2136] "Application example 2 of combining emotional engines"

[2137] (Claim 1)

[2138] Means for collecting conversation content, emails, or chats in real time,

[2139] A means of analyzing the collected data and determining the risk of customer harassment,

[2140] A means of issuing a warning based on the judgment result,

[2141] A means of providing feedback on warnings and generating appropriate alternative text,

[2142] A means of analyzing user emotions and recognizing their emotional state,

[2143] A means of submitting improvement suggestions when the risk of customer harassment exceeds a certain level,

[2144] A system that includes this.

[2145] (Claim 2)

[2146] The system according to claim 1, wherein the means for collecting conversation content includes speech recognition means for converting speech data into text data.

[2147] (Claim 3)

[2148] The system according to claim 1, wherein the means for determining the risk of customer harassment includes an artificial intelligence model that performs sentiment analysis and detects specific keywords. [Explanation of Symbols]

[2149] 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. Means for collecting conversation content, emails, or chats in real time, A means of analyzing the collected data and determining the risk of customer harassment, A means of issuing a warning based on the judgment result, A system including feedback on warnings and means for generating appropriate alternative text.

2. The system according to claim 1, wherein the means for collecting conversation content includes speech recognition means for converting speech data into text data.

3. The system according to claim 1, wherein the means for determining the risk of customer harassment includes an artificial intelligence model that performs sentiment analysis and detects specific keywords.

Images