System

Abstract

A system is provided.SOLUTION: A system comprising: means for acquiring voice data; means for converting the voice data into text data; means for detecting customer harassment based on the text data; means for providing a notification when the customer harassment is detected; means for generating a detection result of the customer harassment and a call content as a report; and means for storing the report in a database.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In modern call centers, operators are expected to provide high-quality customer service on a daily basis, but psychological stress caused by customer harassment (kasuhara) has become a major problem. This psychological stress has a negative impact on the mental health and productivity of operators. Conventional systems have had difficulty effectively detecting customer harassment in real time and responding immediately. For this reason, there is a need for a system that improves the quality of customer service while reducing the mental burden on operators. [Means for solving the problem]

[0005] The present invention provides a system that includes a means for acquiring voice data and converting it into text data, a means for detecting customer harassment based on the text data, a means for notifying customers when customer harassment is detected, and a means for generating a report of the customer harassment detection results and the contents of the call and storing the report in a database. In particular, the means for detecting customer harassment is configured to identify customer harassment using specific keywords and sentiment analysis. The means for notifying customers can also include an escalation notification to an administrator. This configuration enables customers to respond to customer harassment in real time while reducing the mental burden on operators, thereby improving the quality of customer service.

[0006] "Voice data" refers to digital audio information that is a recording of the conversation between a customer and an operator.

[0007] "Text data" is digital data in which voice data is converted into text information using voice recognition technology.

[0008] "Customer harassment" refers to aggressive, insulting, or inappropriate behavior or remarks made by a customer toward an operator.

[0009] "Means for notification" refers to functions or devices that send real-time warnings or instructions to operators or administrators when customer harassment is detected.

[0010] A "report" is a written or digital document generated after a call ends that records the results of the customer harassment detection and the content of the call.

[0011] "Database" refers to an information storage system that stores generated reports and detection results and allows for searching and analysis as needed.

[0012] "Keywords" refer to offensive or inappropriate words or phrases used to identify customer harassment.

[0013] "Sentiment analysis" is the process of analyzing customer emotions (e.g., anger, irritation, etc.) from text data to determine the possibility of customer harassment.

[0014] "Escalation" is the process of reporting a problem to a higher-level manager or specialist and requesting that the problem be addressed. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0016] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0017] First, the terms used in the following description will be explained.

[0018] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0019] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0020] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0021] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0022] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0027] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0028] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0029] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0030] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0033] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0034] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0035] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0036] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. This system acquires voice data, converts it into text data using speech recognition technology, detects customer harassment using natural language processing (NLP), and has the ability to notify or escalate the situation in real time as necessary.

[0037] System configuration

[0038] 1. Acquiring audio data

[0039] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0040] 2. Voice recognition (ASR)

[0041] The server inputs the received voice data into a voice recognition engine, which converts the voice data into text data. For example, a voice saying "Hello, this is customer support" is converted into text "Hello, this is customer support."

[0042] 3. Natural Language Processing (NLP)

[0043] The server then analyzes the converted text data using advanced natural language processing (NLP) algorithms. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs further context and sentiment analysis to determine the likelihood of customer harassment.

[0044] 4. Real-time detection and notification

[0045] If customer harassment is detected, the server immediately raises a flag for response. The server then sends a notification to the device. For example, the device may display a message saying, "A customer is making inappropriate remarks. Please respond calmly." If serious customer harassment occurs, the server sends an escalation notification to an administrator so that the administrator can intervene.

[0046] 5. Report generation and data accumulation

[0047] After the call ends, the server logs all conversations and detection results, and compiles a report containing the time of the incident, the offensive keywords used, the customer sentiment analysis results, and the response. The server stores this report in a database.

[0048] 6. Data analysis and feedback

[0049] The server periodically analyzes the information stored in the database and uses it as material for training programs aimed at improving the mental health and quality of customer service, thereby reducing the mental burden on operators and enabling the continuous improvement of customer service quality.

[0050] Specific operation example

[0051] Assume that a user (operator) starts a call. The device records the contents of the call and sends them to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable," and determines that the customer is irritated through context and sentiment analysis. As this is determined to be customer harassment, the server sends a notification to the device, displaying a warning message saying, "A customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing the administrator to intervene.

[0052] After the call ends, the server logs all call content and any harassment findings, generates a report, and stores it in a database. This information is later used for operator training and support, and to improve the quality of customer service.

[0053] As described above, the present invention can detect customer harassment in real time and take appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service.

[0054] The processing flow will be explained below.

[0055] Step 1:

[0056] The user (operator) starts a call. The device detects this action and automatically starts recording the call.

[0057] Step 2:

[0058] The device transmits real-time recorded voice data to the edge server, which includes all conversations between the customer and the operator.

[0059] Step 3:

[0060] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[0061] Step 4:

[0062] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0063] Step 5:

[0064] The server evaluates the possibility of customer harassment based on the analysis results from the NLP engine. If it is determined to be customer harassment, it sets a response flag. This response flag will serve as a guide for the next action.

[0065] Step 6:

[0066] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[0067] Step 7:

[0068] If the harassment becomes serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[0069] Step 8:

[0070] The user (operator) takes appropriate action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[0071] Step 9:

[0072] The server generates a report containing the time when the customer harassment occurred, the offensive keywords used, the results of the customer sentiment analysis, and the details of the response, and stores the report in a database.

[0073] Step 10:

[0074] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[0075] Example 1

[0076] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0077] In conventional customer support systems, operators must directly respond to customer harassment, placing a heavy mental burden on the operators. Furthermore, even when customer harassment occurs, it is difficult to detect and respond in real time, which can lead to delays in appropriate escalation and reporting. This leads to a decline in the quality of customer support and a deterioration in overall service. This invention aims to reduce the mental burden on operators and improve the quality of customer support by detecting customer harassment in real time and providing appropriate notifications and escalations.

[0078] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0079] In this invention, the server includes: a means for the terminal to acquire voice data when a user initiates a call and transmit it to the server; a means for the server to input the voice data into a voice recognition engine and convert it into text data; a means for the server to use a natural language processing algorithm to detect customer harassment in real time based on the converted text data; a means for the server to notify the terminal when customer harassment is detected and, if necessary, send an escalation notice to an administrator; a means for recording the customer harassment detection results and the content of the call in a log file and generating a report; and a means for storing the generated report in a database. This enables customer harassment to be detected in real time and responded to promptly. Furthermore, by recording the content of the call and the detection results in detail and utilizing them for future analysis and feedback, the quality of customer service can be continuously improved.

[0080] A "user" is a person who acts as an operator at a call center and handles calls with customers.

[0081] A "terminal" refers to a device used by a user, including a calling device and a recording device, that acquires and transmits voice data.

[0082] The "server" refers to a central processing unit that receives voice data sent from a terminal and performs voice recognition and natural language processing.

[0083] "Voice data" refers to digital data that is a recording of the contents of a conversation between a user and a customer.

[0084] A "voice recognition engine" is software or hardware that converts voice data into text data.

[0085] "Text data" refers to character information converted from voice data by a voice recognition engine.

[0086] A "natural language processing algorithm" is a programming technology that analyzes text data and detects customer harassment based on specific keywords and context.

[0087] "Customer harassment" refers to offensive, insulting or inappropriate comments or behavior from a customer toward an operator.

[0088] "Notification" refers to a warning message or information transmission from the server to a terminal or administrator when customer harassment is detected.

[0089] An "escalation notification" is a message sent from the server to an administrator urging emergency action when customer harassment is serious.

[0090] "Log file" refers to a digital file used to record the content of calls and the results of customer harassment detection.

[0091] A "report" is a document that compiles detailed records of calls and customer harassment, and is used for future analysis and feedback.

[0092] "Database" means a digital information storage system for storing generated reports and log files for retrieval and analysis as needed.

[0093] The present invention is implemented as follows to detect customer harassment in real time in a call center, reduce the mental burden on operators, and improve the quality of customer service.

[0094] First, when a user starts a call at a call center, the terminal records the conversation and transmits the resulting voice data to the server in real time. The terminal is usually composed of a call device and a recording device provided as part of the operator tools.

[0095] The server then inputs the received voice data into a voice recognition engine, such as Google® Cloud Speech-to-Text or Microsoft® Azure® Cognitive Services. The voice recognition engine converts the voice data into text data, which the server uses for further processing.

[0096] The server then analyzes the converted text data using a natural language processing (NLP) algorithm, such as the BERT model or an emotion analysis API. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs context and sentiment analysis to determine whether the text is likely to be customer harassment.

[0097] If customer harassment is detected, the server immediately flags it and sends a notification message to the device. For example, a warning message such as "A customer is making inappropriate remarks. Please respond calmly" will be displayed on the device. In serious cases, the server will send an escalation notification to the administrator, urging them to take urgent action.

[0098] After the call ends, the server logs all conversations and detection results, generates a report, and stores it in a database that stores the time of the customer harassment incident, the offensive keywords used, the customer sentiment analysis results, and the response, which can be used for future analysis and training.

[0099] As a specific example, the user (operator) starts a call, the device starts recording, and sends the voice data to the server in real time. The server uses a speech recognition engine to convert customer comments such as "This product is completely unusable!" into text. The NLP engine then detects the offensive keyword "completely unusable" and determines that the customer is irritated. This is judged to be customer harassment, and the server sends a notification to the device saying, "A customer is making inappropriate comments. Please respond calmly." The server also sends an escalation notification to an administrator, who then intervenes. After the call ends, the server records all content and detection results and generates a report.

[0100] Prompting a generative AI model for input can be done as follows:

[0101] "Please explain in detail the specific processing steps of a system for detecting customer harassment in a call center in real time."

[0102] As described above, this invention utilizes voice recognition and natural language processing technology to detect customer harassment in real time and provide appropriate notifications and escalations, thereby reducing the mental burden on operators and improving the quality of customer service.

[0103] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0104] Step 1:

[0105] When a user starts a call at a call center, the device records the conversation. The device then sends the recorded voice data to the server in real time. At this time, the device uses a microphone to capture the voice data and transfers it to the server in streaming format. The input is the user's voice, and the output is the transmission of voice data to the server. Specifically, when the user presses the "Start Call" button, the device starts recording and continuously sends voice data to the server.

[0106] Step 2:

[0107] The server inputs the received voice data into a voice recognition engine. The voice recognition engine converts the voice data into text data. In this case, Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services are often used. The input is voice data, and the output is the converted text data. Specifically, the server starts the voice recognition engine and inputs the received voice data into the engine. The engine analyzes the voice data and generates the corresponding text data.

[0108] Step 3:

[0109] The server analyzes the converted text data using a natural language processing (NLP) algorithm. The NLP engine detects specific keywords and phrases and also analyzes context and sentiment. Technologies used include the BERT model and emotion analysis API. The input is text data, and the output is the analysis results, including possible customer harassment. Specifically, the text data is input into the NLP engine, which returns the analysis results for keywords and sentiment.

[0110] Step 4:

[0111] If customer harassment is detected, the server sets a flag for response. The server then sends a notification to the device and, if necessary, an escalation notification to the administrator. The input is the NLP analysis result, and the output is the sending of a notification. Specifically, the server sends a message to the device saying, "A customer is making inappropriate remarks. Please respond calmly," and if the situation is serious, it sends a notification to the administrator saying, "Urgent action is required."

[0112] Step 5:

[0113] When the call ends, the server logs all conversation content and detection results and generates a report. The input is the text data of the call content and the detection results of customer harassment, and the output is the generation of a report. Specifically, the server receives a call end trigger, aggregates all data, creates a log file, and launches the report generation module.

[0114] Step 6:

[0115] The server stores the generated reports in a database for later analysis and feedback. The input is the generated report, and the output is saving it to the database. Specifically, the server writes the reports to the database and periodically scans the database as needed to analyze newly accumulated data.

[0116] (Application example 1)

[0117] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0118] When employees deal with customers in physical stores, it is difficult for them to respond appropriately to customer harassment. If appropriate responses to customer harassment are delayed, the mental burden on employees increases, which can result in a decline in the quality of customer service. Furthermore, if customer harassment occurs frequently, a means of quickly communicating its impact to managers is also required. To solve these issues, a system is needed that can detect customer harassment in real time and notify employees appropriately.

[0119] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0120] In this invention, the server includes means for acquiring voice data, means for converting the voice data into text data, means equipped with a natural language processing algorithm for detecting customer harassment based on the converted text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the conversation, means for saving the report in a database, and means for displaying a notification of customer harassment on the smart glasses. This enables employees at physical stores to detect customer harassment in conversations with customers in real time and respond promptly and appropriately, thereby reducing the mental burden on employees and improving the quality of customer service.

[0121] The "means for acquiring voice data" refers to a means having the function of collecting the contents of conversations between employees and customers in a physical store in voice format using a device such as a microphone.

[0122] The "means for converting voice data into text data" refers to a means having a function for converting acquired voice data into text format data using voice recognition technology.

[0123] A "means for detecting customer harassment" is a means that is equipped with a natural language processing algorithm and has the ability to analyze text data and identify customer harassment in real time.

[0124] "Means for notification" refers to means that have the function of immediately notifying employees and managers of warnings and information when customer harassment is detected.

[0125] The "means for generating a report of the detection results of customer harassment and the content of the conversation" is a means having a function for creating a report summarizing the analyzed content of the conversation and the detection results of customer harassment.

[0126] "Means for saving to a database" refers to means having the function of recording the generated report in a database for long-term storage and management.

[0127] "Means for displaying on smart glasses" refers to a means for displaying a notice regarding customer harassment on the display of smart glasses, thereby visually informing employees.

[0128] "Natural language processing algorithms" are a set of methods and technologies that analyze text data, identify emotions and keywords, and detect customer harassment.

[0129] This invention is a system that detects customer harassment (kasuhara) in real time while serving customers in a brick-and-mortar store, reduces the mental burden on employees, and improves the quality of customer service. This system allows employees wearing smart glasses to monitor conversations with customers, and is realized using the following hardware and software.

[0130] Hardware

[0131] 1. Smart glasses: A device worn by employees to capture audio and display notifications.

[0132] 2. Server: A central server for processing voice data and managing the database.

[0133] 3. Microphone: Built into the smart glasses, it records conversations with customers.

[0134] software

[0135] 1. Speech recognition engine: Software that converts voice data into text data (e.g., Google Speech Recognition).

[0136] 2. Natural language processing algorithms: Software that analyzes text data and detects customer harassment (e.g., sentiment analysis models for Hugging Face transformers).

[0137] 3. Database Management System: Software that stores and manages logs and reports.

[0138] Program processing flow (natural language explanation)

[0139] 1. Voice data capture: When a customer and an employee start a conversation, the built-in microphone in the smart glasses captures the voice data, which is then sent to the server in real time.

[0140] 2. Speech recognition: The server inputs the received voice data into a voice recognition engine and converts the voice data into text data. For example, a customer saying, "This product is completely unusable!" is converted into text.

[0141] 3. Natural Language Processing: The server inputs the converted text data into a natural language processing algorithm to detect customer harassment using specific keywords and sentiment analysis, such as "idiot," "fired," and "useless."

[0142] 4. Displaying a notification: If customer harassment is detected, the server will display a warning message on the smart glasses saying, "This may be customer harassment. Please respond calmly." In serious cases, the server will also escalate the situation to an administrator.

[0143] 5. Generate and save report: After the conversation is over, the server generates a report of all the conversation contents and the results of customer harassment detection and saves it in the database. This report includes the time of the customer harassment, the offensive keywords used, the customer sentiment analysis results, and the response.

[0144] This allows employees to respond to customer harassment promptly and appropriately, reducing their mental stress and improving the quality of customer service. The generated reports can also be used for training and support at a later date.

[0145] Adding specific examples

[0146] For example, suppose an employee wearing smart glasses is serving a customer and the customer says, "This product is completely unusable!" This comment is picked up by the smart glasses' microphone and sent to the server as voice data. The voice recognition engine then converts the voice into text data, and the natural language processing algorithm detects the offensive keyword "completely unusable" and negative emotion. This causes the smart glasses to display a message saying, "This may be customer harassment. Please respond calmly."

[0147] Example prompts to input to the generative AI model

[0148] For example, by inputting a prompt such as, "Please analyze what emotion is contained in a customer's statement, 'I can't use this product at all!' and whether it constitutes customer harassment," into the generative AI model, it is possible to perform appropriate emotion analysis and detect customer harassment.

[0149] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0150] Step 1: Acquire audio data

[0151] The user (employee) puts on the smart glasses and starts talking with the customer. At this stage, the microphone built into the smart glasses picks up the voice data. The picked up voice data is sent to the server in real time. The input is the voice of the conversation between the customer and the employee, and the output is the voice data.

[0152] Step 2: Voice Recognition

[0153] The voice data received by the server is input into a voice recognition engine, which converts the voice data into text data. For example, a voice saying, "This product is completely unusable!" is converted into text. The input is voice data, and the output is text data. Voice recognition software such as Google Speech Recognition is used for this process.

[0154] Step 3: Natural Language Processing and Harassment Detection

[0155] The server inputs the converted text data into a natural language processing algorithm to perform specific keyword and sentiment analysis. The natural language processing algorithm (e.g., the sentiment analysis model of the Hugging Face transformers) analyzes the text and determines whether it is likely to be customer harassment. The input is the text data, and the output is the sentiment analysis results and the detection results of customer harassment.

[0156] Step 4: View notifications

[0157] If customer harassment is detected, the server displays a warning message on the smart glasses. For example, a message such as "This may be customer harassment. Please respond calmly" may be displayed on the smart glasses' display. In serious cases, an escalation notification may be sent to an administrator. The input is the customer harassment detection result, and the output is a warning message and an escalation notification.

[0158] Step 5: Generate and save the report

[0159] Once the conversation is over, the server generates a report containing all the conversation content and the results of customer harassment detection. This report includes the time of the harassment, the offensive keywords used, the customer sentiment analysis results, and the response. The generated report is saved in a database. The input is the conversation content and the results of customer harassment detection, and the output is the report data.

[0160] Step 6: Analyze the report and provide feedback

[0161] The server periodically analyzes the report data stored in the database and uses it as material for training programs to improve the quality of operator responses. This analysis aims to improve the mental health of employees and the quality of customer service. The input is the report data from the database, and the output is the analysis results and material for training programs.

[0162] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0163] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. In particular, this system incorporates an emotion engine that recognizes the user's emotions, making it possible to more accurately detect stress caused by customer harassment and take appropriate action at the right time.

[0164] System configuration

[0165] 1. Acquiring audio data

[0166] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0167] 2. Voice recognition (ASR)

[0168] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from "voice to text." For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[0169] 3. Natural Language Processing (NLP)

[0170] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0171] 4. Emotion Recognition by Emotion Engine

[0172] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. For example, emotions such as "anger" or "irritation" can be recognized in real time from the customer's tone of voice and the content of their comments.

[0173] 5. Real-time detection and notification

[0174] If customer harassment or negative emotions are detected, the server immediately flags the situation and sends a notification to the device, warning the operator and providing instructions on how to respond. For example, a message such as "A customer is making inappropriate comments. Please respond calmly."

[0175] 6. Escalation and Administrator Notification

[0176] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[0177] 7. Report generation and data accumulation

[0178] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[0179] 8. Data analysis and feedback

[0180] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[0181] Specific operation example

[0182] Specifically, when the user (operator) starts a call, the device records the call and sends it to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable" and determines that the customer is irritated through context and sentiment analysis. The server then uses an emotion engine to recognize the customer's emotional state in detail and confirms that the emotion is "anger." If this is determined to be customer harassment, the server sends a notification to the device, warning, "The customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing them to intervene.

[0183] After the call ends, the server logs the conversation, the customer harassment detection results, and the emotion analysis results, then generates a report and stores it in a database. The data is then periodically analyzed to help improve the mental health of operators and the quality of customer service. In this way, the system of the present invention can detect customer harassment in real time and respond appropriately, thereby reducing the mental burden on operators and improving the quality of customer service.

[0184] The processing flow will be explained below.

[0185] Step 1:

[0186] The user (operator) starts a call. The device detects this action and automatically starts recording the conversation between the customer and the operator.

[0187] Step 2:

[0188] The device transmits the recorded audio data to the server in real time, and the transmitted audio data includes the entire content of the conversation.

[0189] Step 3:

[0190] The server inputs the received voice data into a voice recognition engine. This engine analyzes the voice data and converts it into text data. For example, a voice saying "This product is completely unusable!" is converted into text data saying "This product is completely unusable!"

[0191] Step 4:

[0192] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0193] Step 5:

[0194] The server uses the emotion engine based on the analysis results of the NLP engine to perform a detailed analysis of the user's emotional state. For example, it can recognize the customer's emotions, such as "anger" or "irritation," in real time, taking into account the tone of voice and the content of what is being said.

[0195] Step 6:

[0196] The server integrates the results from the emotion engine and the NLP engine to evaluate the possibility of customer harassment. If it is determined to be customer harassment, it sets a corresponding flag. This flag will guide the next action.

[0197] Step 7:

[0198] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[0199] Step 8:

[0200] If the customer harassment is serious, the server will send an escalation notification to an administrator, who can monitor the call content in real time and directly intervene if necessary.

[0201] Step 9:

[0202] The user (operator) takes the specified action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[0203] Step 10:

[0204] The server generates a report containing the time of occurrence of the customer harassment, the offensive keywords used, the results of the sentiment analysis, and the details of the response, and stores this in a database.

[0205] Step 11:

[0206] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[0207] Example 2

[0208] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0209] Customer harassment in call centers increases the mental burden on operators and reduces the quality of customer service. Current systems have difficulty detecting customer harassment in real time and responding appropriately, and are insufficiently effective in reducing operator stress. For this reason, there is a need for a system that can accurately detect customer harassment and take prompt action.

[0210] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: a means for the user to acquire voice data; a means for the terminal to transmit the voice data to the server in real time; a means for the server to convert the voice data into text data; a means for the server to detect customer harassment based on the text data; a means for the server to notify the terminal when customer harassment is detected; a means for the server to generate a report on the detection results of the customer harassment and the contents of the call; a means for the server to store the report in a database; and a means for the server to analyze the information stored in the database and provide feedback to reduce the user's mental stress. This enables customer harassment to be detected in real time, immediately notified to an operator, and escalated to a manager. Furthermore, recording the detection results and the contents of the call in a report contributes to future countermeasures and improving the mental health of operators.

[0211] "Voice data" is digital data that records the content of a conversation between a user (operator) and a customer.

[0212] "Text data" is character string data converted from voice data by a voice recognition engine.

[0213] "Customer harassment" refers to inappropriate words or actions directed at operators by customers, which increase the mental stress on the operators.

[0214] A "server" is a central computer system that processes voice data and performs speech recognition, natural language processing, and sentiment analysis.

[0215] A "terminal" is a computer device used by a user (operator) and is a device that acquires voice data and transmits it to a server.

[0216] "Notification" means a warning or instruction message sent to an operator or administrator when customer harassment is detected.

[0217] A "report" is written data that records the content of calls and the results of customer harassment detection, and is analysis material that is stored in a database.

[0218] A "database" is a system for efficiently storing and managing large amounts of data, and is a device for accumulating reports and call records.

[0219] "User" refers to an operator who handles customer service at a call center, and is the entity that acquires voice data and receives notifications from the system.

[0220] "Emotion analysis" is the process of analyzing and recognizing a user's emotional state from text data and voice tone.

[0221] "Escalation" is the process of notifying an administrator and encouraging direct intervention when customer harassment is deemed severe.

[0222] "Feedback" is the process of analyzing information stored in a database and providing information that helps improve the quality of the user's response and reduce mental stress.

[0223] This invention is a system that detects customer harassment in a call center in real time and takes appropriate action to reduce the mental burden on operators and improve the quality of customer service. Specific embodiments are described below.

[0224] Hardware and software used

[0225] Hardware:

[0226] Terminal: A computer used by an operator. This terminal has the function of acquiring voice data and sending it to a server.

[0227] Server: A central processing unit for speech recognition, natural language processing, and sentiment analysis.

[0228] software:

[0229] Speech recognition engine (e.g. Google Speech-to-Text)

[0230] Natural Language Processing (NLP) engines (e.g., SpaCy, NLTK)

[0231] Emotion engine (e.g. IBM Watson® Tone Analyzer)

[0232] Data processing and calculation

[0233] 1. Acquiring audio data

[0234] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0235] 2. Voice Recognition

[0236] The server inputs the received voice data into a voice recognition engine, which converts the voice data into "voice to text."

[0237] 3. Natural Language Processing

[0238] The server passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and performs contextual and sentiment analysis.

[0239] 4. Emotion Recognition by Emotion Engine

[0240] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state.

[0241] 5. Real-time detection and notification

[0242] If customer harassment and negative emotions are detected, the server immediately raises a response flag and sends a notification to the device.

[0243] 6. Escalation and Administrator Notification

[0244] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who will monitor the call in real time and intervene directly if necessary.

[0245] 7. Report generation and data accumulation

[0246] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[0247] 8. Data analysis and feedback

[0248] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[0249] Specific examples

[0250] Acquiring audio data:

[0251] When the operator starts a call with a customer, the device starts recording using a VoIP application and transmits the data to the server in real time.

[0252] Voice Recognition:

[0253] The server uses the Google Speech-to-Text API to analyze the received voice data and generate text data such as, "This product is completely unusable!"

[0254] Natural Language Processing:

[0255] The server uses an NLP engine (e.g., SpaCy) to detect offensive keywords and phrases, such as "totally unusable," and analyzes customer frustration.

[0256] Emotion Recognition with Emotion Engine:

[0257] The server uses IBM Watson Tone Analyzer to recognize the emotion of "anger" in real time from the customer's voice tone and text.

[0258] Real-time detection and notification:

[0259] When the server recognizes customer harassment, it displays a message on the operator's terminal saying, "A customer is making inappropriate remarks. Please respond calmly."

[0260] An example of a prompt for a generative AI model is:

[0261] "Develop a system that records call content in the call center in real time and detects customer harassment. Specific steps include voice recognition, natural language processing, and sentiment analysis, and notifying operators or managers as necessary. After the call ends, a report should be generated and saved in a database."

[0262] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0263] Step 1: Acquire audio data

[0264] When a user (operator) starts a call, the device records the call. The recorded voice data is sent to the server in real time as input. This process captures the voice call data between the customer and the operator. Specifically, when an operator makes a call using a VoIP application, the software automatically records the call and sends it to the server as digital voice data.

[0265] Step 2: Voice Recognition

[0266] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. In this process, the voice data is converted into character string data. For example, the Google Speech-to-Text API is used to convert the voice "This product is completely unusable!" into text. Text data is generated as the output.

[0267] Step 3: Natural Language Processing

[0268] The server passes the converted text data to a natural language processing (NLP) engine. The NLP engine detects specific keywords and phrases and performs contextual and sentiment analysis. Based on the input text data, keyword extraction and context understanding are performed. For example, an NLP engine (e.g., SpaCy) can be used to detect the offensive keyword "completely unusable" and infer customer frustration. The results of the contextual and sentiment analysis are generated as output.

[0269] Step 4: Emotion Recognition with the Emotion Engine

[0270] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. The text data and analysis results obtained in the previous step are used as input. For example, IBM Watson Tone Analyzer is used to recognize the emotion of "anger" in real time from text and voice tone. This process clarifies the customer's specific emotional state, and the emotion analysis results are obtained as output.

[0271] Step 5: Real-time detection and notification

[0272] If the server detects customer harassment or negative emotions, it immediately raises a flag and sends a notification to the terminal. The results of emotion analysis are used as input. Specifically, based on the detected emotional state (e.g., anger), a message such as "A customer is making inappropriate remarks. Please respond calmly" is displayed on the operator's terminal. A notification message is sent as output.

[0273] Step 6: Escalation and Administrator Notification

[0274] If the customer harassment is deemed to be serious, the server sends an escalation notification to the administrator. The further analyzed sentiment analysis results are used as input. Specifically, an alert is sent to the administrator stating, "Customer harassment has been detected in the current call. Please monitor the call and intervene if necessary." The escalation notification is sent as output.

[0275] Step 7: Report generation and data accumulation

[0276] When the call ends, the server records the content of the conversation, the results of customer harassment detection, and the results of sentiment analysis in a log and generates a report. The call data and analysis results are used as input. In this process, a detailed report is created and saved in the database. Specifically, a report stating "Customer harassment was detected in the call with the customer. Please refer to the log for details" is generated and saved in the database. Report data is generated as output.

[0277] Step 8: Data analysis and feedback

[0278] The server periodically analyzes the information stored in the database to help improve the quality of operator support and customer service. As input, multiple reports stored in the database are analyzed. This process uses the analysis results as material for training programs and provides feedback to operators. Specifically, monthly reports are generated, and the response status of each operator and successful examples of customer harassment responses are analyzed and used as material for the next training session. As output, an analysis report and improvement proposals are created.

[0279] (Application example 2)

[0280] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0281] Customer harassment in call centers increases the mental burden on operators and can be a factor in reducing the quality of customer service. In particular, in urgent tasks such as security services, customer stress levels tend to rise, potentially increasing the frequency of customer harassment. Effective countermeasures are needed, but current systems struggle to respond in real time, and there is no mechanism in place for operators or managers to intervene quickly. For this reason, there is a strong demand for the development of a system that can quickly and accurately detect customer harassment and respond in real time.

[0282] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0283] In this invention, the server includes means for acquiring voice data, means for converting voice data into text data, means for detecting customer harassment based on the text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the call, means for saving the report in a database, means for analyzing emotions in the content of the call, and means for issuing response instructions to an operator and a manager based on the content of the call. This makes it possible to detect customer harassment in real time and respond quickly.

[0284] "Means for acquiring voice data" refers to a device or software that records the conversation between the customer and the operator and transmits the voice data to a server in real time.

[0285] The "means for converting voice data into text data" refers to a device or software that analyzes acquired voice data and converts it into text data using voice recognition technology.

[0286] A "means for detecting customer harassment based on text data" is a device or software that uses natural language processing technology on text data to evaluate whether or not there is customer harassment through specific keyword and sentiment analysis.

[0287] "Means for notifying when customer harassment is detected" refers to a device or software that issues appropriate warnings and instructions to operators or administrators in real time when customer harassment is detected.

[0288] The "means for generating a report of the customer harassment detection results and the content of the call" is a device or software that formats the customer harassment detection results and the related content of the call and outputs them as a report that can be referenced at a later date.

[0289] "Means for storing reports in a database" refers to a device or software that stores generated reports in a database system for managing and storing them.

[0290] The "means for analyzing the emotion of the call content" is a device or software that analyzes data including voice-recognized text data and voice tone to evaluate the emotional state of the customer.

[0291] "Means for issuing response instructions to operators and administrators based on the content of calls" refers to devices or software that issue specific response instructions to operators and administrators based on analyzed customer emotional data and the results of customer harassment detection.

[0292] The present invention provides a system that can detect customer harassment in a call center in real time and take appropriate action. Specific embodiments for carrying out the present invention will be described in detail below.

[0293] System configuration

[0294] The system of the present invention mainly acquires and analyzes voice data and implements a series of processes to detect customer harassment, which includes the following components:

[0295] 1. How to obtain audio data

[0296] It uses a microphone (built-in or external) installed on the device, such as a smartphone or head-mounted display (HMD).

[0297] When the call starts, the voice data is acquired and sent to the server in real time.

[0298] 2. Means of converting audio data into text data

[0299] On the server side, voice data is converted into text data using speech recognition technology, primarily using speech recognition engines such as Google Speech Recognition API.

[0300] 3. How to detect customer harassment based on text data

[0301] On the server side, the text data is analyzed using an NLP (Natural Language Processing) engine, using libraries such as TextBlob and NLTK.

[0302] 4. Means of notification if customer harassment is detected

[0303] Based on the detection results, a real-time warning is sent to the operator, with the message "A customer is making inappropriate remarks. Please respond calmly." appearing on the terminal display.

[0304] Administrators are notified of the escalation and are encouraged to prepare for intervention.

[0305] 5. A method for generating reports on customer harassment detection results and call content

[0306] After the call is completed, the server generates a report summarizing the detection results and the contents of the call, which is then stored in a database.

[0307] 6. How to save the report to a database

[0308] Reports are stored using a database system such as MySQL (registered trademark) or PostgreSQL.

[0309] 7. A way to analyze the sentiment of calls

[0310] Based on the content and tone of the customer's speech, sentiment is analyzed in detail using sentiment analysis engines such as TextBlob and VADER.

[0311] 8. Means of issuing instructions to operators and administrators based on the content of calls

[0312] Based on the results of sentiment analysis and customer harassment detection, the system sends specific instructions to operators and administrators, encouraging operators to remain calm and informing administrators of the need for intervention.

[0313] Specific examples

[0314] The following scenarios can be considered as examples of actual operation.

[0315] 1. An operator working at a security service anomaly reporting call center receives a call from a customer using a smartphone.

[0316] 2. A customer says something like, "This security system isn't working at all! What can you do about it?"

[0317] 3. The system converts the speech into text in real time and detects the emotion of "anger."

[0318] 4. A notification will appear on the operator's smartphone saying, "A customer is making inappropriate remarks. Please respond calmly."

[0319] 5. An escalation notification is sent to the administrator, who prepares to intervene.

[0320] Prompt Sentence Examples

[0321] An example of a prompt sentence to input to the generative AI model is as follows:

[0322] Develop an application that detects specific phrases and emotions in real time during calls with customers and notifies them if customer harassment occurs. This application uses Python for speech recognition and natural language processing. Specifically, it detects phrases containing the emotion "anger" from the converted speech and notifies the operator and administrator. Examples of customer harassment include "It's completely unusable" and "What are you going to do about it?"

[0323] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0324] Step 1:

[0325] The terminal captures the voice data of the customer and the operator as soon as the call starts, and the captured voice data is sent to the server in real time.

[0326] Input: Voice of the customer and operator

[0327] Output: Audio data

[0328] Specific operation: The device's microphone captures audio and transmits the audio data to the server in real time.

[0329] Step 2:

[0330] The server inputs the received voice data into a voice recognition engine (for example, Google Speech Recognition API) and converts the voice data into text data.

[0331] Input: Audio data

[0332] Output: Text data

[0333] Specific operation: The speech recognition engine analyzes the voice data and converts it into corresponding text data.

[0334] Step 3:

[0335] The server inputs the text data into a natural language processing (NLP) engine (e.g., TextBlob or NLTK) to analyze the text data, performing specific keyword detection and sentiment analysis.

[0336] Input: Text data

[0337] Output: Analysis results (keywords and emotional information)

[0338] How it works: The NLP engine analyzes text data and extracts information through keyword extraction and sentiment analysis.

[0339] Step 4:

[0340] The server detects customer harassment based on the analysis results. If negative emotions such as irritation or anger are detected, it determines that there is a high possibility of customer harassment.

[0341] Input: Analysis results

[0342] Output: Customer harassment judgment

[0343] Specific operation: The emotion engine evaluates the analysis results and determines whether or not there are any negative emotions.

[0344] Step 5:

[0345] If customer harassment is detected, the server sends a notification to the terminal and instructs the operator to respond. It also sends an escalation notification to the administrator.

[0346] Input: Customer Harassment Judgment

[0347] Output: Notification to operators and administrators

[0348] What happens: The server generates a notification message, displays a warning on the operator's screen, and sends an escalation notification to the administrator.

[0349] Step 6:

[0350] Once the call ends, the server generates a report based on the content of the call, the results of customer harassment detection, and the results of sentiment analysis.

[0351] Input: Call content, customer harassment detection results, emotion analysis results

[0352] Output: Call Report

[0353] Specific operation: The server compiles the call content and analysis results and creates a formatted report.

[0354] Step 7:

[0355] The server generates reports that are saved in a database and can be used later to improve the mental health of operators and the quality of customer service.

[0356] Input: Call Report

[0357] Output: Save to database

[0358] Specific operation: The server stores the report in a database system (e.g., MySQL or PostgreSQL).

[0359] Step 8:

[0360] The server periodically analyzes the information stored in the database and uses it to improve the quality of operator support and customer service. This information is also used as material for training programs.

[0361] Input: Reports in the database

[0362] Output: Analysis results, training materials

[0363] What it does: The server retrieves information from the database and extracts useful insights through data analysis.

[0364] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0365] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0366] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0367] [Second embodiment]

[0368] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0369] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0370] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0371] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0372] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0373] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0374] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0375] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0376] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0377] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0378] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0379] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0380] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. This system acquires voice data, converts it into text data using speech recognition technology, detects customer harassment using natural language processing (NLP), and has the ability to notify or escalate the situation in real time as necessary.

[0381] System configuration

[0382] 1. Acquiring audio data

[0383] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0384] 2. Voice recognition (ASR)

[0385] The server inputs the received voice data into a voice recognition engine, which converts the voice data into text data. For example, a voice saying "Hello, this is customer support" is converted into text "Hello, this is customer support."

[0386] 3. Natural Language Processing (NLP)

[0387] The server then analyzes the converted text data using advanced natural language processing (NLP) algorithms. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs further context and sentiment analysis to determine the likelihood of customer harassment.

[0388] 4. Real-time detection and notification

[0389] If customer harassment is detected, the server immediately raises a flag for response. The server then sends a notification to the device. For example, the device may display a message saying, "A customer is making inappropriate remarks. Please respond calmly." If serious customer harassment occurs, the server sends an escalation notification to an administrator so that the administrator can intervene.

[0390] 5. Report generation and data accumulation

[0391] After the call ends, the server logs all conversations and detection results, and compiles a report containing the time of the incident, the offensive keywords used, the customer sentiment analysis results, and the response. The server stores this report in a database.

[0392] 6. Data analysis and feedback

[0393] The server periodically analyzes the information stored in the database and uses it as material for training programs aimed at improving the mental health and quality of customer service, thereby reducing the mental burden on operators and enabling the continuous improvement of customer service quality.

[0394] Specific operation example

[0395] Assume that a user (operator) starts a call. The device records the contents of the call and sends them to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable," and determines that the customer is irritated through context and sentiment analysis. As this is determined to be customer harassment, the server sends a notification to the device, displaying a warning message saying, "A customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing the administrator to intervene.

[0396] After the call ends, the server logs all call content and any harassment findings, generates a report, and stores it in a database. This information is later used for operator training and support, and to improve the quality of customer service.

[0397] As described above, the present invention can detect customer harassment in real time and take appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service.

[0398] The processing flow will be explained below.

[0399] Step 1:

[0400] The user (operator) starts a call. The device detects this action and automatically starts recording the call.

[0401] Step 2:

[0402] The device transmits real-time recorded voice data to the edge server, which includes all conversations between the customer and the operator.

[0403] Step 3:

[0404] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[0405] Step 4:

[0406] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0407] Step 5:

[0408] The server evaluates the possibility of customer harassment based on the analysis results from the NLP engine. If it is determined to be customer harassment, it sets a response flag. This response flag will serve as a guide for the next action.

[0409] Step 6:

[0410] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[0411] Step 7:

[0412] If the harassment becomes serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[0413] Step 8:

[0414] The user (operator) takes appropriate action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[0415] Step 9:

[0416] The server generates a report containing the time when the customer harassment occurred, the offensive keywords used, the results of the customer sentiment analysis, and the details of the response, and stores the report in a database.

[0417] Step 10:

[0418] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[0419] Example 1

[0420] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0421] In conventional customer support systems, operators must directly respond to customer harassment, placing a heavy mental burden on the operators. Furthermore, even when customer harassment occurs, it is difficult to detect and respond in real time, which can lead to delays in appropriate escalation and reporting. This leads to a decline in the quality of customer support and a deterioration in overall service. This invention aims to reduce the mental burden on operators and improve the quality of customer support by detecting customer harassment in real time and providing appropriate notifications and escalations.

[0422] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0423] In this invention, the server includes: a means for the terminal to acquire voice data when a user initiates a call and transmit it to the server; a means for the server to input the voice data into a voice recognition engine and convert it into text data; a means for the server to use a natural language processing algorithm to detect customer harassment in real time based on the converted text data; a means for the server to notify the terminal when customer harassment is detected and, if necessary, send an escalation notice to an administrator; a means for recording the customer harassment detection results and the content of the call in a log file and generating a report; and a means for storing the generated report in a database. This enables customer harassment to be detected in real time and responded to promptly. Furthermore, by recording the content of the call and the detection results in detail and utilizing them for future analysis and feedback, the quality of customer service can be continuously improved.

[0424] A "user" is a person who acts as an operator at a call center and handles calls with customers.

[0425] A "terminal" refers to a device used by a user, including a calling device and a recording device, that acquires and transmits voice data.

[0426] The "server" refers to a central processing unit that receives voice data sent from a terminal and performs voice recognition and natural language processing.

[0427] "Voice data" refers to digital data that is a recording of the contents of a conversation between a user and a customer.

[0428] A "voice recognition engine" is software or hardware that converts voice data into text data.

[0429] "Text data" refers to character information converted from voice data by a voice recognition engine.

[0430] A "natural language processing algorithm" is a programming technology that analyzes text data and detects customer harassment based on specific keywords and context.

[0431] "Customer harassment" refers to offensive, insulting or inappropriate comments or behavior from a customer toward an operator.

[0432] "Notification" refers to a warning message or information transmission from the server to a terminal or administrator when customer harassment is detected.

[0433] An "escalation notification" is a message sent from the server to an administrator urging emergency action when customer harassment is serious.

[0434] "Log file" refers to a digital file used to record the content of calls and the results of customer harassment detection.

[0435] A "report" is a document that compiles detailed records of calls and customer harassment, and is used for future analysis and feedback.

[0436] "Database" means a digital information storage system for storing generated reports and log files for retrieval and analysis as needed.

[0437] The present invention is implemented as follows to detect customer harassment in real time in a call center, reduce the mental burden on operators, and improve the quality of customer service.

[0438] First, when a user starts a call at a call center, the terminal records the conversation and transmits the resulting voice data to the server in real time. The terminal is usually composed of a call device and a recording device provided as part of the operator tools.

[0439] The server then inputs the received voice data into a voice recognition engine, which can be Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services. The voice recognition engine converts the voice data into text data, which the server uses for further processing.

[0440] The server then analyzes the converted text data using a natural language processing (NLP) algorithm, such as the BERT model or an emotion analysis API. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs context and sentiment analysis to determine whether the text is likely to be customer harassment.

[0441] If customer harassment is detected, the server immediately flags it and sends a notification message to the device. For example, a warning message such as "A customer is making inappropriate remarks. Please respond calmly" will be displayed on the device. In serious cases, the server will send an escalation notification to the administrator, urging them to take urgent action.

[0442] After the call ends, the server logs all conversations and detection results, generates a report, and stores it in a database that stores the time of the customer harassment incident, the offensive keywords used, the customer sentiment analysis results, and the response, which can be used for future analysis and training.

[0443] As a specific example, the user (operator) starts a call, the device starts recording, and sends the voice data to the server in real time. The server uses a speech recognition engine to convert customer comments such as "This product is completely unusable!" into text. The NLP engine then detects the offensive keyword "completely unusable" and determines that the customer is irritated. This is judged to be customer harassment, and the server sends a notification to the device saying, "A customer is making inappropriate comments. Please respond calmly." The server also sends an escalation notification to an administrator, who then intervenes. After the call ends, the server records all content and detection results and generates a report.

[0444] Prompting a generative AI model for input can be done as follows:

[0445] "Please explain in detail the specific processing steps of a system for detecting customer harassment in a call center in real time."

[0446] As described above, this invention utilizes voice recognition and natural language processing technology to detect customer harassment in real time and provide appropriate notifications and escalations, thereby reducing the mental burden on operators and improving the quality of customer service.

[0447] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0448] Step 1:

[0449] When a user starts a call at a call center, the device records the conversation. The device then sends the recorded voice data to the server in real time. At this time, the device uses a microphone to capture the voice data and transfers it to the server in streaming format. The input is the user's voice, and the output is the transmission of voice data to the server. Specifically, when the user presses the "Start Call" button, the device starts recording and continuously sends voice data to the server.

[0450] Step 2:

[0451] The server inputs the received voice data into a voice recognition engine. The voice recognition engine converts the voice data into text data. In this case, Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services are often used. The input is voice data, and the output is the converted text data. Specifically, the server starts the voice recognition engine and inputs the received voice data into the engine. The engine analyzes the voice data and generates the corresponding text data.

[0452] Step 3:

[0453] The server analyzes the converted text data using a natural language processing (NLP) algorithm. The NLP engine detects specific keywords and phrases and also analyzes context and sentiment. Technologies used include the BERT model and emotion analysis API. The input is text data, and the output is the analysis results, including possible customer harassment. Specifically, the text data is input into the NLP engine, which returns the analysis results for keywords and sentiment.

[0454] Step 4:

[0455] If customer harassment is detected, the server sets a flag for response. The server then sends a notification to the device and, if necessary, an escalation notification to the administrator. The input is the NLP analysis result, and the output is the sending of a notification. Specifically, the server sends a message to the device saying, "A customer is making inappropriate remarks. Please respond calmly," and if the situation is serious, it sends a notification to the administrator saying, "Urgent action is required."

[0456] Step 5:

[0457] When the call ends, the server logs all conversation content and detection results and generates a report. The input is the text data of the call content and the detection results of customer harassment, and the output is the generation of a report. Specifically, the server receives a call end trigger, aggregates all data, creates a log file, and launches the report generation module.

[0458] Step 6:

[0459] The server stores the generated reports in a database for later analysis and feedback. The input is the generated report, and the output is saving it to the database. Specifically, the server writes the reports to the database and periodically scans the database as needed to analyze newly accumulated data.

[0460] (Application example 1)

[0461] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0462] When employees deal with customers in physical stores, it is difficult for them to respond appropriately to customer harassment. If appropriate responses to customer harassment are delayed, the mental burden on employees increases, which can result in a decline in the quality of customer service. Furthermore, if customer harassment occurs frequently, a means of quickly communicating its impact to managers is also required. To solve these issues, a system is needed that can detect customer harassment in real time and notify employees appropriately.

[0463] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0464] In this invention, the server includes means for acquiring voice data, means for converting the voice data into text data, means equipped with a natural language processing algorithm for detecting customer harassment based on the converted text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the conversation, means for saving the report in a database, and means for displaying a notification of customer harassment on the smart glasses. This enables employees at physical stores to detect customer harassment in conversations with customers in real time and respond promptly and appropriately, thereby reducing the mental burden on employees and improving the quality of customer service.

[0465] The "means for acquiring voice data" refers to a means having the function of collecting the contents of conversations between employees and customers in a physical store in voice format using a device such as a microphone.

[0466] The "means for converting voice data into text data" refers to a means having a function for converting acquired voice data into text format data using voice recognition technology.

[0467] A "means for detecting customer harassment" is a means that is equipped with a natural language processing algorithm and has the ability to analyze text data and identify customer harassment in real time.

[0468] "Means for notification" refers to means that have the function of immediately notifying employees and managers of warnings and information when customer harassment is detected.

[0469] The "means for generating a report of the detection results of customer harassment and the content of the conversation" is a means having a function for creating a report summarizing the analyzed content of the conversation and the detection results of customer harassment.

[0470] "Means for saving to a database" refers to means having the function of recording the generated report in a database for long-term storage and management.

[0471] "Means for displaying on smart glasses" refers to a means for displaying a notice regarding customer harassment on the display of smart glasses, thereby visually informing employees.

[0472] "Natural language processing algorithms" are a set of methods and technologies that analyze text data, identify emotions and keywords, and detect customer harassment.

[0473] This invention is a system that detects customer harassment (kasuhara) in real time while serving customers in a brick-and-mortar store, reduces the mental burden on employees, and improves the quality of customer service. This system allows employees wearing smart glasses to monitor conversations with customers, and is realized using the following hardware and software.

[0474] Hardware

[0475] 1. Smart glasses: A device worn by employees to capture audio and display notifications.

[0476] 2. Server: A central server for processing voice data and managing the database.

[0477] 3. Microphone: Built into the smart glasses, it records conversations with customers.

[0478] software

[0479] 1. Speech recognition engine: Software that converts voice data into text data (e.g., Google Speech Recognition).

[0480] 2. Natural language processing algorithms: Software that analyzes text data and detects customer harassment (e.g., sentiment analysis models for Hugging Face transformers).

[0481] 3. Database Management System: Software that stores and manages logs and reports.

[0482] Program processing flow (natural language explanation)

[0483] 1. Voice data capture: When a customer and an employee start a conversation, the built-in microphone in the smart glasses captures the voice data, which is then sent to the server in real time.

[0484] 2. Speech recognition: The server inputs the received voice data into a voice recognition engine and converts the voice data into text data. For example, a customer saying, "This product is completely unusable!" is converted into text.

[0485] 3. Natural Language Processing: The server inputs the converted text data into a natural language processing algorithm to detect customer harassment using specific keywords and sentiment analysis, such as "idiot," "fired," and "useless."

[0486] 4. Displaying a notification: If customer harassment is detected, the server will display a warning message on the smart glasses saying, "This may be customer harassment. Please respond calmly." In serious cases, the server will also escalate the situation to an administrator.

[0487] 5. Generate and save report: After the conversation is over, the server generates a report of all the conversation contents and the results of customer harassment detection and saves it in the database. This report includes the time of the customer harassment, the offensive keywords used, the customer sentiment analysis results, and the response.

[0488] This allows employees to respond to customer harassment promptly and appropriately, reducing their mental stress and improving the quality of customer service. The generated reports can also be used for training and support at a later date.

[0489] Adding specific examples

[0490] For example, suppose an employee wearing smart glasses is serving a customer and the customer says, "This product is completely unusable!" This comment is picked up by the smart glasses' microphone and sent to the server as voice data. The voice recognition engine then converts the voice into text data, and the natural language processing algorithm detects the offensive keyword "completely unusable" and negative emotion. This causes the smart glasses to display a message saying, "This may be customer harassment. Please respond calmly."

[0491] Example prompts to input to the generative AI model

[0492] For example, by inputting a prompt such as, "Please analyze what emotion is contained in a customer's statement, 'I can't use this product at all!' and whether it constitutes customer harassment," into the generative AI model, it is possible to perform appropriate emotion analysis and detect customer harassment.

[0493] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0494] Step 1: Acquire audio data

[0495] The user (employee) puts on the smart glasses and starts talking with the customer. At this stage, the microphone built into the smart glasses picks up the voice data. The picked up voice data is sent to the server in real time. The input is the voice of the conversation between the customer and the employee, and the output is the voice data.

[0496] Step 2: Voice Recognition

[0497] The voice data received by the server is input into a voice recognition engine, which converts the voice data into text data. For example, a voice saying, "This product is completely unusable!" is converted into text. The input is voice data, and the output is text data. Voice recognition software such as Google Speech Recognition is used for this process.

[0498] Step 3: Natural Language Processing and Harassment Detection

[0499] The server inputs the converted text data into a natural language processing algorithm to perform specific keyword and sentiment analysis. The natural language processing algorithm (e.g., the sentiment analysis model of the Hugging Face transformers) analyzes the text and determines whether it is likely to be customer harassment. The input is the text data, and the output is the sentiment analysis results and the detection results of customer harassment.

[0500] Step 4: View notifications

[0501] If customer harassment is detected, the server displays a warning message on the smart glasses. For example, a message such as "This may be customer harassment. Please respond calmly" may be displayed on the smart glasses' display. In serious cases, an escalation notification may be sent to an administrator. The input is the customer harassment detection result, and the output is a warning message and an escalation notification.

[0502] Step 5: Generate and save the report

[0503] Once the conversation is over, the server generates a report containing all the conversation content and the results of customer harassment detection. This report includes the time of the harassment, the offensive keywords used, the customer sentiment analysis results, and the response. The generated report is saved in a database. The input is the conversation content and the results of customer harassment detection, and the output is the report data.

[0504] Step 6: Analyze the report and provide feedback

[0505] The server periodically analyzes the report data stored in the database and uses it as material for training programs to improve the quality of operator responses. This analysis aims to improve the mental health of employees and the quality of customer service. The input is the report data from the database, and the output is the analysis results and material for training programs.

[0506] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0507] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. In particular, this system incorporates an emotion engine that recognizes the user's emotions, making it possible to more accurately detect stress caused by customer harassment and take appropriate action at the right time.

[0508] System configuration

[0509] 1. Acquiring audio data

[0510] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0511] 2. Voice recognition (ASR)

[0512] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from "voice to text." For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[0513] 3. Natural Language Processing (NLP)

[0514] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0515] 4. Emotion Recognition by Emotion Engine

[0516] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. For example, emotions such as "anger" or "irritation" can be recognized in real time from the customer's tone of voice and the content of their comments.

[0517] 5. Real-time detection and notification

[0518] If customer harassment or negative emotions are detected, the server immediately flags the situation and sends a notification to the device, warning the operator and providing instructions on how to respond. For example, a message such as "A customer is making inappropriate comments. Please respond calmly."

[0519] 6. Escalation and Administrator Notification

[0520] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[0521] 7. Report generation and data accumulation

[0522] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[0523] 8. Data analysis and feedback

[0524] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[0525] Specific operation example

[0526] Specifically, when the user (operator) starts a call, the device records the call and sends it to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable" and determines that the customer is irritated through context and sentiment analysis. The server then uses an emotion engine to recognize the customer's emotional state in detail and confirms that the emotion is "anger." If this is determined to be customer harassment, the server sends a notification to the device, warning, "The customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing them to intervene.

[0527] After the call ends, the server logs the conversation, the customer harassment detection results, and the emotion analysis results, then generates a report and stores it in a database. The data is then periodically analyzed to help improve the mental health of operators and the quality of customer service. In this way, the system of the present invention can detect customer harassment in real time and respond appropriately, thereby reducing the mental burden on operators and improving the quality of customer service.

[0528] The processing flow will be explained below.

[0529] Step 1:

[0530] The user (operator) starts a call. The device detects this action and automatically starts recording the conversation between the customer and the operator.

[0531] Step 2:

[0532] The device transmits the recorded audio data to the server in real time, and the transmitted audio data includes the entire content of the conversation.

[0533] Step 3:

[0534] The server inputs the received voice data into a voice recognition engine. This engine analyzes the voice data and converts it into text data. For example, a voice saying "This product is completely unusable!" is converted into text data saying "This product is completely unusable!"

[0535] Step 4:

[0536] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0537] Step 5:

[0538] The server uses the emotion engine based on the analysis results of the NLP engine to perform a detailed analysis of the user's emotional state. For example, it can recognize the customer's emotions, such as "anger" or "irritation," in real time, taking into account the tone of voice and the content of what is being said.

[0539] Step 6:

[0540] The server integrates the results from the emotion engine and the NLP engine to evaluate the possibility of customer harassment. If it is determined to be customer harassment, it sets a corresponding flag. This flag will guide the next action.

[0541] Step 7:

[0542] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[0543] Step 8:

[0544] If the customer harassment is serious, the server will send an escalation notification to an administrator, who can monitor the call content in real time and directly intervene if necessary.

[0545] Step 9:

[0546] The user (operator) takes the specified action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[0547] Step 10:

[0548] The server generates a report containing the time of occurrence of the customer harassment, the offensive keywords used, the results of the sentiment analysis, and the details of the response, and stores this in a database.

[0549] Step 11:

[0550] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[0551] Example 2

[0552] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0553] Customer harassment in call centers increases the mental burden on operators and reduces the quality of customer service. Current systems have difficulty detecting customer harassment in real time and responding appropriately, and are insufficiently effective in reducing operator stress. For this reason, there is a need for a system that can accurately detect customer harassment and take prompt action.

[0554] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: a means for the user to acquire voice data; a means for the terminal to transmit the voice data to the server in real time; a means for the server to convert the voice data into text data; a means for the server to detect customer harassment based on the text data; a means for the server to notify the terminal when customer harassment is detected; a means for the server to generate a report on the detection results of the customer harassment and the contents of the call; a means for the server to store the report in a database; and a means for the server to analyze the information stored in the database and provide feedback to reduce the user's mental stress. This enables customer harassment to be detected in real time, immediately notified to an operator, and escalated to a manager. Furthermore, recording the detection results and the contents of the call in a report contributes to future countermeasures and improving the mental health of operators.

[0555] "Voice data" is digital data that records the content of a conversation between a user (operator) and a customer.

[0556] "Text data" is character string data converted from voice data by a voice recognition engine.

[0557] "Customer harassment" refers to inappropriate words or actions directed at operators by customers, which increase the mental stress on the operators.

[0558] A "server" is a central computer system that processes voice data and performs speech recognition, natural language processing, and sentiment analysis.

[0559] A "terminal" is a computer device used by a user (operator) and is a device that acquires voice data and transmits it to a server.

[0560] "Notification" means a warning or instruction message sent to an operator or administrator when customer harassment is detected.

[0561] A "report" is written data that records the content of calls and the results of customer harassment detection, and is analysis material that is stored in a database.

[0562] A "database" is a system for efficiently storing and managing large amounts of data, and is a device for accumulating reports and call records.

[0563] "User" refers to an operator who handles customer service at a call center, and is the entity that acquires voice data and receives notifications from the system.

[0564] "Emotion analysis" is the process of analyzing and recognizing a user's emotional state from text data and voice tone.

[0565] "Escalation" is the process of notifying an administrator and encouraging direct intervention when customer harassment is deemed severe.

[0566] "Feedback" is the process of analyzing information stored in a database and providing information that helps improve the quality of the user's response and reduce mental stress.

[0567] This invention is a system that detects customer harassment in a call center in real time and takes appropriate action to reduce the mental burden on operators and improve the quality of customer service. Specific embodiments are described below.

[0568] Hardware and software used

[0569] Hardware:

[0570] Terminal: A computer used by an operator. This terminal has the function of acquiring voice data and sending it to a server.

[0571] Server: A central processing unit for speech recognition, natural language processing, and sentiment analysis.

[0572] software:

[0573] Speech recognition engine (e.g. Google Speech-to-Text)

[0574] Natural Language Processing (NLP) engines (e.g., SpaCy, NLTK)

[0575] Emotion engine (e.g. IBM Watson Tone Analyzer)

[0576] Data processing and calculation

[0577] 1. Acquiring audio data

[0578] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0579] 2. Voice Recognition

[0580] The server inputs the received voice data into a voice recognition engine, which converts the voice data into "voice to text."

[0581] 3. Natural Language Processing

[0582] The server passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and performs contextual and sentiment analysis.

[0583] 4. Emotion Recognition by Emotion Engine

[0584] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state.

[0585] 5. Real-time detection and notification

[0586] If customer harassment and negative emotions are detected, the server immediately raises a response flag and sends a notification to the device.

[0587] 6. Escalation and Administrator Notification

[0588] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who will monitor the call in real time and intervene directly if necessary.

[0589] 7. Report generation and data accumulation

[0590] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[0591] 8. Data analysis and feedback

[0592] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[0593] Specific examples

[0594] Acquiring audio data:

[0595] When the operator starts a call with a customer, the device starts recording using a VoIP application and transmits the data to the server in real time.

[0596] Voice Recognition:

[0597] The server uses the Google Speech-to-Text API to analyze the received voice data and generate text data such as, "This product is completely unusable!"

[0598] Natural Language Processing:

[0599] The server uses an NLP engine (e.g., SpaCy) to detect offensive keywords and phrases, such as "totally unusable," and analyzes customer frustration.

[0600] Emotion Recognition with Emotion Engine:

[0601] The server uses IBM Watson Tone Analyzer to recognize the emotion of "anger" in real time from the customer's voice tone and text.

[0602] Real-time detection and notification:

[0603] When the server recognizes customer harassment, it displays a message on the operator's terminal saying, "A customer is making inappropriate remarks. Please respond calmly."

[0604] An example of a prompt for a generative AI model is:

[0605] "Develop a system that records call content in the call center in real time and detects customer harassment. Specific steps include voice recognition, natural language processing, and sentiment analysis, and notifying operators or managers as necessary. After the call ends, a report should be generated and saved in a database."

[0606] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0607] Step 1: Acquire audio data

[0608] When a user (operator) starts a call, the device records the call. The recorded voice data is sent to the server in real time as input. This process captures the voice call data between the customer and the operator. Specifically, when an operator makes a call using a VoIP application, the software automatically records the call and sends it to the server as digital voice data.

[0609] Step 2: Voice Recognition

[0610] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. In this process, the voice data is converted into character string data. For example, the Google Speech-to-Text API is used to convert the voice "This product is completely unusable!" into text. Text data is generated as the output.

[0611] Step 3: Natural Language Processing

[0612] The server passes the converted text data to a natural language processing (NLP) engine. The NLP engine detects specific keywords and phrases and performs contextual and sentiment analysis. Based on the input text data, keyword extraction and context understanding are performed. For example, an NLP engine (e.g., SpaCy) can be used to detect the offensive keyword "completely unusable" and infer customer frustration. The results of the contextual and sentiment analysis are generated as output.

[0613] Step 4: Emotion Recognition with the Emotion Engine

[0614] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. The text data and analysis results obtained in the previous step are used as input. For example, IBM Watson Tone Analyzer is used to recognize the emotion of "anger" in real time from text and voice tone. This process clarifies the customer's specific emotional state, and the emotion analysis results are obtained as output.

[0615] Step 5: Real-time detection and notification

[0616] If the server detects customer harassment or negative emotions, it immediately raises a flag and sends a notification to the terminal. The results of emotion analysis are used as input. Specifically, based on the detected emotional state (e.g., anger), a message such as "A customer is making inappropriate remarks. Please respond calmly" is displayed on the operator's terminal. A notification message is sent as output.

[0617] Step 6: Escalation and Administrator Notification

[0618] If the customer harassment is deemed to be serious, the server sends an escalation notification to the administrator. The further analyzed sentiment analysis results are used as input. Specifically, an alert is sent to the administrator stating, "Customer harassment has been detected in the current call. Please monitor the call and intervene if necessary." The escalation notification is sent as output.

[0619] Step 7: Report generation and data accumulation

[0620] When the call ends, the server records the content of the conversation, the results of customer harassment detection, and the results of sentiment analysis in a log and generates a report. The call data and analysis results are used as input. In this process, a detailed report is created and saved in the database. Specifically, a report stating "Customer harassment was detected in the call with the customer. Please refer to the log for details" is generated and saved in the database. Report data is generated as output.

[0621] Step 8: Data analysis and feedback

[0622] The server periodically analyzes the information stored in the database to help improve the quality of operator support and customer service. As input, multiple reports stored in the database are analyzed. This process uses the analysis results as material for training programs and provides feedback to operators. Specifically, monthly reports are generated, and the response status of each operator and successful examples of customer harassment responses are analyzed and used as material for the next training session. As output, an analysis report and improvement proposals are created.

[0623] (Application example 2)

[0624] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0625] Customer harassment in call centers increases the mental burden on operators and can be a factor in reducing the quality of customer service. In particular, in urgent tasks such as security services, customer stress levels tend to rise, potentially increasing the frequency of customer harassment. Effective countermeasures are needed, but current systems struggle to respond in real time, and there is no mechanism in place for operators or managers to intervene quickly. For this reason, there is a strong demand for the development of a system that can quickly and accurately detect customer harassment and respond in real time.

[0626] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0627] In this invention, the server includes means for acquiring voice data, means for converting voice data into text data, means for detecting customer harassment based on the text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the call, means for saving the report in a database, means for analyzing emotions in the content of the call, and means for issuing response instructions to an operator and a manager based on the content of the call. This makes it possible to detect customer harassment in real time and respond quickly.

[0628] "Means for acquiring voice data" refers to a device or software that records the conversation between the customer and the operator and transmits the voice data to a server in real time.

[0629] The "means for converting voice data into text data" refers to a device or software that analyzes acquired voice data and converts it into text data using voice recognition technology.

[0630] A "means for detecting customer harassment based on text data" is a device or software that uses natural language processing technology on text data to evaluate whether or not there is customer harassment through specific keyword and sentiment analysis.

[0631] "Means for notifying when customer harassment is detected" refers to a device or software that issues appropriate warnings and instructions to operators or administrators in real time when customer harassment is detected.

[0632] The "means for generating a report of the customer harassment detection results and the content of the call" is a device or software that formats the customer harassment detection results and the related content of the call and outputs them as a report that can be referenced at a later date.

[0633] "Means for storing reports in a database" refers to a device or software that stores generated reports in a database system for managing and storing them.

[0634] The "means for analyzing the emotion of the call content" is a device or software that analyzes data including voice-recognized text data and voice tone to evaluate the emotional state of the customer.

[0635] "Means for issuing response instructions to operators and administrators based on the content of calls" refers to devices or software that issue specific response instructions to operators and administrators based on analyzed customer emotional data and the results of customer harassment detection.

[0636] The present invention provides a system that can detect customer harassment in a call center in real time and take appropriate action. Specific embodiments for carrying out the present invention will be described in detail below.

[0637] System configuration

[0638] The system of the present invention mainly acquires and analyzes voice data and implements a series of processes to detect customer harassment, which includes the following components:

[0639] 1. How to obtain audio data

[0640] It uses a microphone (built-in or external) installed on the device, such as a smartphone or head-mounted display (HMD).

[0641] When the call starts, the voice data is acquired and sent to the server in real time.

[0642] 2. Means of converting audio data into text data

[0643] On the server side, voice data is converted into text data using speech recognition technology, primarily using speech recognition engines such as Google Speech Recognition API.

[0644] 3. How to detect customer harassment based on text data

[0645] On the server side, the text data is analyzed using an NLP (Natural Language Processing) engine, using libraries such as TextBlob and NLTK.

[0646] 4. Means of notification if customer harassment is detected

[0647] Based on the detection results, a real-time warning is sent to the operator, with the message "A customer is making inappropriate remarks. Please respond calmly." appearing on the terminal display.

[0648] Administrators are notified of the escalation and are encouraged to prepare for intervention.

[0649] 5. A method for generating reports on customer harassment detection results and call content

[0650] After the call is completed, the server generates a report summarizing the detection results and the contents of the call, which is then stored in a database.

[0651] 6. How to save the report to a database

[0652] Use a database system such as MySQL or PostgreSQL to store reports.

[0653] 7. A way to analyze the sentiment of calls

[0654] Based on the content and tone of the customer's speech, sentiment is analyzed in detail using sentiment analysis engines such as TextBlob and VADER.

[0655] 8. Means of issuing instructions to operators and administrators based on the content of calls

[0656] Based on the results of sentiment analysis and customer harassment detection, the system sends specific instructions to operators and administrators, encouraging operators to remain calm and informing administrators of the need for intervention.

[0657] Specific examples

[0658] The following scenarios can be considered as examples of actual operation.

[0659] 1. An operator working at a security service anomaly reporting call center receives a call from a customer using a smartphone.

[0660] 2. A customer says something like, "This security system isn't working at all! What can you do about it?"

[0661] 3. The system converts the speech into text in real time and detects the emotion of "anger."

[0662] 4. A notification will appear on the operator's smartphone saying, "A customer is making inappropriate remarks. Please respond calmly."

[0663] 5. An escalation notification is sent to the administrator, who prepares to intervene.

[0664] Prompt Sentence Examples

[0665] An example of a prompt sentence to input to the generative AI model is as follows:

[0666] Develop an application that detects specific phrases and emotions in real time during calls with customers and notifies them if customer harassment occurs. This application uses Python for speech recognition and natural language processing. Specifically, it detects phrases containing the emotion "anger" from the converted speech and notifies the operator and administrator. Examples of customer harassment include "It's completely unusable" and "What are you going to do about it?"

[0667] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0668] Step 1:

[0669] The terminal captures the voice data of the customer and the operator as soon as the call starts, and the captured voice data is sent to the server in real time.

[0670] Input: Voice of the customer and operator

[0671] Output: Audio data

[0672] Specific operation: The device's microphone captures audio and transmits the audio data to the server in real time.

[0673] Step 2:

[0674] The server inputs the received voice data into a voice recognition engine (for example, Google Speech Recognition API) and converts the voice data into text data.

[0675] Input: Audio data

[0676] Output: Text data

[0677] Specific operation: The speech recognition engine analyzes the voice data and converts it into corresponding text data.

[0678] Step 3:

[0679] The server inputs the text data into a natural language processing (NLP) engine (e.g., TextBlob or NLTK) to analyze the text data, performing specific keyword detection and sentiment analysis.

[0680] Input: Text data

[0681] Output: Analysis results (keywords and emotional information)

[0682] How it works: The NLP engine analyzes text data and extracts information through keyword extraction and sentiment analysis.

[0683] Step 4:

[0684] The server detects customer harassment based on the analysis results. If negative emotions such as irritation or anger are detected, it determines that there is a high possibility of customer harassment.

[0685] Input: Analysis results

[0686] Output: Customer harassment judgment

[0687] Specific operation: The emotion engine evaluates the analysis results and determines whether or not there are any negative emotions.

[0688] Step 5:

[0689] If customer harassment is detected, the server sends a notification to the terminal and instructs the operator to respond. It also sends an escalation notification to the administrator.

[0690] Input: Customer Harassment Judgment

[0691] Output: Notification to operators and administrators

[0692] What happens: The server generates a notification message, displays a warning on the operator's screen, and sends an escalation notification to the administrator.

[0693] Step 6:

[0694] Once the call ends, the server generates a report based on the content of the call, the results of customer harassment detection, and the results of sentiment analysis.

[0695] Input: Call content, customer harassment detection results, emotion analysis results

[0696] Output: Call Report

[0697] Specific operation: The server compiles the call content and analysis results and creates a formatted report.

[0698] Step 7:

[0699] The server generates reports that are saved in a database and can be used later to improve the mental health of operators and the quality of customer service.

[0700] Input: Call Report

[0701] Output: Save to database

[0702] Specific operation: The server stores the report in a database system (e.g., MySQL or PostgreSQL).

[0703] Step 8:

[0704] The server periodically analyzes the information stored in the database and uses it to improve the quality of operator support and customer service. This information is also used as material for training programs.

[0705] Input: Reports in the database

[0706] Output: Analysis results, training materials

[0707] What it does: The server retrieves information from the database and extracts useful insights through data analysis.

[0708] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0709] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0710] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0711] [Third embodiment]

[0712] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0713] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0714] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0715] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0716] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0717] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0718] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0719] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0720] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0721] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0722] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0723] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0724] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. This system acquires voice data, converts it into text data using speech recognition technology, detects customer harassment using natural language processing (NLP), and has the ability to notify or escalate the situation in real time as necessary.

[0725] System configuration

[0726] 1. Acquiring audio data

[0727] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0728] 2. Voice recognition (ASR)

[0729] The server inputs the received voice data into a voice recognition engine, which converts the voice data into text data. For example, a voice saying "Hello, this is customer support" is converted into text "Hello, this is customer support."

[0730] 3. Natural Language Processing (NLP)

[0731] The server then analyzes the converted text data using advanced natural language processing (NLP) algorithms. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs further context and sentiment analysis to determine the likelihood of customer harassment.

[0732] 4. Real-time detection and notification

[0733] If customer harassment is detected, the server immediately raises a flag for response. The server then sends a notification to the device. For example, the device may display a message saying, "A customer is making inappropriate remarks. Please respond calmly." If serious customer harassment occurs, the server sends an escalation notification to an administrator so that the administrator can intervene.

[0734] 5. Report generation and data accumulation

[0735] After the call ends, the server logs all conversations and detection results, and compiles a report containing the time of the incident, the offensive keywords used, the customer sentiment analysis results, and the response. The server stores this report in a database.

[0736] 6. Data analysis and feedback

[0737] The server periodically analyzes the information stored in the database and uses it as material for training programs aimed at improving the mental health and quality of customer service, thereby reducing the mental burden on operators and enabling the continuous improvement of customer service quality.

[0738] Specific operation example

[0739] Assume that a user (operator) starts a call. The device records the contents of the call and sends them to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable," and determines that the customer is irritated through context and sentiment analysis. As this is determined to be customer harassment, the server sends a notification to the device, displaying a warning message saying, "A customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing the administrator to intervene.

[0740] After the call ends, the server logs all call content and any harassment findings, generates a report, and stores it in a database. This information is later used for operator training and support, and to improve the quality of customer service.

[0741] As described above, the present invention can detect customer harassment in real time and take appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service.

[0742] The processing flow will be explained below.

[0743] Step 1:

[0744] The user (operator) starts a call. The device detects this action and automatically starts recording the call.

[0745] Step 2:

[0746] The device transmits real-time recorded voice data to the edge server, which includes all conversations between the customer and the operator.

[0747] Step 3:

[0748] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[0749] Step 4:

[0750] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0751] Step 5:

[0752] The server evaluates the possibility of customer harassment based on the analysis results from the NLP engine. If it is determined to be customer harassment, it sets a response flag. This response flag will serve as a guide for the next action.

[0753] Step 6:

[0754] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[0755] Step 7:

[0756] If the harassment becomes serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[0757] Step 8:

[0758] The user (operator) takes appropriate action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[0759] Step 9:

[0760] The server generates a report containing the time when the customer harassment occurred, the offensive keywords used, the results of the customer sentiment analysis, and the details of the response, and stores the report in a database.

[0761] Step 10:

[0762] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[0763] Example 1

[0764] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0765] In conventional customer support systems, operators must directly respond to customer harassment, placing a heavy mental burden on the operators. Furthermore, even when customer harassment occurs, it is difficult to detect and respond in real time, which can lead to delays in appropriate escalation and reporting. This leads to a decline in the quality of customer support and a deterioration in overall service. This invention aims to reduce the mental burden on operators and improve the quality of customer support by detecting customer harassment in real time and providing appropriate notifications and escalations.

[0766] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0767] In this invention, the server includes: a means for the terminal to acquire voice data when a user initiates a call and transmit it to the server; a means for the server to input the voice data into a voice recognition engine and convert it into text data; a means for the server to use a natural language processing algorithm to detect customer harassment in real time based on the converted text data; a means for the server to notify the terminal when customer harassment is detected and, if necessary, send an escalation notice to an administrator; a means for recording the customer harassment detection results and the content of the call in a log file and generating a report; and a means for storing the generated report in a database. This enables customer harassment to be detected in real time and responded to promptly. Furthermore, by recording the content of the call and the detection results in detail and utilizing them for future analysis and feedback, the quality of customer service can be continuously improved.

[0768] A "user" is a person who acts as an operator at a call center and handles calls with customers.

[0769] A "terminal" refers to a device used by a user, including a calling device and a recording device, that acquires and transmits voice data.

[0770] The "server" refers to a central processing unit that receives voice data sent from a terminal and performs voice recognition and natural language processing.

[0771] "Voice data" refers to digital data that is a recording of the contents of a conversation between a user and a customer.

[0772] A "voice recognition engine" is software or hardware that converts voice data into text data.

[0773] "Text data" refers to character information converted from voice data by a voice recognition engine.

[0774] A "natural language processing algorithm" is a programming technology that analyzes text data and detects customer harassment based on specific keywords and context.

[0775] "Customer harassment" refers to offensive, insulting or inappropriate comments or behavior from a customer toward an operator.

[0776] "Notification" refers to a warning message or information transmission from the server to a terminal or administrator when customer harassment is detected.

[0777] An "escalation notification" is a message sent from the server to an administrator urging emergency action when customer harassment is serious.

[0778] "Log file" refers to a digital file used to record the content of calls and the results of customer harassment detection.

[0779] A "report" is a document that compiles detailed records of calls and customer harassment, and is used for future analysis and feedback.

[0780] "Database" means a digital information storage system for storing generated reports and log files for retrieval and analysis as needed.

[0781] The present invention is implemented as follows to detect customer harassment in real time in a call center, reduce the mental burden on operators, and improve the quality of customer service.

[0782] First, when a user starts a call at a call center, the terminal records the conversation and transmits the resulting voice data to the server in real time. The terminal is usually composed of a call device and a recording device provided as part of the operator tools.

[0783] The server then inputs the received voice data into a voice recognition engine, which can be Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services. The voice recognition engine converts the voice data into text data, which the server uses for further processing.

[0784] The server then analyzes the converted text data using a natural language processing (NLP) algorithm, such as the BERT model or an emotion analysis API. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs context and sentiment analysis to determine whether the text is likely to be customer harassment.

[0785] If customer harassment is detected, the server immediately flags it and sends a notification message to the device. For example, a warning message such as "A customer is making inappropriate remarks. Please respond calmly" will be displayed on the device. In serious cases, the server will send an escalation notification to the administrator, urging them to take urgent action.

[0786] After the call ends, the server logs all conversations and detection results, generates a report, and stores it in a database that stores the time of the customer harassment incident, the offensive keywords used, the customer sentiment analysis results, and the response, which can be used for future analysis and training.

[0787] As a specific example, the user (operator) starts a call, the device starts recording, and sends the voice data to the server in real time. The server uses a speech recognition engine to convert customer comments such as "This product is completely unusable!" into text. The NLP engine then detects the offensive keyword "completely unusable" and determines that the customer is irritated. This is judged to be customer harassment, and the server sends a notification to the device saying, "A customer is making inappropriate comments. Please respond calmly." The server also sends an escalation notification to an administrator, who then intervenes. After the call ends, the server records all content and detection results and generates a report.

[0788] Prompting a generative AI model for input can be done as follows:

[0789] "Please explain in detail the specific processing steps of a system for detecting customer harassment in a call center in real time."

[0790] As described above, this invention utilizes voice recognition and natural language processing technology to detect customer harassment in real time and provide appropriate notifications and escalations, thereby reducing the mental burden on operators and improving the quality of customer service.

[0791] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0792] Step 1:

[0793] When a user starts a call at a call center, the device records the conversation. The device then sends the recorded voice data to the server in real time. At this time, the device uses a microphone to capture the voice data and transfers it to the server in streaming format. The input is the user's voice, and the output is the transmission of voice data to the server. Specifically, when the user presses the "Start Call" button, the device starts recording and continuously sends voice data to the server.

[0794] Step 2:

[0795] The server inputs the received voice data into a voice recognition engine. The voice recognition engine converts the voice data into text data. In this case, Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services are often used. The input is voice data, and the output is the converted text data. Specifically, the server starts the voice recognition engine and inputs the received voice data into the engine. The engine analyzes the voice data and generates the corresponding text data.

[0796] Step 3:

[0797] The server analyzes the converted text data using a natural language processing (NLP) algorithm. The NLP engine detects specific keywords and phrases and also analyzes context and sentiment. Technologies used include the BERT model and emotion analysis API. The input is text data, and the output is the analysis results, including possible customer harassment. Specifically, the text data is input into the NLP engine, which returns the analysis results for keywords and sentiment.

[0798] Step 4:

[0799] If customer harassment is detected, the server sets a flag for response. The server then sends a notification to the device and, if necessary, an escalation notification to the administrator. The input is the NLP analysis result, and the output is the sending of a notification. Specifically, the server sends a message to the device saying, "A customer is making inappropriate remarks. Please respond calmly," and if the situation is serious, it sends a notification to the administrator saying, "Urgent action is required."

[0800] Step 5:

[0801] When the call ends, the server logs all conversation content and detection results and generates a report. The input is the text data of the call content and the detection results of customer harassment, and the output is the generation of a report. Specifically, the server receives a call end trigger, aggregates all data, creates a log file, and launches the report generation module.

[0802] Step 6:

[0803] The server stores the generated reports in a database for later analysis and feedback. The input is the generated report, and the output is saving it to the database. Specifically, the server writes the reports to the database and periodically scans the database as needed to analyze newly accumulated data.

[0804] (Application example 1)

[0805] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0806] When employees deal with customers in physical stores, it is difficult for them to respond appropriately to customer harassment. If appropriate responses to customer harassment are delayed, the mental burden on employees increases, which can result in a decline in the quality of customer service. Furthermore, if customer harassment occurs frequently, a means of quickly communicating its impact to managers is also required. To solve these issues, a system is needed that can detect customer harassment in real time and notify employees appropriately.

[0807] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0808] In this invention, the server includes means for acquiring voice data, means for converting the voice data into text data, means equipped with a natural language processing algorithm for detecting customer harassment based on the converted text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the conversation, means for saving the report in a database, and means for displaying a notification of customer harassment on the smart glasses. This enables employees at physical stores to detect customer harassment in conversations with customers in real time and respond promptly and appropriately, thereby reducing the mental burden on employees and improving the quality of customer service.

[0809] The "means for acquiring voice data" refers to a means having the function of collecting the contents of conversations between employees and customers in a physical store in voice format using a device such as a microphone.

[0810] The "means for converting voice data into text data" refers to a means having a function for converting acquired voice data into text format data using voice recognition technology.

[0811] A "means for detecting customer harassment" is a means that is equipped with a natural language processing algorithm and has the ability to analyze text data and identify customer harassment in real time.

[0812] "Means for notification" refers to means that have the function of immediately notifying employees and managers of warnings and information when customer harassment is detected.

[0813] The "means for generating a report of the detection results of customer harassment and the content of the conversation" is a means having a function for creating a report summarizing the analyzed content of the conversation and the detection results of customer harassment.

[0814] "Means for saving to a database" refers to means having the function of recording the generated report in a database for long-term storage and management.

[0815] "Means for displaying on smart glasses" refers to a means for displaying a notice regarding customer harassment on the display of smart glasses, thereby visually informing employees.

[0816] "Natural language processing algorithms" are a set of methods and technologies that analyze text data, identify emotions and keywords, and detect customer harassment.

[0817] This invention is a system that detects customer harassment (kasuhara) in real time while serving customers in a brick-and-mortar store, reduces the mental burden on employees, and improves the quality of customer service. This system allows employees wearing smart glasses to monitor conversations with customers, and is realized using the following hardware and software.

[0818] Hardware

[0819] 1. Smart glasses: A device worn by employees to capture audio and display notifications.

[0820] 2. Server: A central server for processing voice data and managing the database.

[0821] 3. Microphone: Built into the smart glasses, it records conversations with customers.

[0822] software

[0823] 1. Speech recognition engine: Software that converts voice data into text data (e.g., Google Speech Recognition).

[0824] 2. Natural language processing algorithms: Software that analyzes text data and detects customer harassment (e.g., sentiment analysis models for Hugging Face transformers).

[0825] 3. Database Management System: Software that stores and manages logs and reports.

[0826] Program processing flow (natural language explanation)

[0827] 1. Voice data capture: When a customer and an employee start a conversation, the built-in microphone in the smart glasses captures the voice data, which is then sent to the server in real time.

[0828] 2. Speech recognition: The server inputs the received voice data into a voice recognition engine and converts the voice data into text data. For example, a customer saying, "This product is completely unusable!" is converted into text.

[0829] 3. Natural Language Processing: The server inputs the converted text data into a natural language processing algorithm to detect customer harassment using specific keywords and sentiment analysis, such as "idiot," "fired," and "useless."

[0830] 4. Displaying a notification: If customer harassment is detected, the server will display a warning message on the smart glasses saying, "This may be customer harassment. Please respond calmly." In serious cases, the server will also escalate the situation to an administrator.

[0831] 5. Generate and save report: After the conversation is over, the server generates a report of all the conversation contents and the results of customer harassment detection and saves it in the database. This report includes the time of the customer harassment, the offensive keywords used, the customer sentiment analysis results, and the response.

[0832] This allows employees to respond to customer harassment promptly and appropriately, reducing their mental stress and improving the quality of customer service. The generated reports can also be used for training and support at a later date.

[0833] Adding specific examples

[0834] For example, suppose an employee wearing smart glasses is serving a customer and the customer says, "This product is completely unusable!" This comment is picked up by the smart glasses' microphone and sent to the server as voice data. The voice recognition engine then converts the voice into text data, and the natural language processing algorithm detects the offensive keyword "completely unusable" and negative emotion. This causes the smart glasses to display a message saying, "This may be customer harassment. Please respond calmly."

[0835] Example prompts to input to the generative AI model

[0836] For example, by inputting a prompt such as, "Please analyze what emotion is contained in a customer's statement, 'I can't use this product at all!' and whether it constitutes customer harassment," into the generative AI model, it is possible to perform appropriate emotion analysis and detect customer harassment.

[0837] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0838] Step 1: Acquire audio data

[0839] The user (employee) puts on the smart glasses and starts talking with the customer. At this stage, the microphone built into the smart glasses picks up the voice data. The picked up voice data is sent to the server in real time. The input is the voice of the conversation between the customer and the employee, and the output is the voice data.

[0840] Step 2: Voice Recognition

[0841] The voice data received by the server is input into a voice recognition engine, which converts the voice data into text data. For example, a voice saying, "This product is completely unusable!" is converted into text. The input is voice data, and the output is text data. Voice recognition software such as Google Speech Recognition is used for this process.

[0842] Step 3: Natural Language Processing and Harassment Detection

[0843] The server inputs the converted text data into a natural language processing algorithm to perform specific keyword and sentiment analysis. The natural language processing algorithm (e.g., the sentiment analysis model of the Hugging Face transformers) analyzes the text and determines whether it is likely to be customer harassment. The input is the text data, and the output is the sentiment analysis results and the detection results of customer harassment.

[0844] Step 4: View notifications

[0845] If customer harassment is detected, the server displays a warning message on the smart glasses. For example, a message such as "This may be customer harassment. Please respond calmly" may be displayed on the smart glasses' display. In serious cases, an escalation notification may be sent to an administrator. The input is the customer harassment detection result, and the output is a warning message and an escalation notification.

[0846] Step 5: Generate and save the report

[0847] Once the conversation is over, the server generates a report containing all the conversation content and the results of customer harassment detection. This report includes the time of the harassment, the offensive keywords used, the customer sentiment analysis results, and the response. The generated report is saved in a database. The input is the conversation content and the results of customer harassment detection, and the output is the report data.

[0848] Step 6: Analyze the report and provide feedback

[0849] The server periodically analyzes the report data stored in the database and uses it as material for training programs to improve the quality of operator responses. This analysis aims to improve the mental health of employees and the quality of customer service. The input is the report data from the database, and the output is the analysis results and material for training programs.

[0850] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0851] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. In particular, this system incorporates an emotion engine that recognizes the user's emotions, making it possible to more accurately detect stress caused by customer harassment and take appropriate action at the right time.

[0852] System configuration

[0853] 1. Acquiring audio data

[0854] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0855] 2. Voice recognition (ASR)

[0856] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from "voice to text." For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[0857] 3. Natural Language Processing (NLP)

[0858] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0859] 4. Emotion Recognition by Emotion Engine

[0860] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. For example, emotions such as "anger" or "irritation" can be recognized in real time from the customer's tone of voice and the content of their comments.

[0861] 5. Real-time detection and notification

[0862] If customer harassment or negative emotions are detected, the server immediately flags the situation and sends a notification to the device, warning the operator and providing instructions on how to respond. For example, a message such as "A customer is making inappropriate comments. Please respond calmly."

[0863] 6. Escalation and Administrator Notification

[0864] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[0865] 7. Report generation and data accumulation

[0866] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[0867] 8. Data analysis and feedback

[0868] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[0869] Specific operation example

[0870] Specifically, when the user (operator) starts a call, the device records the call and sends it to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable" and determines that the customer is irritated through context and sentiment analysis. The server then uses an emotion engine to recognize the customer's emotional state in detail and confirms that the emotion is "anger." If this is determined to be customer harassment, the server sends a notification to the device, warning, "The customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing them to intervene.

[0871] After the call ends, the server logs the conversation, the customer harassment detection results, and the emotion analysis results, then generates a report and stores it in a database. The data is then periodically analyzed to help improve the mental health of operators and the quality of customer service. In this way, the system of the present invention can detect customer harassment in real time and respond appropriately, thereby reducing the mental burden on operators and improving the quality of customer service.

[0872] The processing flow will be explained below.

[0873] Step 1:

[0874] The user (operator) starts a call. The device detects this action and automatically starts recording the conversation between the customer and the operator.

[0875] Step 2:

[0876] The device transmits the recorded audio data to the server in real time, and the transmitted audio data includes the entire content of the conversation.

[0877] Step 3:

[0878] The server inputs the received voice data into a voice recognition engine. This engine analyzes the voice data and converts it into text data. For example, a voice saying "This product is completely unusable!" is converted into text data saying "This product is completely unusable!"

[0879] Step 4:

[0880] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[0881] Step 5:

[0882] The server uses the emotion engine based on the analysis results of the NLP engine to perform a detailed analysis of the user's emotional state. For example, it can recognize the customer's emotions, such as "anger" or "irritation," in real time, taking into account the tone of voice and the content of what is being said.

[0883] Step 6:

[0884] The server integrates the results from the emotion engine and the NLP engine to evaluate the possibility of customer harassment. If it is determined to be customer harassment, it sets a corresponding flag. This flag will guide the next action.

[0885] Step 7:

[0886] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[0887] Step 8:

[0888] If the customer harassment is serious, the server will send an escalation notification to an administrator, who can monitor the call content in real time and directly intervene if necessary.

[0889] Step 9:

[0890] The user (operator) takes the specified action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[0891] Step 10:

[0892] The server generates a report containing the time of occurrence of the customer harassment, the offensive keywords used, the results of the sentiment analysis, and the details of the response, and stores this in a database.

[0893] Step 11:

[0894] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[0895] Example 2

[0896] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0897] Customer harassment in call centers increases the mental burden on operators and reduces the quality of customer service. Current systems have difficulty detecting customer harassment in real time and responding appropriately, and are insufficiently effective in reducing operator stress. For this reason, there is a need for a system that can accurately detect customer harassment and take prompt action.

[0898] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: a means for the user to acquire voice data; a means for the terminal to transmit the voice data to the server in real time; a means for the server to convert the voice data into text data; a means for the server to detect customer harassment based on the text data; a means for the server to notify the terminal when customer harassment is detected; a means for the server to generate a report on the detection results of the customer harassment and the contents of the call; a means for the server to store the report in a database; and a means for the server to analyze the information stored in the database and provide feedback to reduce the user's mental stress. This enables customer harassment to be detected in real time, immediately notified to an operator, and escalated to a manager. Furthermore, recording the detection results and the contents of the call in a report contributes to future countermeasures and improving the mental health of operators.

[0899] "Voice data" is digital data that records the content of a conversation between a user (operator) and a customer.

[0900] "Text data" is character string data converted from voice data by a voice recognition engine.

[0901] "Customer harassment" refers to inappropriate words or actions directed at operators by customers, which increase the mental stress on the operators.

[0902] A "server" is a central computer system that processes voice data and performs speech recognition, natural language processing, and sentiment analysis.

[0903] A "terminal" is a computer device used by a user (operator) and is a device that acquires voice data and transmits it to a server.

[0904] "Notification" means a warning or instruction message sent to an operator or administrator when customer harassment is detected.

[0905] A "report" is written data that records the content of calls and the results of customer harassment detection, and is analysis material that is stored in a database.

[0906] A "database" is a system for efficiently storing and managing large amounts of data, and is a device for accumulating reports and call records.

[0907] "User" refers to an operator who handles customer service at a call center, and is the entity that acquires voice data and receives notifications from the system.

[0908] "Emotion analysis" is the process of analyzing and recognizing a user's emotional state from text data and voice tone.

[0909] "Escalation" is the process of notifying an administrator and encouraging direct intervention when customer harassment is deemed severe.

[0910] "Feedback" is the process of analyzing information stored in a database and providing information that helps improve the quality of the user's response and reduce mental stress.

[0911] This invention is a system that detects customer harassment in a call center in real time and takes appropriate action to reduce the mental burden on operators and improve the quality of customer service. Specific embodiments are described below.

[0912] Hardware and software used

[0913] Hardware:

[0914] Terminal: A computer used by an operator. This terminal has the function of acquiring voice data and sending it to a server.

[0915] Server: A central processing unit for speech recognition, natural language processing, and sentiment analysis.

[0916] software:

[0917] Speech recognition engine (e.g. Google Speech-to-Text)

[0918] Natural Language Processing (NLP) engines (e.g., SpaCy, NLTK)

[0919] Emotion engine (e.g. IBM Watson Tone Analyzer)

[0920] Data processing and calculation

[0921] 1. Acquiring audio data

[0922] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[0923] 2. Voice Recognition

[0924] The server inputs the received voice data into a voice recognition engine, which converts the voice data into "voice to text."

[0925] 3. Natural Language Processing

[0926] The server passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and performs contextual and sentiment analysis.

[0927] 4. Emotion Recognition by Emotion Engine

[0928] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state.

[0929] 5. Real-time detection and notification

[0930] If customer harassment and negative emotions are detected, the server immediately raises a response flag and sends a notification to the device.

[0931] 6. Escalation and Administrator Notification

[0932] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who will monitor the call in real time and intervene directly if necessary.

[0933] 7. Report generation and data accumulation

[0934] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[0935] 8. Data analysis and feedback

[0936] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[0937] Specific examples

[0938] Acquiring audio data:

[0939] When the operator starts a call with a customer, the device starts recording using a VoIP application and transmits the data to the server in real time.

[0940] Voice Recognition:

[0941] The server uses the Google Speech-to-Text API to analyze the received voice data and generate text data such as, "This product is completely unusable!"

[0942] Natural Language Processing:

[0943] The server uses an NLP engine (e.g., SpaCy) to detect offensive keywords and phrases, such as "totally unusable," and analyzes customer frustration.

[0944] Emotion Recognition with Emotion Engine:

[0945] The server uses IBM Watson Tone Analyzer to recognize the emotion of "anger" in real time from the customer's voice tone and text.

[0946] Real-time detection and notification:

[0947] When the server recognizes customer harassment, it displays a message on the operator's terminal saying, "A customer is making inappropriate remarks. Please respond calmly."

[0948] An example of a prompt for a generative AI model is:

[0949] "Develop a system that records call content in the call center in real time and detects customer harassment. Specific steps include voice recognition, natural language processing, and sentiment analysis, and notifying operators or managers as necessary. After the call ends, a report should be generated and saved in a database."

[0950] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0951] Step 1: Acquire audio data

[0952] When a user (operator) starts a call, the device records the call. The recorded voice data is sent to the server in real time as input. This process captures the voice call data between the customer and the operator. Specifically, when an operator makes a call using a VoIP application, the software automatically records the call and sends it to the server as digital voice data.

[0953] Step 2: Voice Recognition

[0954] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. In this process, the voice data is converted into character string data. For example, the Google Speech-to-Text API is used to convert the voice "This product is completely unusable!" into text. Text data is generated as the output.

[0955] Step 3: Natural Language Processing

[0956] The server passes the converted text data to a natural language processing (NLP) engine. The NLP engine detects specific keywords and phrases and performs contextual and sentiment analysis. Based on the input text data, keyword extraction and context understanding are performed. For example, an NLP engine (e.g., SpaCy) can be used to detect the offensive keyword "completely unusable" and infer customer frustration. The results of the contextual and sentiment analysis are generated as output.

[0957] Step 4: Emotion Recognition with the Emotion Engine

[0958] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. The text data and analysis results obtained in the previous step are used as input. For example, IBM Watson Tone Analyzer is used to recognize the emotion of "anger" in real time from text and voice tone. This process clarifies the customer's specific emotional state, and the emotion analysis results are obtained as output.

[0959] Step 5: Real-time detection and notification

[0960] If the server detects customer harassment or negative emotions, it immediately raises a flag and sends a notification to the terminal. The results of emotion analysis are used as input. Specifically, based on the detected emotional state (e.g., anger), a message such as "A customer is making inappropriate remarks. Please respond calmly" is displayed on the operator's terminal. A notification message is sent as output.

[0961] Step 6: Escalation and Administrator Notification

[0962] If the customer harassment is deemed to be serious, the server sends an escalation notification to the administrator. The further analyzed sentiment analysis results are used as input. Specifically, an alert is sent to the administrator stating, "Customer harassment has been detected in the current call. Please monitor the call and intervene if necessary." The escalation notification is sent as output.

[0963] Step 7: Report generation and data accumulation

[0964] When the call ends, the server records the content of the conversation, the results of customer harassment detection, and the results of sentiment analysis in a log and generates a report. The call data and analysis results are used as input. In this process, a detailed report is created and saved in the database. Specifically, a report stating "Customer harassment was detected in the call with the customer. Please refer to the log for details" is generated and saved in the database. Report data is generated as output.

[0965] Step 8: Data analysis and feedback

[0966] The server periodically analyzes the information stored in the database to help improve the quality of operator support and customer service. As input, multiple reports stored in the database are analyzed. This process uses the analysis results as material for training programs and provides feedback to operators. Specifically, monthly reports are generated, and the response status of each operator and successful examples of customer harassment responses are analyzed and used as material for the next training session. As output, an analysis report and improvement proposals are created.

[0967] (Application example 2)

[0968] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0969] Customer harassment in call centers increases the mental burden on operators and can be a factor in reducing the quality of customer service. In particular, in urgent tasks such as security services, customer stress levels tend to rise, potentially increasing the frequency of customer harassment. Effective countermeasures are needed, but current systems struggle to respond in real time, and there is no mechanism in place for operators or managers to intervene quickly. For this reason, there is a strong demand for the development of a system that can quickly and accurately detect customer harassment and respond in real time.

[0970] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0971] In this invention, the server includes means for acquiring voice data, means for converting voice data into text data, means for detecting customer harassment based on the text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the call, means for saving the report in a database, means for analyzing emotions in the content of the call, and means for issuing response instructions to an operator and a manager based on the content of the call. This makes it possible to detect customer harassment in real time and respond quickly.

[0972] "Means for acquiring voice data" refers to a device or software that records the conversation between the customer and the operator and transmits the voice data to a server in real time.

[0973] The "means for converting voice data into text data" refers to a device or software that analyzes acquired voice data and converts it into text data using voice recognition technology.

[0974] A "means for detecting customer harassment based on text data" is a device or software that uses natural language processing technology on text data to evaluate whether or not there is customer harassment through specific keyword and sentiment analysis.

[0975] "Means for notifying when customer harassment is detected" refers to a device or software that issues appropriate warnings and instructions to operators or administrators in real time when customer harassment is detected.

[0976] The "means for generating a report of the customer harassment detection results and the content of the call" is a device or software that formats the customer harassment detection results and the related content of the call and outputs them as a report that can be referenced at a later date.

[0977] "Means for storing reports in a database" refers to a device or software that stores generated reports in a database system for managing and storing them.

[0978] The "means for analyzing the emotion of the call content" is a device or software that analyzes data including voice-recognized text data and voice tone to evaluate the emotional state of the customer.

[0979] "Means for issuing response instructions to operators and administrators based on the content of calls" refers to devices or software that issue specific response instructions to operators and administrators based on analyzed customer emotional data and the results of customer harassment detection.

[0980] The present invention provides a system that can detect customer harassment in a call center in real time and take appropriate action. Specific embodiments for carrying out the present invention will be described in detail below.

[0981] System configuration

[0982] The system of the present invention mainly acquires and analyzes voice data and implements a series of processes to detect customer harassment, which includes the following components:

[0983] 1. How to obtain audio data

[0984] It uses a microphone (built-in or external) installed on the device, such as a smartphone or head-mounted display (HMD).

[0985] When the call starts, the voice data is acquired and sent to the server in real time.

[0986] 2. Means of converting audio data into text data

[0987] On the server side, voice data is converted into text data using speech recognition technology, primarily using speech recognition engines such as Google Speech Recognition API.

[0988] 3. How to detect customer harassment based on text data

[0989] On the server side, the text data is analyzed using an NLP (Natural Language Processing) engine, using libraries such as TextBlob and NLTK.

[0990] 4. Means of notification if customer harassment is detected

[0991] Based on the detection results, a real-time warning is sent to the operator, with the message "A customer is making inappropriate remarks. Please respond calmly." appearing on the terminal display.

[0992] Administrators are notified of the escalation and are encouraged to prepare for intervention.

[0993] 5. A method for generating reports on customer harassment detection results and call content

[0994] After the call is completed, the server generates a report summarizing the detection results and the contents of the call, which is then stored in a database.

[0995] 6. How to save the report to a database

[0996] Use a database system such as MySQL or PostgreSQL to store reports.

[0997] 7. A way to analyze the sentiment of calls

[0998] Based on the content and tone of the customer's speech, sentiment is analyzed in detail using sentiment analysis engines such as TextBlob and VADER.

[0999] 8. Means of issuing instructions to operators and administrators based on the content of calls

[1000] Based on the results of sentiment analysis and customer harassment detection, the system sends specific instructions to operators and administrators, encouraging operators to remain calm and informing administrators of the need for intervention.

[1001] Specific examples

[1002] The following scenarios can be considered as examples of actual operation.

[1003] 1. An operator working at a security service anomaly reporting call center receives a call from a customer using a smartphone.

[1004] 2. A customer says something like, "This security system isn't working at all! What can you do about it?"

[1005] 3. The system converts the speech into text in real time and detects the emotion of "anger."

[1006] 4. A notification will appear on the operator's smartphone saying, "A customer is making inappropriate remarks. Please respond calmly."

[1007] 5. An escalation notification is sent to the administrator, who prepares to intervene.

[1008] Prompt Sentence Examples

[1009] An example of a prompt sentence to input to the generative AI model is as follows:

[1010] Develop an application that detects specific phrases and emotions in real time during calls with customers and notifies them if customer harassment occurs. This application uses Python for speech recognition and natural language processing. Specifically, it detects phrases containing the emotion "anger" from the converted speech and notifies the operator and administrator. Examples of customer harassment include "It's completely unusable" and "What are you going to do about it?"

[1011] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1012] Step 1:

[1013] The terminal captures the voice data of the customer and the operator as soon as the call starts, and the captured voice data is sent to the server in real time.

[1014] Input: Voice of the customer and operator

[1015] Output: Audio data

[1016] Specific operation: The device's microphone captures audio and transmits the audio data to the server in real time.

[1017] Step 2:

[1018] The server inputs the received voice data into a voice recognition engine (for example, Google Speech Recognition API) and converts the voice data into text data.

[1019] Input: Audio data

[1020] Output: Text data

[1021] Specific operation: The speech recognition engine analyzes the voice data and converts it into corresponding text data.

[1022] Step 3:

[1023] The server inputs the text data into a natural language processing (NLP) engine (e.g., TextBlob or NLTK) to analyze the text data, performing specific keyword detection and sentiment analysis.

[1024] Input: Text data

[1025] Output: Analysis results (keywords and emotional information)

[1026] How it works: The NLP engine analyzes text data and extracts information through keyword extraction and sentiment analysis.

[1027] Step 4:

[1028] The server detects customer harassment based on the analysis results. If negative emotions such as irritation or anger are detected, it determines that there is a high possibility of customer harassment.

[1029] Input: Analysis results

[1030] Output: Customer harassment judgment

[1031] Specific operation: The emotion engine evaluates the analysis results and determines whether or not there are any negative emotions.

[1032] Step 5:

[1033] If customer harassment is detected, the server sends a notification to the terminal and instructs the operator to respond. It also sends an escalation notification to the administrator.

[1034] Input: Customer Harassment Judgment

[1035] Output: Notification to operators and administrators

[1036] What happens: The server generates a notification message, displays a warning on the operator's screen, and sends an escalation notification to the administrator.

[1037] Step 6:

[1038] Once the call ends, the server generates a report based on the content of the call, the results of customer harassment detection, and the results of sentiment analysis.

[1039] Input: Call content, customer harassment detection results, emotion analysis results

[1040] Output: Call Report

[1041] Specific operation: The server compiles the call content and analysis results and creates a formatted report.

[1042] Step 7:

[1043] The server generates reports that are saved in a database and can be used later to improve the mental health of operators and the quality of customer service.

[1044] Input: Call Report

[1045] Output: Save to database

[1046] Specific operation: The server stores the report in a database system (e.g., MySQL or PostgreSQL).

[1047] Step 8:

[1048] The server periodically analyzes the information stored in the database and uses it to improve the quality of operator support and customer service. This information is also used as material for training programs.

[1049] Input: Reports in the database

[1050] Output: Analysis results, training materials

[1051] What it does: The server retrieves information from the database and extracts useful insights through data analysis.

[1052] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1053] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1054] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1055] [Fourth embodiment]

[1056] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1057] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1058] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1059] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1060] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1061] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1062] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1063] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1064] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1065] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1066] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1067] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1068] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1069] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. This system acquires voice data, converts it into text data using speech recognition technology, detects customer harassment using natural language processing (NLP), and has the ability to notify or escalate the situation in real time as necessary.

[1070] System configuration

[1071] 1. Acquiring audio data

[1072] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[1073] 2. Voice recognition (ASR)

[1074] The server inputs the received voice data into a voice recognition engine, which converts the voice data into text data. For example, a voice saying "Hello, this is customer support" is converted into text "Hello, this is customer support."

[1075] 3. Natural Language Processing (NLP)

[1076] The server then analyzes the converted text data using advanced natural language processing (NLP) algorithms. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs further context and sentiment analysis to determine the likelihood of customer harassment.

[1077] 4. Real-time detection and notification

[1078] If customer harassment is detected, the server immediately raises a flag for response. The server then sends a notification to the device. For example, the device may display a message saying, "A customer is making inappropriate remarks. Please respond calmly." If serious customer harassment occurs, the server sends an escalation notification to an administrator so that the administrator can intervene.

[1079] 5. Report generation and data accumulation

[1080] After the call ends, the server logs all conversations and detection results, and compiles a report containing the time of the incident, the offensive keywords used, the customer sentiment analysis results, and the response. The server stores this report in a database.

[1081] 6. Data analysis and feedback

[1082] The server periodically analyzes the information stored in the database and uses it as material for training programs aimed at improving the mental health and quality of customer service, thereby reducing the mental burden on operators and enabling the continuous improvement of customer service quality.

[1083] Specific operation example

[1084] Assume that a user (operator) starts a call. The device records the contents of the call and sends them to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable," and determines that the customer is irritated through context and sentiment analysis. As this is determined to be customer harassment, the server sends a notification to the device, displaying a warning message saying, "A customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing the administrator to intervene.

[1085] After the call ends, the server logs all call content and any harassment findings, generates a report, and stores it in a database. This information is later used for operator training and support, and to improve the quality of customer service.

[1086] As described above, the present invention can detect customer harassment in real time and take appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service.

[1087] The processing flow will be explained below.

[1088] Step 1:

[1089] The user (operator) starts a call. The device detects this action and automatically starts recording the call.

[1090] Step 2:

[1091] The device transmits real-time recorded voice data to the edge server, which includes all conversations between the customer and the operator.

[1092] Step 3:

[1093] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[1094] Step 4:

[1095] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[1096] Step 5:

[1097] The server evaluates the possibility of customer harassment based on the analysis results from the NLP engine. If it is determined to be customer harassment, it sets a response flag. This response flag will serve as a guide for the next action.

[1098] Step 6:

[1099] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[1100] Step 7:

[1101] If the harassment becomes serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[1102] Step 8:

[1103] The user (operator) takes appropriate action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[1104] Step 9:

[1105] The server generates a report containing the time when the customer harassment occurred, the offensive keywords used, the results of the customer sentiment analysis, and the details of the response, and stores the report in a database.

[1106] Step 10:

[1107] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[1108] Example 1

[1109] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1110] In conventional customer support systems, operators must directly respond to customer harassment, placing a heavy mental burden on the operators. Furthermore, even when customer harassment occurs, it is difficult to detect and respond in real time, which can lead to delays in appropriate escalation and reporting. This leads to a decline in the quality of customer support and a deterioration in overall service. This invention aims to reduce the mental burden on operators and improve the quality of customer support by detecting customer harassment in real time and providing appropriate notifications and escalations.

[1111] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1112] In this invention, the server includes: a means for the terminal to acquire voice data when a user initiates a call and transmit it to the server; a means for the server to input the voice data into a voice recognition engine and convert it into text data; a means for the server to use a natural language processing algorithm to detect customer harassment in real time based on the converted text data; a means for the server to notify the terminal when customer harassment is detected and, if necessary, send an escalation notice to an administrator; a means for recording the customer harassment detection results and the content of the call in a log file and generating a report; and a means for storing the generated report in a database. This enables customer harassment to be detected in real time and responded to promptly. Furthermore, by recording the content of the call and the detection results in detail and utilizing them for future analysis and feedback, the quality of customer service can be continuously improved.

[1113] A "user" is a person who acts as an operator at a call center and handles calls with customers.

[1114] A "terminal" refers to a device used by a user, including a calling device and a recording device, that acquires and transmits voice data.

[1115] The "server" refers to a central processing unit that receives voice data sent from a terminal and performs voice recognition and natural language processing.

[1116] "Voice data" refers to digital data that is a recording of the contents of a conversation between a user and a customer.

[1117] A "voice recognition engine" is software or hardware that converts voice data into text data.

[1118] "Text data" refers to character information converted from voice data by a voice recognition engine.

[1119] A "natural language processing algorithm" is a programming technology that analyzes text data and detects customer harassment based on specific keywords and context.

[1120] "Customer harassment" refers to offensive, insulting or inappropriate comments or behavior from a customer toward an operator.

[1121] "Notification" refers to a warning message or information transmission from the server to a terminal or administrator when customer harassment is detected.

[1122] An "escalation notification" is a message sent from the server to an administrator urging emergency action when customer harassment is serious.

[1123] "Log file" refers to a digital file used to record the content of calls and the results of customer harassment detection.

[1124] A "report" is a document that compiles detailed records of calls and customer harassment, and is used for future analysis and feedback.

[1125] "Database" means a digital information storage system for storing generated reports and log files for retrieval and analysis as needed.

[1126] The present invention is implemented as follows to detect customer harassment in real time in a call center, reduce the mental burden on operators, and improve the quality of customer service.

[1127] First, when a user starts a call at a call center, the terminal records the conversation and transmits the resulting voice data to the server in real time. The terminal is usually composed of a call device and a recording device provided as part of the operator tools.

[1128] The server then inputs the received voice data into a voice recognition engine, which can be Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services. The voice recognition engine converts the voice data into text data, which the server uses for further processing.

[1129] The server then analyzes the converted text data using a natural language processing (NLP) algorithm, such as the BERT model or an emotion analysis API. The NLP engine detects specific keywords (e.g., "idiot," "fired," "useless") and performs context and sentiment analysis to determine whether the text is likely to be customer harassment.

[1130] If customer harassment is detected, the server immediately flags it and sends a notification message to the device. For example, a warning message such as "A customer is making inappropriate remarks. Please respond calmly" will be displayed on the device. In serious cases, the server will send an escalation notification to the administrator, urging them to take urgent action.

[1131] After the call ends, the server logs all conversations and detection results, generates a report, and stores it in a database that stores the time of the customer harassment incident, the offensive keywords used, the customer sentiment analysis results, and the response, which can be used for future analysis and training.

[1132] As a specific example, the user (operator) starts a call, the device starts recording, and sends the voice data to the server in real time. The server uses a speech recognition engine to convert customer comments such as "This product is completely unusable!" into text. The NLP engine then detects the offensive keyword "completely unusable" and determines that the customer is irritated. This is judged to be customer harassment, and the server sends a notification to the device saying, "A customer is making inappropriate comments. Please respond calmly." The server also sends an escalation notification to an administrator, who then intervenes. After the call ends, the server records all content and detection results and generates a report.

[1133] Prompting a generative AI model for input can be done as follows:

[1134] "Please explain in detail the specific processing steps of a system for detecting customer harassment in a call center in real time."

[1135] As described above, this invention utilizes voice recognition and natural language processing technology to detect customer harassment in real time and provide appropriate notifications and escalations, thereby reducing the mental burden on operators and improving the quality of customer service.

[1136] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1137] Step 1:

[1138] When a user starts a call at a call center, the device records the conversation. The device then sends the recorded voice data to the server in real time. At this time, the device uses a microphone to capture the voice data and transfers it to the server in streaming format. The input is the user's voice, and the output is the transmission of voice data to the server. Specifically, when the user presses the "Start Call" button, the device starts recording and continuously sends voice data to the server.

[1139] Step 2:

[1140] The server inputs the received voice data into a voice recognition engine. The voice recognition engine converts the voice data into text data. In this case, Google Cloud Speech-to-Text or Microsoft Azure Cognitive Services are often used. The input is voice data, and the output is the converted text data. Specifically, the server starts the voice recognition engine and inputs the received voice data into the engine. The engine analyzes the voice data and generates the corresponding text data.

[1141] Step 3:

[1142] The server analyzes the converted text data using a natural language processing (NLP) algorithm. The NLP engine detects specific keywords and phrases and also analyzes context and sentiment. Technologies used include the BERT model and emotion analysis API. The input is text data, and the output is the analysis results, including possible customer harassment. Specifically, the text data is input into the NLP engine, which returns the analysis results for keywords and sentiment.

[1143] Step 4:

[1144] If customer harassment is detected, the server sets a flag for response. The server then sends a notification to the device and, if necessary, an escalation notification to the administrator. The input is the NLP analysis result, and the output is the sending of a notification. Specifically, the server sends a message to the device saying, "A customer is making inappropriate remarks. Please respond calmly," and if the situation is serious, it sends a notification to the administrator saying, "Urgent action is required."

[1145] Step 5:

[1146] When the call ends, the server logs all conversation content and detection results and generates a report. The input is the text data of the call content and the detection results of customer harassment, and the output is the generation of a report. Specifically, the server receives a call end trigger, aggregates all data, creates a log file, and launches the report generation module.

[1147] Step 6:

[1148] The server stores the generated reports in a database for later analysis and feedback. The input is the generated report, and the output is saving it to the database. Specifically, the server writes the reports to the database and periodically scans the database as needed to analyze newly accumulated data.

[1149] (Application example 1)

[1150] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1151] When employees deal with customers in physical stores, it is difficult for them to respond appropriately to customer harassment. If appropriate responses to customer harassment are delayed, the mental burden on employees increases, which can result in a decline in the quality of customer service. Furthermore, if customer harassment occurs frequently, a means of quickly communicating its impact to managers is also required. To solve these issues, a system is needed that can detect customer harassment in real time and notify employees appropriately.

[1152] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1153] In this invention, the server includes means for acquiring voice data, means for converting the voice data into text data, means equipped with a natural language processing algorithm for detecting customer harassment based on the converted text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the conversation, means for saving the report in a database, and means for displaying a notification of customer harassment on the smart glasses. This enables employees at physical stores to detect customer harassment in conversations with customers in real time and respond promptly and appropriately, thereby reducing the mental burden on employees and improving the quality of customer service.

[1154] The "means for acquiring voice data" refers to a means having the function of collecting the contents of conversations between employees and customers in a physical store in voice format using a device such as a microphone.

[1155] The "means for converting voice data into text data" refers to a means having a function for converting acquired voice data into text format data using voice recognition technology.

[1156] A "means for detecting customer harassment" is a means that is equipped with a natural language processing algorithm and has the ability to analyze text data and identify customer harassment in real time.

[1157] "Means for notification" refers to means that have the function of immediately notifying employees and managers of warnings and information when customer harassment is detected.

[1158] The "means for generating a report of the detection results of customer harassment and the content of the conversation" is a means having a function for creating a report summarizing the analyzed content of the conversation and the detection results of customer harassment.

[1159] "Means for saving to a database" refers to means having the function of recording the generated report in a database for long-term storage and management.

[1160] "Means for displaying on smart glasses" refers to a means for displaying a notice regarding customer harassment on the display of smart glasses, thereby visually informing employees.

[1161] "Natural language processing algorithms" are a set of methods and technologies that analyze text data, identify emotions and keywords, and detect customer harassment.

[1162] This invention is a system that detects customer harassment (kasuhara) in real time while serving customers in a brick-and-mortar store, reduces the mental burden on employees, and improves the quality of customer service. This system allows employees wearing smart glasses to monitor conversations with customers, and is realized using the following hardware and software.

[1163] Hardware

[1164] 1. Smart glasses: A device worn by employees to capture audio and display notifications.

[1165] 2. Server: A central server for processing voice data and managing the database.

[1166] 3. Microphone: Built into the smart glasses, it records conversations with customers.

[1167] software

[1168] 1. Speech recognition engine: Software that converts voice data into text data (e.g., Google Speech Recognition).

[1169] 2. Natural language processing algorithms: Software that analyzes text data and detects customer harassment (e.g., sentiment analysis models for Hugging Face transformers).

[1170] 3. Database Management System: Software that stores and manages logs and reports.

[1171] Program processing flow (natural language explanation)

[1172] 1. Voice data capture: When a customer and an employee start a conversation, the built-in microphone in the smart glasses captures the voice data, which is then sent to the server in real time.

[1173] 2. Speech recognition: The server inputs the received voice data into a voice recognition engine and converts the voice data into text data. For example, a customer saying, "This product is completely unusable!" is converted into text.

[1174] 3. Natural Language Processing: The server inputs the converted text data into a natural language processing algorithm to detect customer harassment using specific keywords and sentiment analysis, such as "idiot," "fired," and "useless."

[1175] 4. Displaying a notification: If customer harassment is detected, the server will display a warning message on the smart glasses saying, "This may be customer harassment. Please respond calmly." In serious cases, the server will also escalate the situation to an administrator.

[1176] 5. Generate and save report: After the conversation is over, the server generates a report of all the conversation contents and the results of customer harassment detection and saves it in the database. This report includes the time of the customer harassment, the offensive keywords used, the customer sentiment analysis results, and the response.

[1177] This allows employees to respond to customer harassment promptly and appropriately, reducing their mental stress and improving the quality of customer service. The generated reports can also be used for training and support at a later date.

[1178] Adding specific examples

[1179] For example, suppose an employee wearing smart glasses is serving a customer and the customer says, "This product is completely unusable!" This comment is picked up by the smart glasses' microphone and sent to the server as voice data. The voice recognition engine then converts the voice into text data, and the natural language processing algorithm detects the offensive keyword "completely unusable" and negative emotion. This causes the smart glasses to display a message saying, "This may be customer harassment. Please respond calmly."

[1180] Example prompts to input to the generative AI model

[1181] For example, by inputting a prompt such as, "Please analyze what emotion is contained in a customer's statement, 'I can't use this product at all!' and whether it constitutes customer harassment," into the generative AI model, it is possible to perform appropriate emotion analysis and detect customer harassment.

[1182] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1183] Step 1: Acquire audio data

[1184] The user (employee) puts on the smart glasses and starts talking with the customer. At this stage, the microphone built into the smart glasses picks up the voice data. The picked up voice data is sent to the server in real time. The input is the voice of the conversation between the customer and the employee, and the output is the voice data.

[1185] Step 2: Voice Recognition

[1186] The voice data received by the server is input into a voice recognition engine, which converts the voice data into text data. For example, a voice saying, "This product is completely unusable!" is converted into text. The input is voice data, and the output is text data. Voice recognition software such as Google Speech Recognition is used for this process.

[1187] Step 3: Natural Language Processing and Harassment Detection

[1188] The server inputs the converted text data into a natural language processing algorithm to perform specific keyword and sentiment analysis. The natural language processing algorithm (e.g., the sentiment analysis model of the Hugging Face transformers) analyzes the text and determines whether it is likely to be customer harassment. The input is the text data, and the output is the sentiment analysis results and the detection results of customer harassment.

[1189] Step 4: View notifications

[1190] If customer harassment is detected, the server displays a warning message on the smart glasses. For example, a message such as "This may be customer harassment. Please respond calmly" may be displayed on the smart glasses' display. In serious cases, an escalation notification may be sent to an administrator. The input is the customer harassment detection result, and the output is a warning message and an escalation notification.

[1191] Step 5: Generate and save the report

[1192] Once the conversation is over, the server generates a report containing all the conversation content and the results of customer harassment detection. This report includes the time of the harassment, the offensive keywords used, the customer sentiment analysis results, and the response. The generated report is saved in a database. The input is the conversation content and the results of customer harassment detection, and the output is the report data.

[1193] Step 6: Analyze the report and provide feedback

[1194] The server periodically analyzes the report data stored in the database and uses it as material for training programs to improve the quality of operator responses. This analysis aims to improve the mental health of employees and the quality of customer service. The input is the report data from the database, and the output is the analysis results and material for training programs.

[1195] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1196] This invention is a system that detects customer harassment in call centers in real time and takes appropriate action, thereby reducing the mental burden on operators and improving the quality of customer service. In particular, this system incorporates an emotion engine that recognizes the user's emotions, making it possible to more accurately detect stress caused by customer harassment and take appropriate action at the right time.

[1197] System configuration

[1198] 1. Acquiring audio data

[1199] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[1200] 2. Voice recognition (ASR)

[1201] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from "voice to text." For example, the voice phrase "This product is completely unusable!" is converted into text data "This product is completely unusable!"

[1202] 3. Natural Language Processing (NLP)

[1203] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[1204] 4. Emotion Recognition by Emotion Engine

[1205] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. For example, emotions such as "anger" or "irritation" can be recognized in real time from the customer's tone of voice and the content of their comments.

[1206] 5. Real-time detection and notification

[1207] If customer harassment or negative emotions are detected, the server immediately flags the situation and sends a notification to the device, warning the operator and providing instructions on how to respond. For example, a message such as "A customer is making inappropriate comments. Please respond calmly."

[1208] 6. Escalation and Administrator Notification

[1209] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who can monitor the call in real time and intervene directly if necessary.

[1210] 7. Report generation and data accumulation

[1211] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[1212] 8. Data analysis and feedback

[1213] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[1214] Specific operation example

[1215] Specifically, when the user (operator) starts a call, the device records the call and sends it to the server in real time. The server uses a speech recognition engine to convert the customer's statement, "This product is completely unusable!" into text. The server then uses an NLP engine to detect the offensive keyword "completely unusable" and determines that the customer is irritated through context and sentiment analysis. The server then uses an emotion engine to recognize the customer's emotional state in detail and confirms that the emotion is "anger." If this is determined to be customer harassment, the server sends a notification to the device, warning, "The customer is making inappropriate remarks. Please respond calmly." The server also sends an escalation notification to an administrator, allowing them to intervene.

[1216] After the call ends, the server logs the conversation, the customer harassment detection results, and the emotion analysis results, then generates a report and stores it in a database. The data is then periodically analyzed to help improve the mental health of operators and the quality of customer service. In this way, the system of the present invention can detect customer harassment in real time and respond appropriately, thereby reducing the mental burden on operators and improving the quality of customer service.

[1217] The processing flow will be explained below.

[1218] Step 1:

[1219] The user (operator) starts a call. The device detects this action and automatically starts recording the conversation between the customer and the operator.

[1220] Step 2:

[1221] The device transmits the recorded audio data to the server in real time, and the transmitted audio data includes the entire content of the conversation.

[1222] Step 3:

[1223] The server inputs the received voice data into a voice recognition engine. This engine analyzes the voice data and converts it into text data. For example, a voice saying "This product is completely unusable!" is converted into text data saying "This product is completely unusable!"

[1224] Step 4:

[1225] The server then passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and then performs contextual and sentiment analysis. For example, the keyword "completely unusable" is detected, and the phrase can be used to infer customer frustration.

[1226] Step 5:

[1227] The server uses the emotion engine based on the analysis results of the NLP engine to perform a detailed analysis of the user's emotional state. For example, it can recognize the customer's emotions, such as "anger" or "irritation," in real time, taking into account the tone of voice and the content of what is being said.

[1228] Step 6:

[1229] The server integrates the results from the emotion engine and the NLP engine to evaluate the possibility of customer harassment. If it is determined to be customer harassment, it sets a corresponding flag. This flag will guide the next action.

[1230] Step 7:

[1231] The server sends a notification to the device based on the response flag. The notification includes a warning message for the operator. For example, a message such as "A customer is making inappropriate remarks. Please respond calmly."

[1232] Step 8:

[1233] If the customer harassment is serious, the server will send an escalation notification to an administrator, who can monitor the call content in real time and directly intervene if necessary.

[1234] Step 9:

[1235] The user (operator) takes the specified action and the call ends. The server detects the end of the call and records all conversation content and the results of the customer harassment detection in a log.

[1236] Step 10:

[1237] The server generates a report containing the time of occurrence of the customer harassment, the offensive keywords used, the results of the sentiment analysis, and the details of the response, and stores this in a database.

[1238] Step 11:

[1239] The server periodically analyzes the information stored in the database and uses it for training programs to support operators and improve the quality of customer service, thereby reducing the mental burden on operators and continuously improving the quality of customer service.

[1240] Example 2

[1241] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1242] Customer harassment in call centers increases the mental burden on operators and reduces the quality of customer service. Current systems have difficulty detecting customer harassment in real time and responding appropriately, and are insufficiently effective in reducing operator stress. For this reason, there is a need for a system that can accurately detect customer harassment and take prompt action.

[1243] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: a means for the user to acquire voice data; a means for the terminal to transmit the voice data to the server in real time; a means for the server to convert the voice data into text data; a means for the server to detect customer harassment based on the text data; a means for the server to notify the terminal when customer harassment is detected; a means for the server to generate a report on the detection results of the customer harassment and the contents of the call; a means for the server to store the report in a database; and a means for the server to analyze the information stored in the database and provide feedback to reduce the user's mental stress. This enables customer harassment to be detected in real time, immediately notified to an operator, and escalated to a manager. Furthermore, recording the detection results and the contents of the call in a report contributes to future countermeasures and improving the mental health of operators.

[1244] "Voice data" is digital data that records the content of a conversation between a user (operator) and a customer.

[1245] "Text data" is character string data converted from voice data by a voice recognition engine.

[1246] "Customer harassment" refers to inappropriate words or actions directed at operators by customers, which increase the mental stress on the operators.

[1247] A "server" is a central computer system that processes voice data and performs speech recognition, natural language processing, and sentiment analysis.

[1248] A "terminal" is a computer device used by a user (operator) and is a device that acquires voice data and transmits it to a server.

[1249] "Notification" means a warning or instruction message sent to an operator or administrator when customer harassment is detected.

[1250] A "report" is written data that records the content of calls and the results of customer harassment detection, and is analysis material that is stored in a database.

[1251] A "database" is a system for efficiently storing and managing large amounts of data, and is a device for accumulating reports and call records.

[1252] "User" refers to an operator who handles customer service at a call center, and is the entity that acquires voice data and receives notifications from the system.

[1253] "Emotion analysis" is the process of analyzing and recognizing a user's emotional state from text data and voice tone.

[1254] "Escalation" is the process of notifying an administrator and encouraging direct intervention when customer harassment is deemed severe.

[1255] "Feedback" is the process of analyzing information stored in a database and providing information that helps improve the quality of the user's response and reduce mental stress.

[1256] This invention is a system that detects customer harassment in a call center in real time and takes appropriate action to reduce the mental burden on operators and improve the quality of customer service. Specific embodiments are described below.

[1257] Hardware and software used

[1258] Hardware:

[1259] Terminal: A computer used by an operator. This terminal has the function of acquiring voice data and sending it to a server.

[1260] Server: A central processing unit for speech recognition, natural language processing, and sentiment analysis.

[1261] software:

[1262] Speech recognition engine (e.g. Google Speech-to-Text)

[1263] Natural Language Processing (NLP) engines (e.g., SpaCy, NLTK)

[1264] Emotion engine (e.g. IBM Watson Tone Analyzer)

[1265] Data processing and calculation

[1266] 1. Acquiring audio data

[1267] When the user (operator) starts a call, the terminal records the conversation between the customer and the operator, and the recorded voice data is sent to the server in real time.

[1268] 2. Voice Recognition

[1269] The server inputs the received voice data into a voice recognition engine, which converts the voice data into "voice to text."

[1270] 3. Natural Language Processing

[1271] The server passes the converted text data to a natural language processing (NLP) engine, which detects specific keywords and phrases and performs contextual and sentiment analysis.

[1272] 4. Emotion Recognition by Emotion Engine

[1273] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state.

[1274] 5. Real-time detection and notification

[1275] If customer harassment and negative emotions are detected, the server immediately raises a response flag and sends a notification to the device.

[1276] 6. Escalation and Administrator Notification

[1277] If the harassment is deemed serious, the server will send an escalation notification to an administrator, who will monitor the call in real time and intervene directly if necessary.

[1278] 7. Report generation and data accumulation

[1279] After the call ends, the server logs the conversation, the results of customer harassment detection, and the results of sentiment analysis, and generates a report. These reports are stored in a database.

[1280] 8. Data analysis and feedback

[1281] The server periodically analyzes the information stored in the database and uses it to improve the quality of support and customer service for operators. This information is also used as material for training programs, reducing the mental burden on operators and continuously improving the quality of customer service.

[1282] Specific examples

[1283] Acquiring audio data:

[1284] When the operator starts a call with a customer, the device starts recording using a VoIP application and transmits the data to the server in real time.

[1285] Voice Recognition:

[1286] The server uses the Google Speech-to-Text API to analyze the received voice data and generate text data such as, "This product is completely unusable!"

[1287] Natural Language Processing:

[1288] The server uses an NLP engine (e.g., SpaCy) to detect offensive keywords and phrases, such as "totally unusable," and analyzes customer frustration.

[1289] Emotion Recognition with Emotion Engine:

[1290] The server uses IBM Watson Tone Analyzer to recognize the emotion of "anger" in real time from the customer's voice tone and text.

[1291] Real-time detection and notification:

[1292] When the server recognizes customer harassment, it displays a message on the operator's terminal saying, "A customer is making inappropriate remarks. Please respond calmly."

[1293] An example of a prompt for a generative AI model is:

[1294] "Develop a system that records call content in the call center in real time and detects customer harassment. Specific steps include voice recognition, natural language processing, and sentiment analysis, and notifying operators or managers as necessary. After the call ends, a report should be generated and saved in a database."

[1295] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1296] Step 1: Acquire audio data

[1297] When a user (operator) starts a call, the device records the call. The recorded voice data is sent to the server in real time as input. This process captures the voice call data between the customer and the operator. Specifically, when an operator makes a call using a VoIP application, the software automatically records the call and sends it to the server as digital voice data.

[1298] Step 2: Voice Recognition

[1299] The server inputs the received voice data into a voice recognition engine. The voice recognition engine analyzes the voice data and converts it from voice to text. In this process, the voice data is converted into character string data. For example, the Google Speech-to-Text API is used to convert the voice "This product is completely unusable!" into text. Text data is generated as the output.

[1300] Step 3: Natural Language Processing

[1301] The server passes the converted text data to a natural language processing (NLP) engine. The NLP engine detects specific keywords and phrases and performs contextual and sentiment analysis. Based on the input text data, keyword extraction and context understanding are performed. For example, an NLP engine (e.g., SpaCy) can be used to detect the offensive keyword "completely unusable" and infer customer frustration. The results of the contextual and sentiment analysis are generated as output.

[1302] Step 4: Emotion Recognition with the Emotion Engine

[1303] Based on the analysis results of the NLP engine, the server uses an emotion engine to perform a detailed analysis of the user's emotional state. The text data and analysis results obtained in the previous step are used as input. For example, IBM Watson Tone Analyzer is used to recognize the emotion of "anger" in real time from text and voice tone. This process clarifies the customer's specific emotional state, and the emotion analysis results are obtained as output.

[1304] Step 5: Real-time detection and notification

[1305] If the server detects customer harassment or negative emotions, it immediately raises a flag and sends a notification to the terminal. The results of emotion analysis are used as input. Specifically, based on the detected emotional state (e.g., anger), a message such as "A customer is making inappropriate remarks. Please respond calmly" is displayed on the operator's terminal. A notification message is sent as output.

[1306] Step 6: Escalation and Administrator Notification

[1307] If the customer harassment is deemed to be serious, the server sends an escalation notification to the administrator. The further analyzed sentiment analysis results are used as input. Specifically, an alert is sent to the administrator stating, "Customer harassment has been detected in the current call. Please monitor the call and intervene if necessary." The escalation notification is sent as output.

[1308] Step 7: Report generation and data accumulation

[1309] When the call ends, the server records the content of the conversation, the results of customer harassment detection, and the results of sentiment analysis in a log and generates a report. The call data and analysis results are used as input. In this process, a detailed report is created and saved in the database. Specifically, a report stating "Customer harassment was detected in the call with the customer. Please refer to the log for details" is generated and saved in the database. Report data is generated as output.

[1310] Step 8: Data analysis and feedback

[1311] The server periodically analyzes the information stored in the database to help improve the quality of operator support and customer service. As input, multiple reports stored in the database are analyzed. This process uses the analysis results as material for training programs and provides feedback to operators. Specifically, monthly reports are generated, and the response status of each operator and successful examples of customer harassment responses are analyzed and used as material for the next training session. As output, an analysis report and improvement proposals are created.

[1312] (Application example 2)

[1313] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1314] Customer harassment in call centers increases the mental burden on operators and can be a factor in reducing the quality of customer service. In particular, in urgent tasks such as security services, customer stress levels tend to rise, potentially increasing the frequency of customer harassment. Effective countermeasures are needed, but current systems struggle to respond in real time, and there is no mechanism in place for operators or managers to intervene quickly. For this reason, there is a strong demand for the development of a system that can quickly and accurately detect customer harassment and respond in real time.

[1315] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1316] In this invention, the server includes means for acquiring voice data, means for converting voice data into text data, means for detecting customer harassment based on the text data, means for notifying when customer harassment is detected, means for generating a report of the customer harassment detection results and the content of the call, means for saving the report in a database, means for analyzing emotions in the content of the call, and means for issuing response instructions to an operator and a manager based on the content of the call. This makes it possible to detect customer harassment in real time and respond quickly.

[1317] "Means for acquiring voice data" refers to a device or software that records the conversation between the customer and the operator and transmits the voice data to a server in real time.

[1318] The "means for converting voice data into text data" refers to a device or software that analyzes acquired voice data and converts it into text data using voice recognition technology.

[1319] A "means for detecting customer harassment based on text data" is a device or software that uses natural language processing technology on text data to evaluate whether or not there is customer harassment through specific keyword and sentiment analysis.

[1320] "Means for notifying when customer harassment is detected" refers to a device or software that issues appropriate warnings and instructions to operators or administrators in real time when customer harassment is detected.

[1321] The "means for generating a report of the customer harassment detection results and the content of the call" is a device or software that formats the customer harassment detection results and the related content of the call and outputs them as a report that can be referenced at a later date.

[1322] "Means for storing reports in a database" refers to a device or software that stores generated reports in a database system for managing and storing them.

[1323] The "means for analyzing the emotion of the call content" is a device or software that analyzes data including voice-recognized text data and voice tone to evaluate the emotional state of the customer.

[1324] "Means for issuing response instructions to operators and administrators based on the content of calls" refers to devices or software that issue specific response instructions to operators and administrators based on analyzed customer emotional data and the results of customer harassment detection.

[1325] The present invention provides a system that can detect customer harassment in a call center in real time and take appropriate action. Specific embodiments for carrying out the present invention will be described in detail below.

[1326] System configuration

[1327] The system of the present invention mainly acquires and analyzes voice data and implements a series of processes to detect customer harassment, which includes the following components:

[1328] 1. How to obtain audio data

[1329] It uses a microphone (built-in or external) installed on the device, such as a smartphone or head-mounted display (HMD).

[1330] When the call starts, the voice data is acquired and sent to the server in real time.

[1331] 2. Means of converting audio data into text data

[1332] On the server side, voice data is converted into text data using speech recognition technology, primarily using speech recognition engines such as Google Speech Recognition API.

[1333] 3. How to detect customer harassment based on text data

[1334] On the server side, the text data is analyzed using an NLP (Natural Language Processing) engine, using libraries such as TextBlob and NLTK.

[1335] 4. Means of notification if customer harassment is detected

[1336] Based on the detection results, a real-time warning is sent to the operator, with the message "A customer is making inappropriate remarks. Please respond calmly." appearing on the terminal display.

[1337] Administrators are notified of the escalation and are encouraged to prepare for intervention.

[1338] 5. A method for generating reports on customer harassment detection results and call content

[1339] After the call is completed, the server generates a report summarizing the detection results and the contents of the call, which is then stored in a database.

[1340] 6. How to save the report to a database

[1341] Use a database system such as MySQL or PostgreSQL to store reports.

[1342] 7. A way to analyze the sentiment of calls

[1343] Based on the content and tone of the customer's speech, sentiment is analyzed in detail using sentiment analysis engines such as TextBlob and VADER.

[1344] 8. Means of issuing instructions to operators and administrators based on the content of calls

[1345] Based on the results of sentiment analysis and customer harassment detection, the system sends specific instructions to operators and administrators, encouraging operators to remain calm and informing administrators of the need for intervention.

[1346] Specific examples

[1347] The following scenarios can be considered as examples of actual operation.

[1348] 1. An operator working at a security service anomaly reporting call center receives a call from a customer using a smartphone.

[1349] 2. A customer says something like, "This security system isn't working at all! What can you do about it?"

[1350] 3. The system converts the speech into text in real time and detects the emotion of "anger."

[1351] 4. A notification will appear on the operator's smartphone saying, "A customer is making inappropriate remarks. Please respond calmly."

[1352] 5. An escalation notification is sent to the administrator, who prepares to intervene.

[1353] Prompt Sentence Examples

[1354] An example of a prompt sentence to input to the generative AI model is as follows:

[1355] Develop an application that detects specific phrases and emotions in real time during calls with customers and notifies them if customer harassment occurs. This application uses Python for speech recognition and natural language processing. Specifically, it detects phrases containing the emotion "anger" from the converted speech and notifies the operator and administrator. Examples of customer harassment include "It's completely unusable" and "What are you going to do about it?"

[1356] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1357] Step 1:

[1358] The terminal captures the voice data of the customer and the operator as soon as the call starts, and the captured voice data is sent to the server in real time.

[1359] Input: Voice of the customer and operator

[1360] Output: Audio data

[1361] Specific operation: The device's microphone captures audio and transmits the audio data to the server in real time.

[1362] Step 2:

[1363] The server inputs the received voice data into a voice recognition engine (for example, Google Speech Recognition API) and converts the voice data into text data.

[1364] Input: Audio data

[1365] Output: Text data

[1366] Specific operation: The speech recognition engine analyzes the voice data and converts it into corresponding text data.

[1367] Step 3:

[1368] The server inputs the text data into a natural language processing (NLP) engine (e.g., TextBlob or NLTK) to analyze the text data, performing specific keyword detection and sentiment analysis.

[1369] Input: Text data

[1370] Output: Analysis results (keywords and emotional information)

[1371] How it works: The NLP engine analyzes text data and extracts information through keyword extraction and sentiment analysis.

[1372] Step 4:

[1373] The server detects customer harassment based on the analysis results. If negative emotions such as irritation or anger are detected, it determines that there is a high possibility of customer harassment.

[1374] Input: Analysis results

[1375] Output: Customer harassment judgment

[1376] Specific operation: The emotion engine evaluates the analysis results and determines whether or not there are any negative emotions.

[1377] Step 5:

[1378] If customer harassment is detected, the server sends a notification to the terminal and instructs the operator to respond. It also sends an escalation notification to the administrator.

[1379] Input: Customer Harassment Judgment

[1380] Output: Notification to operators and administrators

[1381] What happens: The server generates a notification message, displays a warning on the operator's screen, and sends an escalation notification to the administrator.

[1382] Step 6:

[1383] Once the call ends, the server generates a report based on the content of the call, the results of customer harassment detection, and the results of sentiment analysis.

[1384] Input: Call content, customer harassment detection results, emotion analysis results

[1385] Output: Call Report

[1386] Specific operation: The server compiles the call content and analysis results and creates a formatted report.

[1387] Step 7:

[1388] The server generates reports that are saved in a database and can be used later to improve the mental health of operators and the quality of customer service.

[1389] Input: Call Report

[1390] Output: Save to database

[1391] Specific operation: The server stores the report in a database system (e.g., MySQL or PostgreSQL).

[1392] Step 8:

[1393] The server periodically analyzes the information stored in the database and uses it to improve the quality of operator support and customer service. This information is also used as material for training programs.

[1394] Input: Reports in the database

[1395] Output: Analysis results, training materials

[1396] What it does: The server retrieves information from the database and extracts useful insights through data analysis.

[1397] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1398] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1399] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1400] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1401] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1402] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1403] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1404] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1405] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1406] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1407] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1408] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1409] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1410] 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.

[1411] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1412] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1413] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1414] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1415] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1416] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1417] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1418] The following is further disclosed regarding the above embodiment.

[1419] (Claim 1)

[1420] means for acquiring audio data;

[1421] means for converting the voice data into text data;

[1422] means for detecting customer harassment based on the text data;

[1423] means for notifying when said customer harassment is detected;

[1424] means for generating a report of the detection result of the customer harassment and the contents of the call;

[1425] The system includes means for storing said report in a database.

[1426] (Claim 2)

[1427] 2. The system of claim 1, wherein the means for detecting customer harassment is configured to identify customer harassment using specific keywords and sentiment analysis.

[1428] (Claim 3)

[1429] 2. The system according to claim 1, wherein the means for notifying includes an escalation notification to an administrator.

[1430] "Example 1"

[1431] (Claim 1)

[1432] When a user initiates a call, the terminal acquires voice data and transmits it to a server;

[1433] A means for the server to input voice data into a voice recognition engine and convert it into text data;

[1434] a means for the server to use a natural language processing algorithm to detect customer harassment in real time based on the converted text data;

[1435] When customer harassment is detected, the server notifies the terminal and, if necessary, sends an escalation notification to the administrator;

[1436] a means for recording the detection result of customer harassment and the contents of the call in a log file and generating a report;

[1437] The system includes a means for storing the generated reports in a database.

[1438] (Claim 2)

[1439] 10. The system of claim 1, wherein the server is configured to identify customer harassment using specific keywords and sentiment analysis.

[1440] (Claim 3)

[1441] 10. The system of claim 1, wherein after the call ends, the server records all text and detection results in a log file, generates a report, and stores it in a database.

[1442] "Application Example 1"

[1443] (Claim 1)

[1444] means for acquiring audio data;

[1445] means for converting the voice data into text data;

[1446] means for detecting customer harassment based on the text data;

[1447] means for notifying when said customer harassment is detected;

[1448] means for generating a report of the detection result of the customer harassment and the content of the conversation;

[1449] means for storing said report in a database;

[1450] means for displaying the customer harassment notice on the smart glasses;

[1451] A system including a means equipped with a natural language processing algorithm including a configuration for analyzing the voice data and text data.

[1452] (Claim 2)

[1453] 2. The system of claim 1, wherein the means for detecting customer harassment is configured to identify customer harassment using specific keywords and sentiment analysis.

[1454] (Claim 3)

[1455] 2. The system according to claim 1, wherein the means for notifying includes an escalation notification to an administrator.

[1456] "Example 2: Combining Emotion Engines"

[1457] (Claim 1)

[1458] a means for a user to obtain audio data;

[1459] means for transmitting the voice data to a server in real time by the terminal;

[1460] A server converts the voice data into text data;

[1461] A means for detecting customer harassment based on the text data in a server;

[1462] A means for notifying the terminal when the server detects customer harassment;

[1463] a means for generating a report on the detection result of the customer harassment and the contents of the call by the server;

[1464] means for the server to store said report in a database;

[1465] The system includes a server that analyzes the information stored in the database and provides feedback to reduce the user's mental load.

[1466] (Claim 2)

[1467] 10. The system of claim 1, wherein the server comprises means for identifying the customer harassment using specific keywords and sentiment analysis.

[1468] (Claim 3)

[1469] 2. The system of claim 1, wherein the server includes means for including said notification in an escalation notification to an administrator.

[1470] "Application example 2 when combining emotion engines"

[1471] (Claim 1)

[1472] means for acquiring audio data;

[1473] means for converting the voice data into text data;

[1474] means for detecting customer harassment based on the text data;

[1475] means for notifying when said customer harassment is detected;

[1476] means for generating a report of the detection result of the customer harassment and the contents of the call;

[1477] means for storing said report in a database;

[1478] means for analyzing emotions in the content of the call;

[1479] A means for issuing instructions to an operator and a manager based on the contents of the call;

[1480] A system including:

[1481] (Claim 2)

[1482] 2. The system of claim 1, wherein the means for detecting customer harassment is configured to identify customer harassment using specific keywords and sentiment analysis, and further configured to include escalation to an administrator.

[1483] (Claim 3)

[1484] 2. The system according to claim 1, wherein the means for notifying includes means for issuing a warning to an operator in real time and means for notifying an administrator of an escalation. [Explanation of symbols]

[1485] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for acquiring audio data; means for converting the voice data into text data; means for detecting customer harassment based on the text data; means for notifying when said customer harassment is detected; means for generating a report of the detection result of the customer harassment and the contents of the call; The system includes means for storing said report in a database.

2. 2. The system of claim 1, wherein the means for detecting customer harassment is configured to identify customer harassment using specific keywords and sentiment analysis.

3. 2. The system according to claim 1, wherein the means for notifying includes an escalation notification to an administrator.

Images